Editor-in-Chief

Dr. Sergey Victorovich Ulyanov

State University “Dubna”, Russian Federation

Editorial Board Members

Federico Félix Hahn-Schlam, Mexico Luis Pérez Domínguez, Mexico

Luiz Carlos Sandoval Góes, Brazil Abderraouf MAOUDJ, Algeria
Reza Javanmard Alitappeh, Iran Ratchatin Chancharoen, Thailand
Brahim Brahmi, Canada Shih-Wen Hsiao, Taiwan
Behzad Moradi, Iran Siti Azfanizam Ahmad, Malaysia
Hassan Alhelou, Syrian Arab Republic Mahmoud Shafik, United Kingdom
Lihong Zheng, Australia Hesham Mohamed Shehata, Egypt
Nguyen-Truc-Dao Nguyen, United States Hafiz Alabi Alaka, United Kingdom
José Miguel Rubio, Chile Abdelhakim DEBOUCHA, Algeria
Fazlollah Abbasi, Iran Karthick Srinivasan, Canada
Chi-Yi Tsai, TaiWan Ozoemena Anthony Ani, Nigeria
Shuo Feng, Canada Rong-Tsu Wang, Taiwan
Mohsen Kaboli, Germany Yu Zhao, China
Dragan Milan Randjelovic, Serbia Aslam Muhammad, Pakistan
Milan Kubina, Slovakia Yong Zhong, China
Yang Sun, China Xin Zhang, China
Yongmin Zhang, China ANISH PANDEY, Bhubaneswar
Mouna Afif, Tunisia Hojat Moayedirad, Iran
Yousef Awwad Daraghmi, Palestinian Mohammed Abdo Hashem Ali, Malaysia
Ahmad Fakharian, Iran Paolo Rocchi, Italy
Kamel Guesmi, Algeria Falah Hassan Ali Al-akashi, Iraq
Yuwen Shou, Taiwan Chien-Ho Ko, Taiwan
Sung-Ja Choi, Korea Bakİ Koyuncu, Turkey
Yahia ElFahem Said, Saudi Arabia Wai Kit Wong, Malaysia
Michał Pająk, Poland Viktor Manahov, United Kingdom
Qinwei Fan, China Riadh Ayachi, Tunisia
Andrey Ivanovich Kostogryzov, Russian Federation Terje Solsvik Kristensen, Norway
Ridha Ben Salah, Tunisia Andrey G. Reshetnikov, Russian Federation
Hussein Chible Chible, Lebanon Mustafa Faisal Abdelwahed, Egypt
Tianxing Cai, United States Ali Khosravi, Finland
Mahmoud Elsisi, Egypt Chen-Wu Wu, China
Jacky Y. K. NG, China Mariam Shah Musavi, France
Li Liu, China Shing Tenqchen, Taiwan
Fushun Liu, China Konstantinos Ilias Kotis, Greece
Ebtehal Turki Alotaibi, Saudi Arabia Junfei Qiu, United Kingdom
 Volume 2 Issue 2 · October 2020 · ISSN 2661-3220 (Online)

Artificial Intelligence
Advances

Editor-in-Chief
Dr. Sergey Victorovich Ulyanov
 Volume 2 | Issue 2 | October 2020 | Page 1-91
Artificial Intelligence Advances

Contents
Article
1 Intelligent Control of Mobile Robot with Redundant Manipulator & Stereovision: Quan-
tum / Soft Computing Toolkit
Kirill V. Koshelev　Alena V. Nikolaeva　Andrey G. Reshetnikov　Sergey V. Ulyanov
32 Intelligent Robust Control of Redundant Smart Robotic Arm Pt II: Quantum Computing
KB Optimizer Supremacy
Alena V. Nikolaeva　Sergey V. Ulyanov
68 Robotic Smart Prosthesis Arm with BCI and Kansei / Kawaii / Affective Engineering
Approach. Pt I: Quantum Soft Computing Supremacy
Alexey V. Nemchaninov　Alena V. Nikolaeva　Sergey V. Ulyanov　Andrey G. Reshetnikov

Review
88 Research and Analysis of Machine Learning Algorithm in Artificial Intelligence
Yang Li　Xueyuan Du　Yiheng Xu

Copyright
Artificial Intelligence Advances is licensed under a Creative Commons-Non-Commercial 4.0 International Copyright
(CC BY- NC4.0). Readers shall have the right to copy and distribute articles in this journal in any form in any medium,
and may also modify, convert or create on the basis of articles. In sharing and using articles in this journal, the user must
indicate the author and source, and mark the changes made in articles. Copyright © BILINGUAL PUBLISHING CO. All
Rights Reserved.
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Artificial Intelligence Advances

https://ojs.bilpublishing.com/index.php/aia

ARTICLE
Intelligent Control of Mobile Robot with Redundant Manipulator &
Stereovision: Quantum / Soft Computing Toolkit
Kirill V. Koshelev1,2　Alena V. Nikolaeva1,2　Andrey G. Reshetnikov1,2　Sergey V. Ulyanov1,2*
1. Dubna State University, Russia
2. INESYS LLC (EFKO GROUP), Russia

ARTICLE INFO ABSTRACT

Article history The task of an intelligent control system design applying soft and quan-
Received: 22 November 2019 tum computational intelligence technologies discussed. An example of a
control object as a mobile robot with redundant robotic manipulator and
Accepted: 3 June 2020 stereovision introduced. Design of robust knowledge bases is performed
Published Online: 30 June 2020 using a developed computational intelligence - quantum/soft computing
toolkit (QC/SCOptKBTM). The knowledge base self-organization process
Keywords: of fuzzy homogeneous regulators through the application of end-to-end
Quantum / Soft computing optimizer IT of quantum computing described. The coordination control between
the mobile robot and redundant manipulator with stereovision based on
Knowledge base soft computing described. The general design methodology of a gener-
Fuzzy controller alizing control unit based on the physical laws of quantum computing
Quantum fuzzy inference (quantum information-thermodynamic trade-off of control quality dis-
tribution and knowledge base self-organization goal) is considered. The
Multi-agent systems
modernization of the pattern recognition system based on stereo vision
Mobile robot stereo vision technology presented. The effectiveness of the proposed methodology is
demonstrated in comparison with the structures of control systems based
on soft computing for unforeseen control situations with sensor system.
The main objective of this article is to demonstrate the advantages of the
approach based on quantum/soft computing.

　
1. Introduction The prototype presented below is used in the education pro-

T
he application of robotic systems in various prob-
lem-oriented domains gives a significant positive
effect; moreover, not only industrial intelligent ro-
bots, but also robots for service use with various degrees of
social responsibility (medical robots, rescue robots etc.) are
spreading [1-3]. The effectiveness of such robot’s application
directly depends on the quality of control systems [3], which
are the most labor-intensive part of the robotic complex.
cess, the software component of each block of the control
system is modernized through the application of soft and
quantum computing IT. A significant increase in robustness
[4]
when using these types of computing allows us to say
that the creation of an effective robust and adaptive control
system for robotic devices requires an additional software
and algorithmic support platform in the form of sophisti-
cated toolkit based on soft and quantum computing (Quan-
tum computational intelligence toolkit) [4]. Specialists in

*Corresponding Author:
Sergey V. Ulyanov,
Dubna State University, Russia;
INESYS LLC (EFKO GROUP), Russia;
Email: ulyanovsv@mail.ru

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440 1

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
the design and development of various robots rely on an
intelligent platform for the application of computational in-
telligence technologies. This approach allows unifying the
process of intellectualization of the created hybrid industrial
service and special purpose robots [1-3].
In the article we are focused the attention on the fact
that the application of quantum computing in the design
process of intelligent control systems (ICS) significantly
increases the reliability of hybrid intelligent controllers
by introducing knowledge self-organization capability.
The basic principles of creating robust knowledge bases
(KB) are described. The paper describes an experiment,
the results of which make it clear that the effectiveness
of a fuzzy controller is significantly reduced in case
of emergency situations. We propose an approach that
allows to solve a similar problem by introducing a quan-
tum generalization of strategies in fuzzy inference in on-
line from a set of pre-defined fuzzy controllers (FC) by
new Quantum Fuzzy Inference (QFI). We consider a
new structure of intelligent control system (ICS) with a
quantum KB self-organization based on QFI. We espe-
cially focus on robustness of control because it is the
background for support the reliability of advanced con-
trol accuracy in uncertainty environments. The main goal
of this work is to provide a brief description of soft com-
puting tools for designing independent FC, then we will
provide the QFI methodology of quantum KB self-orga-
nization in unforeseen situations. Quantum supremacy of
classical intelligent control system design in unexpected
control situations demonstrated.
1.1 State of Art of Robotic Systems Development
The basis of the intelligent robot control system is to
achieve a synergistic effect from the interaction of various
on-board robot systems (a movement system, a mobile
manipulator control system, an image processing system
etc.). This approach was identified and supported by re-
search and development in Japan in the mid-90s of the
last century [1-3]. It should be noted that now this particular
approach is fundamental in robotics.
In particular case, in [1-3] is described how the mobile
robot for service use works in buildings with different
scenes of rooms and moves in unstructured environments
in presence of many human being operators and unexpect-
ed obstacles applied cognitive graphic of path planning
and soft computing for control of different technological
operations (open door, getting on an elevator, room con-
trol operations and so on). It was suggested to construct a
simulation system for mobile service robot behavior based
on cognitive graphics with virtual and complimentary re-
ality. This system is used for possible world’s simulation
2 Distributed under creative commons license 4.0
in the robot artificial life. This allows us to evaluate the
control algorithms in online robot behavior and to reduce
difficulties connected with such troubles as robot colli-
sions with obstacles and robot hardware damages. In this
Item we describe a new approach to intelligent control
system design using soft computing technology. A new
form of direct human - robot communications (including
emotion, instinct and intuition) and an autonomous lo-
comotion control system were developed (see Figure 1).
Authors considered direct human-robot communications
based on natural language (NL) and construct the simula-
tion system of spatial scenes and robot behavior in virtual
reality (VR).
Figure 1. Structure of Artificial Intelligence (AI) control
system with distributed knowledge representation (on
control signal levels). a - Intelligent control “in large” b
- Intelligent control “in small” c - Control on executive
level
It was explained also the managing system which
controls cooperatively three sub-systems of the service
robot, as the locomotion system, the handling system for
a mobile manipulator and the image processing system
as human vision system. Also, three sub-systems which
organize the service robot system for its autonomous nav-
igation and these soft computing are described in [3,5]. The
locomotion control system is composed of four functions,
i.e. locomotion control, planning for works, learning and
recognition. These four functions are related to each other.
By using the handling system for a mobile manipulator
and the image processing system as human vision system,
the robot can realize some technology operations, for ex-
ample, opening a door and getting on an elevator. These
three sub-systems are based on fuzzy control, fuzzy neural
networks (FNN) and genetic algorithm (GA). Experimen-
tal results on the developed robot show that the proposed
methods are very useful for autonomous locomotion con-
trol of the robot.
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Example. Intelligent control and soft computing for
avoidance of obstacles and execution of technology op-
erations - robot is power-wheeled steering type which is
achieved by two driving wheels and a caster with passive
suspension for stable locomotion. Thirteen ultrasonic (US)
sensors, nine infrared (IR) sensors, a five degree-of-free-
dom (DOF) manipulator with a three-finger hand and a
CCD camera are equipped on the robot for conducting
tasks and works in buildings including human being,
opening door and getting on an elevator. In process of
robot’s locomotion from point to point mobile robot must
avoid obstacles in room, and achieve starting position for
successful opening room door (as technology operation
from one point to another point) and go out the room to
elevator in presence of obstacles in a corridor. Movement
process is planned by the managing system which was de-
scribed in [5].
It must be said that the management of poorly formal-
ized and poorly structured control objects (CO) with a
variable and complex structure, variable parameters and
uncertain operating conditions, in a group or an autono-
mous object requires the use of end-to-end (in the general
case - quantum) information technologies. Studying the
problems of implementing control systems for such new
and extremely complex CO as NICA, LHC, data process-
ing of telescopes, autopilots, objects of intelligent robotics
indicates a significant dependence of the robustness of the
system on the embedded computing basis of the most ICS
[6]
. It is necessary to revise the science platform that is em-
bedded in end-to-end information technology (IT) for the
design of ICS.
Traditional modern automatic control systems fun-
damentally do not take into account the occurrence of
contingency management situations and do not include
the human being cognitive factor in the control loop,
which does not guarantee the timely achievement of the
management goal - obtaining the maximum intensity of
functioning and stable repetition of the required operating
modes with minimal cost of useful resources, such as:
time settings, power consumption and more other. In the
conditions of uncertainty or inaccuracy (imperfect) of the
initial information, unforeseen situations or information
risk, the traditional (using the principle of global negative
feedback) and widely used in industry PID-controller of-
ten does not cope with the management task. At the same
time, the solution to the problem of global robustness of
the PID-controller is still unknown, despite the relevance
of this problem.
Remark. Systems based on biological systems and
mimicking their working conditions and behavior are
found to be more adaptive and systematically. Their
Distributed under creative commons license 4.0
responses are faster and more accurate as compared to
the conventional systems. The major drawbacks found
while dealing with the static controllers like PI, PID
etc. are that they are less efficient for dealing with the
complexities and non-linear disturbances. Also, they are
non-adaptive and less robust. However, the computation-
al model based on the mammalian decision system could
deal with these complexities and non-linearity more ide-
ally and robustly. In online process control systems, all
the conditions are uncertain. Also, some complex pro-
cesses are very dynamic and conditionals are not known
with accurate precisions. Thus, in these systems, the
design of the controller is very crucial. Also, solving this
non-linearity through conventional computational model
proves to be very slow. Also, in online scenarios the dis-
turbance sources are unknown and thus any prior model
cannot be predicted. Adaptive models are required for
dealing such scenarios. Thus, various AI techniques like
neuro-fuzzy techniques, Genetic Algorithms, PSO are
proved. A model based on limbic-system of mammalian
brain emotions has been proposed by Caro Lucas in 2004
[7]
. He has demonstrated that emotion-based decisions
are quick and more satisfying all the constraints. He has
proposed the model based on dynamic limbic system
of brain of mammalian. A BELBIC model presents the
mimics of the limbic systems components - amygdala,
orbito-frontal cortex, thalamus, sensory cortex. It has
been implemented for various SISO, MIMO, and other
non-linear systems. Various results have demonstrated
its very fast control action, better disturbance handling
capacity, and robustness (see Appendix 1).
This is also due to the fact that the classical methods
of control theory have synthesis and design methods for
well-formalized, well-defined and well-described CO
that operate in previously known conditions and situ-
ations. However, unforeseen circumstances and their
individual characteristics determine the secrecy and
underdetermination of the parameters of physical and
mathematical models, and should be taken into account
in the KB of intelligent FC. Quantum end-to-end IT al-
lows designing hierarchical ICS’s, which make possible
to redistribute the degree of decision-making responsi-
bility between ICS’s depending on a dynamically chang-
ing situation [6,8].
This article also provides a brief description of the de-
veloped system of pattern recognition, which is based on
stereovision technology. The use of stereovision allows
obtaining data of the depth of the image, the distance to
objects, provides an opportunity to build a 3Dpicture of
the surrounding world. This article discusses the archi-
tecture of convolutional neural networks and its applica-
DOI: https://doi.org/10.30564/aia.v2i2.1440 3
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

tion in intelligent robotics. Advantages of convolutional

networks are used for recognition with a high degree of
invariance to transformations, distortions and scaling.
The convolutional neural network is a widespread and
effective tool for deep machine learning, with the help of
which computer vision problems are successfully solved.
The process of classifying images of a convolutional
neural network is close to a similar process occurring in
the cortex of the human brain. This paper presents a brief
description of the modernization of the pattern recognition
system based on stereo vision technology.

1.2 IT for the Design of Robust Intelligent Con-

trol Systems
The use of FC in conjunction with the PID-controller led
to the creation of hybrid fuzzy ICS with various levels of
intelligence depending on the completeness and correct-
ness of the designed KB. The use of soft computing tech-
nology (based on GA and FNN) has expanded the field of
effective application of FC by adding new functions in the
form of learning and adaptation.
The developed technology and intelligent tools Soft
Computing Optimizer (SCO &IICS) SCOptKBTM made
it possible to design robust KB based on the solution of
one of the algorithmically difficult problems of the theory
of artificial intelligence - the extraction, processing and
formation of objective knowledge without expert esti-
mates. Three GA’s are used in this SCO, which allow us
to design the optimal FC structure (type and number of
Membership function (MF), their parameters, number of
fuzzy inference rules), which approximates the training
(learning) signal with the required error. In this case, the
optimal FNN structure is automatically designed and a
universal approximator model is formed in the form of FC
with a finite number of production rules in KB [8].
The technology uses new types of computational intel-
ligence (see Figure 2). The design process for robust KB
consists of two interconnected stages based on soft and
quantum computing. Design is carried out on the basis of
Computational Intelligence Toolkits - SCO & QCO (QC/
SCOptKBTM).
At the first stage of the design process individual KB’s
are formed for two (or more) FC’s, which operating in
specific learning control situations. At this stage an evolu-
tionary multi-criteria GA with soft computing technology
and fuzzy stochastic modeling is used. The mathematical
model of the system and the CO (functioning in conditions
of training with reinforcement), according to the measured
output signals of sensors and control actions, can act as an
information source of the design process.
Figure 2. Structure of information design technology of
IFICS
From the point of view of hierarchical multi-agent
management, the developed IT allows managing both au-
tonomous objects and hierarchically structurally connect-
ed teams of autonomous robots (Multiple KB design in
Figure 2). On Figure 2 is presented the information design
technology of robust integrated fuzzy intelligent control
systems (IFICS). Main problem in this technology is the
design of robust KB of FC that can include the self-orga-
nization of knowledge in unpredicted control situations.
The background of this design processes is KB optimizer
based on quantum/soft computing [9]. Concrete industrial
Benchmarks (as “cart - pole” system, robotic unicycle,
robotic motorcycle, mobile robot for service use, semi-ac-
tive car suspension system etc. [5,10-13]) are tested success-
fully with the developed design technology.
2. Unconventional Computational Intelli-
gence Toolkit: Soft and Quantum Computing
Technologies
Soft computing and quantum computing are new types
of unconventional computational intelligence toolkit
(details see in http://www.qcoptimizer.com). Technology
of soft computing is based on GA, FNN and fuzzy logic
inference. Quantum computational intelligence is used
quantum genetic search algorithm, quantum neural net-
work and QFI. These algorithms are including three main
operators. In GA selection, crossover and mutation opera-
tors are used; in quantum search algorithm superposition,
entanglement and interference are applied.
On Figure 3 is presented the structure of robust intel-
ligent control system in unpredicted control situations.
This structure is the particular case of general structure of

4 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
IFICS (see Figure 2).
Figure 3. Structure of robust intelligent control system in
unpredicted control situations
Remark. An application of developed toolkit in design
of “Hu-Machine technology” based on Kansei Engineer-
ing is demonstrated for emotion generating enterprise
(purpose of enterprise). We are considered the humanized
technology of intelligent robotic systems design based
on Kansei Engineering and Quantum/Soft Computing [14].
As well known the subject of humanized technology or
human-related systems has been actively researched. With
the increasing concern regarding human factors in system
development Kansei Engineering and Soft Computing
are the most representative research fields on this subject.
Soft computing toolkit is developed for emotion, instinct,
and intuition recognition and expression generation. In
particular, with GA (as effective random search of solu-
tion) an intuition process is modeled. FNN is used for de-
scription of instinct process that modeled approximation
of optimal solution in unpredicted control situation. Fuzzy
logic control is used for design of emotion according to
corresponding look-up table. Quantum computing toolkit
is used for increasing of robustness in intelligent control
systems based on superposition and correlations of affec-
tive operations. Detail description of quantum computing
toolkit and QFI are described in [8]. Results of simulations
[4,8]
are shown that from two unstable fuzzy controllers it
is possible to design in online a new robust fuzzy control-
ler. It is a pure quantum effect and do not have a classical
analogy [15]. These results of quantum game simulation
show that the winner is quantum fuzzy controller (QFC)
designed from two imperfect KB controller with mini-
mum of generalized entropy production. Therefore, QFI
supports optimal thermodynamic trade-off between sta-
bility, controllability and robustness in self-organization
process (from viewpoint of physical background of global
robustness in ICSs). Also, important the new result for
Distributed under creative commons license 4.0
advanced control system that all other controllers (FC1,
FC2) are failed but QFC is designed in online a new FC
with increasing robustness. This approach was applied to
other complex robotic systems [11-13].
2.1 Structure and Main Functions of SCO
As above mentioned in ICS design soft computing tech-
nology is a combination of the following approaches:
fuzzy systems theory, GA and FNN. FC is the central
element of ICS and generates time-dependent control sig-
nals (control laws) with gain kp, kd, ki (coefficient gain’s
schedule) of the PID-controller.ICS structure with FC and
SCO blocks in soft computing simultaneously includes
the following management qualities: controllability, pre-
cision accuracy and stability (lower control level - ACS),
learning and adaptation (upper intelligent control level -
FC with robust KB in learning control situation). Fuzzy
controller is the universal approximator with bounded set
of production logic inference rules (look-up table). In the
case of approximating a certain control signal, the input
components can be a control error, the error integral and
its derivative, and the output component can be the re-
quired value of the control action or adjustable parameters
of the control system, for example, the coefficient gains of
the PID controller. The result of the approximation of the
learning signal is the designed KB for FC, including the
optimal finite set of production rules and optimally formed
parameters of the membership function of the input and
output variables of FC.
The developed technology supports a software cli-
ent-server architecture with remote configuration and
transfer of KB, which allows the CO to receive KB from
the SCO unit or from other CO’s remotely, this makes
it possible to remotely control structurally new objects,
such as swarm of robots, multi-agent systems, complex
distributed automated production, physical installations
such as Mega-Science. In addition, this technology makes
it possible to accumulate (and later to acquire new knowl-
edge), update and adapt KB for a specific CO and control
situation (including emergency) in online. The result of
the toolkit application at the first stage of the ICS design
process is the required type of universal approximator in
the form of FC with optimal KB structure (see Figure 2).
2.2 Structure and Main Functions of QFI
The purpose of applying quantum computing and creating
a self-organizing quantum controller is to combine the
intelligent controllers of various sensors obtained in the
first stage into a self-organizing connected multi-agent
network based on a quantum controller and cognitive-in-
DOI: https://doi.org/10.30564/aia.v2i2.1440 5
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
formation interaction between KBs. Quantum computing
technologies for creating self-organizing KB intelligent
regulators are considered in [16,17]. In [16] a new type of
quantum search algorithm on the generalized space KB
of FC was described (KB designed on the basis of soft
computing technologies). The QFI model implements the
self-organization of KB, based on the physical laws of the
theory of quantum computing [18,19] and the application of
four operators: superposition, quantum correlation, in-
terference, and measurement. The first three are unitary,
reversible quantum operators, and the fourth (measurement
operator) is classical (irreversible).
The QFI design process includes the design of a quan-
tum algorithmic gate - the matrix form of three quantum
operators: superposition, entanglement and interference,
which are part of the structure of quantum search algo-
rithms, and a source of quantum information that is a hid-
den variable in classical states. The main unit of such ICS
is the quantum genetic search algorithm (QGSA). QFI
operators can be implemented both on a quantum and on
a classical processor, which in the second case allows you
to integrate them into various control systems and built-in
intelligent controllers, taking into account the limitations
of the computing resource of the elements of an exper-
imental laboratory bench or on-board system. The very
process of creating such an algorithmic gate requires the
attraction of large computational resources, in connection
with which a methodology has been developed for pro-
cessing big data and conducting quantum computing on a
supercomputer (for example, “Govorun” LIT JINR).
The QFI algorithm (see Figure 4) for determining new
control parameters (a special case of implementing quan-
tum computing on a classical processor) consists of such
stages as normalization, formation of a quantum bit, after
which the optimal structure of the quantum algorithmic
gate is selected, and the state with maximum amplitude is
selected probability, and decoding the signal to obtain new
control parameters.
Figure 4. QFI algorithm
Studies conducted in conjunction with ST Microelec-
tronics & Yamaha Motor Co and testing at various CO have
6 Distributed under creative commons license 4.0
confirmed the existence of a synergistic effect of self-orga-
nization during the formation of robust KB from designed
non-robust imperfect KB using quantum computing. More-
over, an additional information resource for control is based
on the extraction of quantum information hidden in classi-
cal states embedded in a quantum algorithmic gate. Such
ICS design methods allowed achieving global robustness in
online, while using the computing resources of the embed-
ded conventional processor of the on-board system.
The use of new quantum computing paradigm and deep
machine learning in the group of interaction robots pro-
vides the following advantages:
(1) the potential acceleration of computing through the
use of quantum evolutionary operators, which will make it
possible to fully use the powerful, but slow GA to get the
possibility of training the system in online;
(2) the introduction of quantum superposition and
quantum-classical correlation operators into the classical
algorithm entails the appearance of unique properties of
the data processing process, which affects the result of the
work of a group of robots;
(3) low dependence on environmental disturbances;
(4) independence from static electricity or movement
of the CO or digital video camera;
(5) registration of changes and redistribution of tasks
during changes in a group of autonomous agents.
2.3 Control Performance Measure: Thermody-
namic Trade-off and Interrelations between Sta-
bility, Controllability and Robustness
According to Figure 5, one of the main tasks of designing
ICS consists in providing that the developed (chosen)
structure possesses the required level of control quality
and robustness (supports the required indices of reliability
and accuracy of control under the conditions of informa-
tion uncertainty).
Figure 5. Performance and interrelations between of con-
trol quality criteria
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Note that one of the most important and hard-to-solve
problems of designing ICS’s is the design of robust KB
that allow the ICS to operate under the conditions of in-
formation uncertainty and risk. The core of technique for
designing robust KB of FC’s is generated by new types of
computing and simulation processes.
Remark. We are witnessing a rapidly growing interest
in the field of advanced computational intelligence, a “soft
computing” technique. Soft computing integrates fuzzy
logic, neural networks, evolutionary computation, and
chaos. Soft computing is the most important technology
available for designing ICSs and cognitive control. The
difficulties of fuzzy logic involve acquiring knowledge
from experts and finding knowledge for unknown tasks.
This is related to design problems in constructing fuzzy
rules. FNN’s and GA’s are attracting attention for their po-
tential in raising the efficiency of knowledge finding and
acquisition. Combining the technologies of fuzzy logic,
FNN’s and GA’s, i.e., soft computing techniques will have
a tremendous impact on the fields of intelligent systems
and control design.
To explain the apparent success of soft computing,
we must determine the basic capabilities of different soft
computing frameworks. Recently, the application of ICS
structures based on new types of computations (such as
soft, quantum computing) has drawn the ever-increasing
attention of researchers. Numerous investigations con-
ducted have shown that soft computing possess the fol-
lowing points of favor: soft computing retain the main ad-
vantages of conventional automatic control systems (such
as stability, controllability, observable ability, etc.); soft
computing have an optimal (from the point of view of a
given control objective performance) KB; soft computing
guarantee the attainability of the required control quality
based on the designed KB
One of the main problems of modern control theory is
to develop and design automatic control systems that meet
the three main requirements: stability, controllability, and
robustness. The listed quality criteria ensure the required
accuracy of control and reliability of operation of the con-
trolled object under the conditions of incomplete infor-
mation about the external perturbations and under noise
in the measurement and control channels, uncertainty in
either the structure or parameters of the control object, or
under limited possibility of a formalized description of the
control goal. Therefore, in practice of advanced control
systems main sources of unpredicted control situations are
as following:
(1) Control object
① Type of unstable dynamic behavior
(a) Local unstable
Distributed under creative commons license 4.0
(b) Global unstable
(c) Partial unstable on generalized coordinate and
non-linear braces
② Time-dependent random structure or parametric ex-
citations
③ Type of model description
(a) Mathematical model
(b) Physical model
(c) Partial mathematical and fuzzy physical model
(2) External random excitations
① Different probability density functions
② Time-dependent probability functions
(3) Measurement system
① Sensor noise
② Time delay
③ Random time delay with sensor noise
(4) Different types of reference signals
(5) Different types of traditional controllers
This problem is solved in three stages as following:
(1) the characteristics of stability of the controlled plant
are determined for fixed conditions of its operation in the
external environment;
(2) a control law is formed that provides the stability of
operation of the controlled plant for a given accuracy of
control (according to a given criteria of the optimal con-
trol);
(3) the sensitivity and robustness of the dynamic be-
havior of the controlled plant are tested for various classes
of random perturbations and noise.
These design stages are considered by modern control
theory as relatively independent. The main problem of
designing automatic control systems is to determine an
optimal interaction between these three quality indices of
control performance. For robust structures of automatic
control systems, a physical control principle can be prov-
en that allows one to establish in an analytic form the
correspondence between the required level of stability,
controllability, and robustness of the control. This allows
one to determine the required intelligence level of the
automatic control system depending on the complexity of
the particular control problem.
Let us briefly consider main physical principles of an
energy-based control processes that allow one to establish
the interrelation between the qualitative dynamic char-
acteristics of the controlled plant and the actuator of the
automatic control system: stability, controllability, and
robustness of control. For this purpose, we are employing
the informational and thermodynamic approaches that
join by a homogeneous condition the criteria of dynamic
stability (the Lyapunov’s function), controllability, and
robustness.
DOI: https://doi.org/10.30564/aia.v2i2.1440 7
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Example: Thermodynamics trade-off between stability, characterized uniquely by the parameters of the nonlinear
controllability, and robustness. Consider a dynamic con- dynamic automatic control system, which results in deter-
trolled plant given by the equation mination of an optimal selective trajectory from the set of
possible trajectories in optimization problems. Thus, the
dq first condition is fulfilled automatically.
ϕ ( q, S ( t ) , t , u , ξ ( t ) ) , u =
= f ( q, qd , t ) ,(1)
dt dV
Assume that the second condition ≤ 0 holds. In
where q is the vector of generalized coordinates de- dt
scribing the dynamics of the controlled plant; S is the gen- this case, the complete derivative of the Lyapunov func-
eralized entropy of dynamic system; u is the control force tion has the form
(the output of the actuator of the automatic control sys-
dV
tem); qd(t) is reference signal, ξ(t) is random disturbance
dt
= ∑ q q
i
i i + SS = ∑ q ϕ ( q, S , t , u ) + ( S
i
i i cob ( )
− Sc ) Scob − Sc .
and t is the time. The necessary and sufficient conditions
of asymptotic stability of dynamic system with ξ(t)≡0 are Thus, taking into account (1) and the notation intro-
determined by the physical constraints (for example, as duced above, we have
for Port-controlled Hamiltonian Systems (PCHS) [20]) on
the form of the Lyapunov function, which possesses two dV
= ∑ qi ϕi ( q, ( Ψ − ϒ ) , t , u ) + ( Ψ − ϒ ) ( Ψ
 − ϒ ) ≤ 0
important properties represented by the following condi- dt    .
 
i

tions: Stability Robustness
Controllability
(I) This is a strictly positive function of generalized co- (3)
ordinates, i.e., V>0;
(II) The complete derivative in time of the Lyapunov’s Figure 6 shows the role of developed thermodynamic
function is a non-positive function, trade-off in robust control design.

dV
≤0.
dt

In general case Lagrangian dynamic system (1) is not

lossless with corresponding outputs. By conditions (I) and
(II), as the generalized Lyapunov function, we take the
function

1 n 2 1 2
=V ∑ qi + 2 S ,(2)
2 i =1

where S=Sp-Sc is the production of entropy in the open

system “control object + controller”; S p = Ψ ( q, q , t ) Figure 6. Physical law of intelligent control as back-
is the entropy production in the controlled plant; and ground of IFICS design technology
Sc = ϒ ( e, t ) is the entropy production in the controller In the case of PCHS [20] we have
(actuator of the automatic control system). It is possible
∂H ( x , t )
T
to introduce the entropy characteristics in Eqs. (1) and (2) dV
because of the scalar property of entropy as a function of dt
∑
= xi  J i ( xi , t ) − Ri ( xi , S , t )  i i
∂xi
+ gi ( xi , t ) ui
 i
   
time, S(t). Stability Controllability
Remark. It is worth noting that the presence of entropy
+ (Ψ − ϒ)(Ψ
 − ϒ ) ≤ 0
production in (1) as a parameter (reflects the dynamics of 
Robustness
the behavior of the CO) and results in a new class of sub-
(4)
stantially nonlinear dynamic automatic control systems.
The choice of the minimum entropy production both in For the definition of the error system, it is obvious that
the control object and in the fuzzy PID controller as a stabilization (settling the state at the origin) of the error
fitness function in the GA allows one to obtain feasible system implies the tracking control of the original system.
robust control laws for the gains in the hybrid fuzzy PID It was proposed a procedure to realize an error system of a
controller. The entropy production of a dynamic system is given PCHS by another PCHS via the generalized canoni-

8 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
cal transformation:
Φ ( x, t ) , H =
x= H ( x, t ) + U ( x, t ) ,
,
y + α ( x, t ) , u =
y= u + β ( x, t )
which preserves the structure of port-controlled Ham-
iltonian systems with dissipation, that is, the system into
an appropriate Hamiltonian system in such a way that the
transformed system
∂H ( x , t )
T
 J ( x , t ) − R ( x , t ) 
x = + g ( x,t )u,
∂x
∂H ( x , t )
T can never increase after the first resetting event. Further-
y = g ( x,t )
T
∂x
satisfies x ( t ) =
0 ⇔ x (t ) =
xd ( t )
Relation (4) relates the stability, controllability, and ro-
bustness properties.
Remark. This approach was firstly presented in [21]. It
was introduced the new physical measure of control qual-
ity (3) to complex non-linear controlled objects described
as non-linear dissipative models. This physical measure of
control quality is based on the physical law of minimum
entropy production rate in ICS and in dynamic behavior
of complex object. The problem of the minimum entropy
production rate is equivalent with the associated problem
of the maximum released mechanical work as the optimal
solutions of corresponding Hamilton-Jacobi-Bellman
equations. It has shown that the variational fixed-end
problem of the maximum work W is equivalent to the
variational fixed-end problem of the minimum entropy
production. In this case both optimal solutions are equiv-
alent for the dynamic control of complex systems and the
principle of minimum of entropy production guarantee
the maximal released mechanical work with intelligent
operations. This new physical measure of control quality
we applied as fitness function of GA in optimal control
system design.
In [22] have studied something similar, what was called
as “equipartition of energy”. Such state corresponds to the
minimum of system entropy. The introduction of physical
criteria (the minimum entropy production rate) can guar-
antee the stability and robustness of control. This method
differs from aforesaid design method in that a new intelli-
gent global feedback in control system is introduced. The
interrelation between the stability of CO (the Lyapunov
function) and controllability (the entropy production rate)
is used. The basic peculiarity of the given method is the
necessity of model investigation for control object and
the calculation of entropy production rate through the pa-
rameters of the developed model. The integration of joint
Distributed under creative commons license 4.0
systems of equations (the equations of mechanical model
motion and the equations of entropy production rate)
enable to use the result as the fitness function in GA as
a new type of CI. Acceleration method of integration for
these equations is described in [23].
Main goal of robust intelligent control is support of
optimal trade-off between stability, controllability and
robustness with thermodynamic relation as thermodynam-
ically stabilizing compensator. The resetting set is thus
defined to be the set of all points in the closed-loop state
space that correspond to decreasing controller emulated
energy. By resetting the controller states, the CO energy
more, if the closed-loop system total energy is conserved
between resetting events, then a decrease in plant energy
is accompanied by a corresponding increase in emulated
energy.
In concluding this section, we formulate the following
conclusions:
(1) The introduced physical law of intelligent control (3)
provides a background of design of robust KB’s of ICS’s
(with different levels of intelligence) based on soft com-
puting.
(2) The technique of soft computing gives the opportu-
nity to develop a universal approximator in the form of a
fuzzy automatic control system, which elicits information
from the data of simulation of the dynamic behavior of the
CO and the actuator of the automatic control system.
(3) The application of soft computing guarantees the
purposeful design of the corresponding robustness level
by an optimal design of the total number of production
rules and types of membership functions in the knowledge
base.
The main components and their interrelations in the
information design technology (IDT) are based on new
types of (soft and quantum) computing. The key point of
this IDT is the use of the method of eliciting objective
knowledge about the control process irrespective of the
subjective experience of experts and the design of objec-
tive knowledge bases of a FC, which is principal compo-
nent of a robust ICS. The output result of application of
this IDT is a robust KB of the FC that allows the ICS to
operate under various types of information uncertainty.
Self-organized ICS based on soft computing technology
was described in [21] that can support thermodynamic trade-
off in interrelations between stability, controllability and
robustness. As particular case Eq. (3) includes the entropic
principle of robustness [24]. The support of optimal thermo-
dynamic trade-off between stability, controllability and ro-
bustness in self-organization processes with (3), (4) can be
realized using a new quantum control algorithm of self-or-
DOI: https://doi.org/10.30564/aia.v2i2.1440 9
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

ganization in KB of robust FC based on quantum comput- situations in an external random environment. The design
ing operations (that absent in soft computing toolkit). of private individual KB with the application of software
tools used for specified control situations is carried out
3. The Role of Robust Intelligent Control Sys- in accordance with the design technology of the ICS and
tems in Advanced Robotics discussed in detail in [26]. The main task solved by QFI is
the formation of KB with an increased level of robustness
The process of developing innovative information tech-
from a finite set of KB for workstations formed using soft
nologies for the design of embedded self-organizing ro-
computing.
bust intelligent control systems based on quantum end-
Peculiarities of quantum approach. Recall that the
to-end artificial intelligence technologies is inextricably
square of the amplitude of the probabilities of a state in
linked with the development of scientific and technolog-
quantum mechanics is equal to the classical probability
ical progress in areas such as particle physics, robotic
of finding a quantum system in a given state (Bohr’s pos-
systems (sensors, computation processors components),
tulate, which has several variants of rigorous justification
new production technologies, creation of innovative smart [27]
). From the point of view of the quantum information
materials, analytics of large experimental data, search
theory, a pure quantum state is characterized, as is known,
in unstructured databases. The development of ICS for
by the von Neumann entropy value of zero. Therefore,
traditional automated control systems with an increased
the intelligent quantum state in the considered case takes
level of robustness is of significant theoretical, practical
place for the minimum informational entropy of the Shan-
and commercial importance. It is imperative that modern
non’s quantum state. The desired minimum is achieved, in
control systems maintain the required levels of accura-
turn, with the maximum probability Pi of the state. Since
cy and reliability in unforeseen situations. The practice
Pi, by definition, is the square of the corresponding proba-
and results of modeling real objects have shown that in
bility amplitude, the principle of the maximum probability
conditions of uncertainty or inaccuracy in the initial infor-
amplitude in the correlated state can be taken as a criteri-
mation, unforeseen situations, or information risk the tra-
on for selecting the priority “intelligent” correlation (co-
ditional (using the principle of global negative feedback)
herent) state in a superposition of possible candidates [28,29].
and widely used in industry PID-controller often fails to
Thus, by calculating the amplitudes of quantum states in
cope with the management task. The application of FC
a superposition of states with mixed types of quantum
in conjunction with the PID-controller led to the creation
correlation and choosing the maximum among them, a
of hybrid fuzzy ICSs with different levels of intelligence
quantum oracle model is implemented that contains the
depending on the completeness and correctness of the de-
necessary information about the desired solution. Using
signed KB.
the standard decoding procedure (the internal product of
The use of soft computing technology (based on GA
vectors in a Hilbert space) and selecting the scaling fac-
and FNN) has expanded the field of effective use of FC
tors for the output values of the projected gain factors,
by adding new functions in the form of learning and ad-
iterative work of algorithm of QFI is carried out. Remote
aptation (see Figure 5). However, in the general case of
connection of the CO to a stationary computing system
contingency management situations, it is very difficult to
opens up the possibility of remote configuration, forma-
design a globally “good” and robust ICS structure. This
tion and self-organization of KB in online. The presented
restriction is especially typical for unforeseen control
model QFI makes it possible to solve the classical prob-
situations when the CO operates in rapidly changing con-
lems of designing robust KB in ICS structures that have
ditions (sensor failure or noise in the measuring system,
no analogues among the family of randomized classical
the presence of a delay time for control or measurement
algorithms and is characterized by polynomial computa-
signals, a sharp change in the structure of the CO or its pa-
tional complexity. The design of ICS on QFI is carried out
rameters, etc.). A solution to this kind of problems can be
using the developed software toolkit “Quantum Optimiz-
found by introducing the principle of self-organization of
er”. The technology of application of quantum fuzzy logic
KB into the design process of FC, which is implemented
allows combining several KB into a single control system,
and programmatically supported by the developed model
thereby allowing FNN to work in parallel as quantum
of QFI using the methodology of quantum soft computing
neural network.
and Intelligent System of System Engineering [25]. The
Remark. From the point of view of computational com-
proposed model of QFI uses private individual KB of FC,
plexity theories [31-33], the developed algorithm belongs to
each of which is obtained by using SCO for the corre-
the class of polynomial algorithms with a limited error
sponding operating conditions of the CO and fixed control
- the BPP class (bounded-error probabilistic polynomial

10 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

time), and its quantum generalization - to the BQP class. self-organization properties of robust knowledge bases.
Accordingly, to quantum algorithm theory and quantum The following levels of those shown in Figure7 were con-
Kolmogorov’s algorithmic complexity, this algorithm is sidered in detail in [21]: level 3 (physical model and objec-
effective. This means that structurally the algorithm has tive of the self-organization control process, and physical
polynomial complexity, i.e. the randomized algorithm interpretation of main operators of the quantum algorithm
is polynomial rather than exponential (classic search al- of KB self-organization control), and level 2 (dynamics of
gorithms) depending on input signals; at the same time, evolution self-organization process).
the limited probability of the accuracy of measuring the
calculation result is enough to an effective decision mak-
ing. The following actions are implemented in QFI model
[25,34,35]
:
(1) the results of the fuzzy inference of each indepen-
dent FC are processed;
(2) based on the methods of the quantum information
theory, valuable quantum information is extracted, hidden
in independent (individual) KB;
(3) in online, a generalized robust output control signal
is projected on all the sets of KB.
In this case, the output signal of QFI in online is the
optimal control signal for changing the coefficient gains
of the PID-controller, which includes the necessary (best)
quality characteristics of the output control signals of each
FC, thereby realizing the principle of self-organization. Figure 7. Hierarchical structure of quantum algorithm of
Therefore, the area of effective functioning of ICS struc- design of self-organization of robust KB’s in ICS
ture can be significantly expanded by including such an On Figure 8 is presented the layouts of robots of the
important characteristic of control quality as robustness. “mobile manipulator” type, which were designed by var-
The robustness of the control signal is the basis for main- ious research groups (HERB: Home Exploring Robotic
taining the reliability and accuracy of control in condi- Butler, STAIR: Stanford Artificial Intelligence Robot and
tions of uncertainty of information or a poorly formalized many others), which are a platform for creating new IT
description of the operating conditions and / or control and software products.
objectives [36].
In natural systems the sought robustness property is
coded in the algorithm of reproduction of the self-orga-
nization process. Therefore, such systems can autono-
mously handle an unforeseen event using different (close,
as regards the idea) approaches: adaptation (learning,
evolution) in the framework of which the system corrects
its behavior in order to handle the event variation; predic-
tion manifested in the fact that the system can “predict”
the change of situation and determine more precisely its
behavior (this property is a special case of adaptation and
does not require that the system estimates the situation
before it occurs); and robustness consisting in the fact that
the system can operate and achieve the objective if contin-
gency perturbations of a certain type occur.
Figure 8. Examples of modern mobile robots with manip-
The robustness property of the KB is achieved by
ulators based on the ROS framework
application of the quantum design algorithm of self-or-
ganization in the course of intelligent control which is One of the most striking examples of developing IT
schematically shown in Figure 7. Figure 7 show the hi- projects is the ROS framework. ROS provides libraries
erarchical levels of the design process using the quantum and tools to help software developers create robot appli-
algorithm, interconnection, and interrelation of the above cations. It provides hardware abstraction, device drivers,

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440 11

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

libraries, visualizers, message-passing, package control,

simultaneous localization and mapping (SLAM) navi-
gation, planning and more. The classic task of a mobile
manipulator is the task of recognizing an object and ma-
nipulating it. In this work, we will consider the mobile
intelligent robot shown in Figure 9.

Figure 11. Decomposition of the intelligent control units

Figure 9. A prototype of a mobile robot with manipulator
designed by the laboratory of intelligent control systems
(INESYS (EFKO Group), Russia)
Achieving a global goal by a mobile robot for service
use with redundant manipulator implies solving the fol-
lowing tasks: pattern recognition in conditions of chang-
ing lighting and noises, controlling a multi-link arm and
trolley, autonomous navigation in the room using laser
rangefinders and stereo vision. The level of robustness of
the control laid down in the solution of individual tasks
affects the quality of interaction of individual modules
and, accordingly, the reliability of the system as a whole.
The block diagram of the mobile robot control system is
shown in the Figure 10.
of the mobile robot control system
Design of an ICS Using the Soft Computing Optimizer
In this section we will consider the process of design
and modeling a control system of a redundant manipulator,
which was performed in the Matlab/Simulink environment.
A fundamental feature in the construction of multi-link
manipulators is modularity, which provides adaptability
and reconfigurability of the dynamic structure in accor-
dance with the task at hand. For a redundant manipulator
the control task can be described as follows:1) ensuring the
specified accuracy of the positioning of the functional de-
vice of the manipulator;2) determination of the spatial con-
figuration of the manipulator links (invariance is ensured
by the redundancy of the number of degrees of freedom),
taking into account unforeseen environmental factors. The
QFI algorithm [19] includes the step of choosing the type
of quantum correlation for constructing the output control
signals. We will consider three types of mixed correlations:
spatial, spatio-temporal, and temporal [37], and in addition,
correlations of various numbers of FC. The dependence of
the output signals is determined by the correlation of the
sets of input coefficients, where each set is an entangled
state. Figure 12 shows the scheme of adding QFI to ICS.

Figure 10. The block diagram of the mobile robot control

system
The decomposition of the intelligent control units of Figure 12. The structure of ICS based on soft and quan-
the mobile robot control system is shown in the Figure 11. tum computing
Here Qref is a reference signal of the navigation and

12 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

pattern recognition system, Q' is a measured variable, Testing the robustness of control system models is car-
E is a control error, s(t) is a control influence limitation, ried out in experiments in standard and unforeseen control
mC(t),mMS(t) are noises in the control and measurement situations. These experiments are described in detail in [37].
system channels, U is a control action, d(t) is delay in MS, We will consider control systems with constant coef-
fCO(t) is external influences on CO, Q is adjustable value. ficients of the PID regulator and ICS based on SCO on
The creation of QFI unit is carried out using the computa- soft computing. A system of quality criteria that takes into
tional intelligence SCO toolkit based on quantum comput- account methods for assessing transients of automatic
ing [19,38]. In the selected configuration of the ICS structure control theory was introduced to evaluate and compare
are implemented seven FC’s, each of which independently the results of testing control systems. These methods have
controls one of the seven links (see Figure 13). been adapted for a specific CO (mobile robot-manipula-
tor). The following criteria were highlighted:
(1) solution of the positioning problem in standard con-
trol situations PTSKCS (Position Task Solution in known
control situations);
(2) solution of the positioning problem in external un-
foreseen control situations PTSACCS1 (Position Task Solu-
tion in the above considered control situations);
(3) solution to the problem of positioning in internal
contingency management situations PTSACCS2;
(4) IT performance;
(5) relative overshoot value σ;
(6) relative error in link positioning at the end of a giv-
en number of iterations ε;
(7) time of one iteration t;
(8) the complexity of the implementation of control P;
Figure 13. ICS of robot-manipulator with seven degrees
of freedom based on SCO (9) overall assessment of FCB management (Full Con-
trol Behavior).
Designations in Figure 13: KPi,KDi,KIi, i = 1,7 are pro- Table 1 shows a comparison of ICS based on SCO with
portional, differential and integral coefficients gain of the quantum computing using spatial correlations with the op-
PID-controller, where i is the number of the corresponding eration of ICS based on SCO with soft computing.
link of the robot-manipulator, TSi is training signal, GAi is
genetic algorithm that generates the training signal for the Table 1. Comparison of ICS based on QCO with ICS
formation of the i-th KB. The standard control situation for based on SCO
the i-th FC is a typical control situation, in the conditions
ICS based on SCO with quantum
of which the training signal TSi is received. ICS designed ICS based on computing
by SCO may contain information about seven control situa- Situation SCO with soft Using the correla- Using the
computing tion of two adjacent correlation
tions (regular or unforeseen) for each of the links. FC of seven FC
The process of integrating QFI unit into intelligent lev- 1 Regular situations 0,923 1,000 1,000
el of the control system is shown in Figure 14, where the External contingen-
2 0,744 0,821 0,821
blocks that are designed using soft and quantum comput- cies
Internal contingen-
ing technologies are highlighted. Further, ICS with QFI 3 0,923 0,846 0,923
cies
will be called SCO based on quantum computing. 4 Performance 0,092 0,477 0,459
5 Overshoot 0,969 0,973 0,971
6 Sustainability 0,911 0,962 0,960
7 Single iteration time 0,973 0,961 0,960
Implementation
8 0,946 0,957 0,986
complexity
9 General control 0,721 0,816 0,826

Using QFI allowed:

Figure 14. Intelligent level of ICS based on SCO with (1) solving the problem of positioning in regular situa-
quantum and soft computing tions;

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440 13

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

(2) improving the results of the positioning task in con-

ditions of external unforeseen situations (under internal
unforeseen situations the results do not change at best);
(3) increase in the criterion Performance by 5 times;
(4) improving the assessment of general management,
the best result is achieved by using spatial correlation of
all seven FC’s.
Figures 15 and 16 show results of simulation.

Figure 16. The movement of manipulator in an external

Figure 15. The movement of manipulator in a standard
control situation: under control of ICS based on SCO with
soft computing (left); ICS based on SCO with quantum
computing (right)
unforeseen situation: under control of ICS based on SCO
(left); ICS based on QCO (right)
Quantum supremacy. We can conclude that ICS based
on QCO (with quantum computing) in the conditions of
a regular and external unforeseen control situation solves
the problem of precision accuracy positioning a manipu-
lator with 7DoF much better (more 2500 times) than ICS
based on SCO (with soft computing). It is also necessary
to present the results of an experiment with the position-
ing of the manipulator under conditions of strong internal
disturbing influences [37].
Example. Consider an example of the influence of

14 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

noise in a measurement system (an example of noise in theory of swarm artificial intelligence [40]. A multi-agent
a measurement system is shown in the Figure 17) on the system is capable of solving complex dynamic tasks for
operation of ICS based on soft and quantum computing. performing collaborative work that does not could be per-
formed by each element of the system individually in a
Noise q1 Noise q2
5
0
5
0
variety of environments without external control, control
-5
0 100 200
Noise q3
300
-5
0 100
Noise q4
200 300 or coordination. In this case, it’s described a net of weakly
5 5
0 0 interconnected robots, working together in order to solve
-5 -5

5
0 100 200
Noise q5
300

5
0 100
Noise q6
200 300
problems that go beyond individual capabilities. Different
0

-5
0 100 200 300
0

-5
0 100 200 300
nodes of such a system, as a rule, have a different level of
5
Noise q7
intellectualization (knowledge, algorithms, computational
bases) and various information resources in the design.
0
-5
0 100 200 300

Each node should be able to modify its behavior depend-

Figure 17. Noise in the measurement system
Figure 18 shows the position of the manipulator during
operation ICS based on soft computing (left) and quantum
computing (right).
Figure 18. The position of manipulator in space with ICS
based on soft computing (left) and quantum computing
(right)
In unforeseen situations with the presence of strong
noise in the measurement system ICS based on QCO with
quantum computing showed an advantage (more 10 times)
over ICS based on SCO with soft computing.
One of the key tasks of modern robotics is the de-
velopment of technologies for the interaction of robotic
systems. Modern approaches to solving this problem are
based on the theory of multi-agent systems [39] and the
ing on the circumstances, as well as plan its communi-
cation and cooperation strategies with other nodes. Here
indicators of the level of cooperation are: the nature of the
distribution of tasks, the combination of various informa-
tion resources and, of course, the ability to solve a com-
mon problem at a given time. The solution of the multi-
agent interaction problem can be found by introducing
the principle of quantum self-organization (in the process
of operation) KB for fuzzy controllers, which is imple-
mented and programmatically supported by the developed
model of QFI using quantum and soft computing method-
ologies and an engineering system - Intelligent System of
Systems Engineering (based on the synergistic principle
of self-organization of knowledge). Multi-agent system
must solve complex dynamic problems of performing
the joint work of various devices. Due to the synergistic
effect, an additional information resource is created in
which the task may not be performed by each element
(agent) of the system separately in various environments
without external control, control or coordination, but the
exchange of knowledge and information allows us to do
useful work together to achieve the management goal un-
der the conditions of uncertainty of the initial information
and restrictions on the consumption of a useful resource
[41]
. A physically synergistic effect is the self-organization
of knowledge and the creation of additional information
that allows a multi-agent system to perform the most use-
ful work with a minimum of loss of useful resource and
the required initial information, without destroying the
lower executive level of the control system. Together with
the informational and thermodynamic laws of intelligent
control (the optimal distribution of qualities of “stability
- controllability - robustness”) ICS is designed for multi-
agent systems, which guarantees the achievement of the
control goal under conditions of uncertainty in the initial
information and a limited useful resource.
The lower level of software and hardware implementa-
tion represents variations of real-time operating systems
and corresponding software in the form of Open source

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440 15

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
solutions frameworks - add-ons, which taking into account
the specific features of the subject area. For example,
the Tango Controls add-in (widely used in mega-nuclear
projects of nuclear physics), or Robotics Operation Sys-
tem (ROS) with real time Linux. By default, parameters
related to the physical features of the implementation of
the system itself, control parameters in feedback loops,
are used as a custom level in such systems installed by ex-
perts or customizable by classical methods. Moreover, the
existing tools of intellectualization have a fairly unified
standard methodological approach and often do not meet
modern requirements for the complexity and required
reliability of the developed systems. One of the main
problems of the effective use of tools and soft computing
technologies (neural networks, evolutionary algorithms,
fuzzy logic) in control problems was the solution of the
following problems:
(1) an objective definition of the type of MF and its pa-
rameters in the production rules in KB;
(2) determination of the optimal structure of FNN in
deep machine learning problems (approximation of the
training signal with the required (given) error and with a
minimum number of production rules in KB);
(3) the use of multicriteria GA in multicriteria control
problems in the presence of discrete restrictions on the pa-
rameters of the CO.
The listed problems were solved and tested on the
basis of SCO software tools using soft computing tech-
nology. The developed intelligent tools made it possible
to design robust KB based on the solution of one of the
algorithmically difficult problems of the theory of artifi-
cial intelligence - the extraction, processing and formation
of objective knowledge without the use of human expert
evaluations.
4. Object Detection and Tracking, Pattern
Recognition, Navigation based on Stereo Vi-
sion and Computational Intelligence
The combined use of various sensors and their combi-
nations with stereo vision technology allows for a more
qualitative and complete construction of the “world
scene” of the robot, thereby improving its interaction with
the environment. The stereo vision technology, which to
some extent repeats the features of the development of
natural vision, allows the on-board system to receive in-
formation not only about the color and brightness of the
object, but also about the distance to it, about its geomet-
ric shape, about obstacles to the object, which plays an
extremely important role in the tasks of a mobile robot.
The intellectualization of the control system, in particular,
16 Distributed under creative commons license 4.0
the use of a neural network approach in the recognition
system, significantly reduced the negative impact of exter-
nal factors on the quality of recognition (recognition error
when changing the angle of the object, changing lighting,
software sensitivity, etc.).
4.1 Convolutional Neural Network Architecture
As the object of recognition was chosen a plastic cork of
an ordinary plastic bottle. With certain image processing,
the module also allows tracking the object. In relation to
the task of controlling a mobile trolley with a manipula-
tor convolutional neural network (CNN) are often used.
The CNN is a multi-layer sensor network, it represents a
further development of the multi-layer perceptron, how-
ever, unlike the latter, the convolutional network has a
much smaller number of weights (the principle of sharing
weights). Quite often the CNN model is divided into two
main parts: the part responsible for the selection of fea-
tures, and the part with which classification is performed
[42]
. Figure 19 shows the structure of the CNN.
Features of the structure of the CNN. The selection of
features of an object occurs using layers of convolution
and subsampling. Each layer is a set of plates of neurons,
which are also called feature maps. The convolution net-
work structure includes two types of layers: convolution
layers and subsampling layers. Convolution and subsam-
pling layers are combined into macro layers (a macro lay-
er is a convolution layer followed by a subsampling layer
[43]
). A set of several architecture-similar macro-layers is
called a feature separator in a sensory neural network.
Each neuron of the convolution layer and the subsample
layer is associated with a receptive field (RF). RF is a
certain square area that includes neurons capable of trans-
mitting signals to a neuron that has a given RF. Figure 20
shows a convolutional layer diagram, and also shows the
process of its interaction with the previous layer.
Figure 19. General structure of the convolutional neural
network
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Figure 20. Scheme of interaction of the convolutional
layer with the previous layer
Figure 20 makes it clear that RF of each neuron of the
convolutional layer is immediately associated with two
feature maps of the previous layer (as an example parallel
processing of each image channel in RGB color format
can be given). A key feature of convolutional layers is that
bonds formed within the same feature map have the same
set of weights. These are the so-called bound weights. Us-
ing the associated weights, certain features are selected in
an arbitrary position on the feature map. The connection
of the convolutional layer card with several cards of the
previous layer provides the opportunity to equally inter-
pret differently presented information [42]. Figure 20 shows
that RF of the neurons intersect (the step is subject to ad-
justment). Reducing the step of applying the RF increas-
es the number of neurons in the map of the next layer.
Features of an object with the help of RF and associated
weights are extracted [44]. If KС is the number of neurons
that make up the RF of the n-th neuron of the convolu-
tional layer, Kernel[k] is the convolution core, b is the dis-
placement of the n-th neuron (b and Kernel [k] retain their
values for the entire map of the convolutional layer), x[n
+ k] is the input for the n-th neuron of the convolutional
layer (k = 0..KC-1), then the convolution operation can be
displayed by the Eq. (5) [42]:
K c −1
p= b + ∑ Kernelk * xn + k (5)
k =0
The weighted sum p is supplied to the input of the acti-
vation function -the response of the neuron is determined
[42]
. The output of the neuron has the following form:
y = f ( p ) (6)
Each neuron of the convolutional layer is a detector
of a certain feature that was isolated during training. The
interaction of convolution with the activation function of
a neuron allows us to assess the degree of presence of a
Distributed under creative commons license 4.0
particular trait in the current RF of this neuron. The con-
volution of an input element with general customizable
parameters is an analogue of passing an image on a map
through some filter [44].
Figure 21 shows the interaction pattern of the subsam-
pling layer with the previous convolutional layer.
Figure 21. The scheme of interaction of the subsample
layer with the previous layer
The main task of the subsampling layer is to reduce the
scale of the processed display obtained using the previous
convolutional layer. Each map of a subsample layer is as-
sociated with only one map of the previous convolutional
layer.
It is important to note that RFs of the neurons of the
subsample layer do not intersect. Configurable parameters
are common to all neurons of each plate. The number of
these parameters is equal to two; it does not depend on the
number of elements included in the RF of these neurons.
Since RFs of the neurons do not intersect, the convolution
p for the n-th neuron of the subsample layer is defined as
follows:
K s −1
p= b + u * ∑ * xn*K s+k (7)
k =0
In Eq. (7) KS is the total number of neurons included in
the RP of the n-th neuron of the subsample layer [43].
The second part of the CNN is a feature classifier. The
classifier, as a rule, is a single-layer or two-layer percep-
tron. The number of neurons in the classifier layer usually
corresponds to the number of classes to which the input
image belongs. There are no associated weights in the
classifier. The weighted sum p for the neuron of the classi-
fier layer can be defined as
K
p= bn + ∑xk * wn , k , (8)
k =1
In Eq. (8) bn is the offset, different for each neuron,
x[k] is the input element, w[n, k] are the custom param-
eters of the n-th neuron (unique to each neuron), K is the
DOI: https://doi.org/10.30564/aia.v2i2.1440 17
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
input size for the classifier layer [42].
Remark. There are many works [45-50] that are devoted
to the creation and training of the CNN. In this work the
recognition system based on stereo vision technology uses
the classical CNN architecture, which includes convolu-
tion and averaging layers. Network training was done with
a teacher. In relation to the recognition problem, a teacher
is the number of a class that is encoded in a vector. This
vector is equal to the size of the output layer of the neural
network. This is the desired result corresponding to this
input pattern. The actual response is obtained as a result
of the reaction of the neural network with the current pa-
rameters on the input pattern. Error signal - the difference
between the desired signal and the current response of the
neural network. It is on the basis of the error signal that
the tunable parameters of the neural network are corrected
[43]
. A significant minus of this training scheme is the great
difficulty in creating training samples, however with a few
number of classes the negative influence of this factor can
be neglected. In the general case a deep machine learning
quantum algorithm is used that contains the quantum su-
perposition operator of all input signals with massive data
processing parallelism [4].
The error function depends on the system settings be-
ing configured. For such a function, one can construct a
multidimensional error surface in the coordinates of free
parameters. In this case the real surface of the error is
averaged over all possible examples, which are present-
ed in the form of input-output pairs. To improve system
performance over time, the error value should shift to the
minimum. This minimum can be both local and global [43].
The most common and reliable methods for achieving a
local or global minimum on the error surface are local op-
timization methods [51-53].
Remark. Many local optimization algorithms can be di-
vided into two classes: local optimization algorithms with
the calculation of partial derivatives of the first order (the
method of steepest descent, the method of one-dimen-
sional and two-dimensional optimization of the objective
function in the direction of the anti-gradient, the method
of conjugate gradients, methods that take into account the
direction of the anti-gradient at several steps of the algo-
rithm, etc. .) and local optimization algorithms with the
calculation of partial derivatives of the first and second or-
ders (Newton's method, optimization methods with sparse
Hessian matrix, quasi-Newton methods, Gauss-Newton
method, Levenberg-Marquardt method). In this work, we
use the classical error back propagation algorithm with
a further generalization to recognition algorithms based
on quantum deep machine learning using quantum neural
networks and quantum GA [54,55]. Quantum computing is a
18 Distributed under creative commons license 4.0
new computational paradigm that promises applications in
several fields, including machine learning. In the last de-
cade, deep learning, and in particular CNN, have become
essential for applications in signal processing and image
recognition. Quantum deep learning, however remains a
challenging problem, as it is difficult to implement non
linearities with quantum unitaries. The quantum CNN
(QCNN) is a shallow circuit, reproducing completely the
classical CNN, by allowing non linearities and pooling
operations. The QCNN is particularly interesting for deep
networks and could allow new frontiers in image recogni-
tion, by using more or larger convolution kernels, larger
or deeper inputs (see Appendix 2).
The main and most important stage in the implemen-
tation of the recognition system based on the CNN is the
stage of formation of the training sample. In a specific case
the training sample included 10 classes of objects (cork
from a plastic bottle, ball, human face, ballpoint pen, etc.).
The total volume of the training sample is 50,000 images
(5,000 per class). The volume of the test sample is 10,000
images. A training sample was formed, partly from person-
ally captured images, partly from images downloaded from
the Internet. All images are pre-processed (32x32, RGB,
resized, mirrored) and divided according to their classifica-
tion. The design, creation and training of CNN were carried
out using the Keras library. The structure of the CNN is set
using this library directly in the program code, which im-
poses questions on the choice of the optimal structure.
At the CNN input, the images are received in matrix
rather than in vector form (which is necessary to save
information about the topology). Input image size 32x32
pixels, format - RGB. The first convolutional layer con-
tains 32 3x3 feature maps (each with its own convolution
kernel), i.e. each convolutional neuron is connected to a
square 3x3 image. The next convolutional layer has a sim-
ilar architecture. It is known that convolution layers and
subsampling layers are responsible for highlighting cer-
tain attributes of various objects (borders, simple colors,
and curves) in images. Deepening into the network (going
through the following layers of convolution and subsam-
pling) allows you to define less abstract (most characteris-
tic of any class of objects) features. The following depen-
dence can be traced - an increase in the number of macro
layers (“Convolution-Subsampling”) makes it possible to
find more and more complex features of certain objects
in the image. The next step after convolution is averag-
ing (subsampling). This operation reduces the dimension
of feature maps obtained from the previous convolution
layer. This method is based on the fact that neighboring
pixels are very slightly different from each other (the so-
called “pixel correlation”). The averaging operation sig-
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
nificantly reduces the dependence of the recognition result
on the scale of the input image, and also significantly
reduces the computational load. A fully connected output
layer of 10 neurons contains the probability that the object
in the analyzed image belongs to a certain class. Con-
volutional neural networks are currently one of the best
application tools for solving recognition and classification
problems. Figure 22 shows that the system correctly rec-
ognizes objects even when lighting is degraded.
Figure 22. The reaction to the lighting shift
4.2 Navigation Module and Building “World
Scene” for a Robot
The navigation module is also an essential part of the mo-
bile robot control system. Using this module, it seems pos-
sible to significantly increase the efficiency of the robot.
The development of the navigation system was carried
out using the ROS framework. The control system recre-
ates a three-dimensional map (world scene) of the room
based on a set of points obtained using a laser rangefinder
and a stereo camera. The objects and obstacles present are
marked on the resulting map, which allows robot to mod-
el its behavior, localize position in space, memorize and
build room maps (SLAM) using the Unified Robot De-
scription Format (URDF). Algorithms for calculating the
motion path are executed at this moment, which allows
the control system to avoid collisions with various objects.
As the robot advances in space, the map is dynamically
supplemented, control actions on the actuators are recal-
culated taking into account deviations from the optimal
Distributed under creative commons license 4.0
trajectory calculated by the planning package. In this case
the movement correction is calculated by PID-controller,
while the control parameters (gain factors) are set in ac-
cordance with the developed KB. The interaction of the
intelligent control modules in the ROS system is provided
by the file system level. In this case it is necessary to cre-
ate a model of a controlled robot and adding this control
model to the ROS navigation system. This is necessary to
establish spatial relationships between the robot and the
objects of the surrounding world. Figure 23 shows a map
of the room obtained using the sensor system.
Figure 23. The functioning of the navigation module
The interaction of the main blocks of the mobile ro-
bot control system (manipulator control system, motion
control system, recognition system) allows us to solve
the problem of detecting an object and interacting with it.
Figure 24 shows that the recognition system selects and
classifies the desired object (bottle cap), and the naviga-
tion system paves the route for the found object.
Figure 24. The solution to the task of controlling a mobile
robot with manipulator
Figure 24 is presented the description of the task of con-
trolling a mobile robot with manipulator from the point of
view of decomposition of control system. Also, the main
functional blocks are highlighted, the areas of application
of intelligent information technologies are shown.
DOI: https://doi.org/10.30564/aia.v2i2.1440 19
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
4.3 Example
The advantages of using SCO toolkit are clearly demon-
strated in an experiment with multi-agent interactions [41].
In this experiment the CO were a robot-manipulator, a
robot-bartender and a robot-inverted pendulum.
The visual channel is the main one for receiving infor-
mation about the surrounding world. Robot receives a lot
of information about the environment using a computer
vision system. The recognition module is an important
part of ICS of a robotic device. The recognition process
for the case of interaction of robots is presented in more
detail in Figure 25.
Figure 25. The recognition system in the task of interac-
tion of robots
A glass is installed on the robot-inverted pendulum,
which is the object of recognition for the stationary ma-
nipulator. The inverted pendulum determines its position
relative to the robot-bartender using an infrared camera
and “beacons”. When the appropriate command is given,
robot-inverted pendulum changes its location and moves
to the manipulator. When the inverted pendulum is in the
visibility range of the stationary manipulator (see Figure
25), recognition system determines the location of the
glass and sends a command to the manipulator fill it with
liquid (see Figure 26).
Figure 26. Robot-bartender pours liquid into a glass
mounted on a robot-inverted pendulum
20 Distributed under creative commons license 4.0
As mentioned above, one of the main purposes of the
effective and practical use of ICS is the possibility of
guaranteed achievement of the control goal with the high-
est quality control and reducing the consumption of useful
system resource. The traditional PID-controller is used in
more than 85% of industrial and non-industrial automatic
control systems (ACS), including facilities with increased
social and economic responsibility. Therefore, one of the
important specific (theoretical and practical) problems in
the creation of ICS is the development of methods and
algorithms to increase the reliability and quality of man-
agement of the executive (lower) level of ACS based on
the traditional PID-controller. One of the difficulties in
developing of ICS for unforeseen control situations is to
solve the problem of designing an appropriate KB using
objective knowledge about the behavior of CO and fuzzy
PID-controllers. The solution of this problem significantly
depends on the availability of the development of an al-
gorithmically solvable, physically/mathematically correct
model and the practical implementation of the process of
extracting, processing and generating objective knowledge
without the participation of an expert. The introduction of
physical and informational restrictions in the formalized
description of CO’s model significantly affects the quality
of the generated KB. Exclusion of these restrictions from
the description of CO’s models leads to a loss of robust-
ness of the designed control laws. Therefore, the main
goal of developing the basis of information technology for
the design of ICS for such a wide class of CO’s is the cre-
ation of a process for designing robust KB in unforeseen
control situations for the executive level, that takes into
account real physical and information limitations in the
production rules of KB.
The possibility of effective application of robust KB’s
design technology for ICS (see. Figure 12) in unforeseen
control situations is based on the idea of using quantum
decision-making strategies in the form of QFI. As a re-
sult, the robustness of ICS is formed using QFI in the
laws of controlling the gain of the PID-controller. The
simulation results (see above - ICS of redundant manip-
ulator) show that the required control quality in unfore-
seen control situations is not achieved when controlling
the FC, while when controlling quantum FC, the control
system has the required control quality. It follows that
from two non-robust FC’s, using quantum self-orga-
nization of knowledge, it is possible to design in real
time a robust FC, the KB of which satisfies both quality
criteria. Therefore, the decomposition of the solution of
the problem of multicriteria optimization of robust KB
in the unforeseen control situation to particular solutions
of optimization subtasks can be physically realized in
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

real time in the form of separate reactions of the corre- bots separately.
sponding individual KB optimized with various fixed
criteria of quality and control situation. Aggregation of 5. Conclusion
the obtained particular solutions in the form of a new ro-
Intensification of the production process, increasing com-
bust KB is carried out on the basis of QFI. QFI contains
plexity and the number of various technical products,
the mechanism for the formation of quantum correlation
increasing the processing power of processors indicate the
between the obtained particular solutions. As a result,
necessity for development of design technologies and the
only reactions of a finite number of individual KB’s
introduction of intelligent systems in the control loop. In
containing extremely achievable control laws in this
turn, it is necessary to have intelligent tools to configure
unforeseen situation are used. The control laws of fuzzy
similar systems in various subject areas. An increasing
PID-controller formed by the new robust KB have a sim-
in the computational basis (types of calculations) of a
pler physical implementation and as a result contain the
software product leads to an increase in the quality and
best characteristics of individual control quality criteria
reliability of control, adding the properties of adaptation
for an unforeseen control situation. It is also necessary
and learning to the feedback loop. However, increasing
to mention that according to the Thermodynamics trade-
the reliability of the control system leads to an increase in
off between stability, controllability, and robustness (see
tunable parameters and, as a result, to increasing the com-
paragraph 2.3), the property of robustness (by its phys-
plexity of tuning such systems. A control system can cope
ical nature) is an integral part of the property of self-or-
with an unpredictable environment autonomously using
ganization. The required level of robustness of ICS is
different but closely related approaches:
achieved by fulfilling the principle of minimum produc-
(1) Adaptation (learning, evolution) - the system chang-
tion of generalized entropy. The principle of minimum
es its behavior to cope with the change.
entropy production in the control system is the physical
(2) Anticipation (cognition) - the system predicts a
principle of optimal functioning with a minimum of use-
change to cope with, and adjusts its behavior accordingly.
ful work. This principle underlies the development of a
This is a special case of adaptation, where the system does
robust control system (which is very important for such
not require experiencing a situation before responding to
complex systems as, for example, a team of robots). The
it.
approach based on QFI (described in this paper) guaran-
(3) Robustness - the system is robust if it continues to
tees the necessary condition for self-organization - the
function in the face of perturbations. This can be achieved
minimum of the required initial information in the train-
with modularity, degeneracy, distributed robustness, or
ing signals. The thermodynamic criterion of the mini-
redundancy.
mum of a new measure of generalized entropy produc-
(4) Modularity: firstly, the approach to incorporating el-
tion provides a sufficient condition for self-organization
ements of intelligent computing into classical control sys-
- the robustness of control processes with a minimum
tems implies non-destruction of the lower executive level
consumption of useful resource.
(which ensures minimal interposal in the CO hardware).
The experiment with multi-agent interaction, in which
Secondly, when working with complex control objects, it
the robot manipulator participated (ICS of this robot is
is necessary: ① to separate the control blocks to reduce
described in this chapter), allows us to draw the follow-
complexity, ② to organize coordination control of the
ing conclusion - the use of SCO based on quantum com-
separated blocks to improve the quality of control, which
puting can significantly increase the robustness of the
QFI does well described on the example of the ICS based
control system. The toolkit allows to configure complex
on SCO on quantum computing of 7DoF manipulator in
control systems with many of control loops; the number
this article.
of input variables and the size of the resulting KB are
Successful self-organizing systems will use combina-
limited only by the hardware characteristics of the com-
tions of these approaches to maintain their integrity in a
puter and CO. The developed interaction options show
changing and unexpected environment. Adaptation will
the possibilities of using these systems in a wide range of
enable the system to modify itself to “fit” better within
tasks, such as automation of warehouses and production
the environment. Robustness will allow the system to
facilities, automatic catering establishments, emergency
withstand changes without losing its function or purpose,
response, etc. The main feature of a multi-agent system
and thus allowing it to adapt. Anticipation will prepare the
is the synergistic effect arising from the combination of
system for changes before these occur, adapting the sys-
several robotic systems, when the result of the functional
tem without it being perturbed. The main components and
interaction far exceeds the capabilities of each of the ro-
their interrelations in the information design technology

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440 21

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
are based on new types of (soft and quantum) computing.
The key point of this information design technology is the
use of the method of eliciting objective knowledge about
the control process irrespective of the subjective experi-
ence of experts and the design of objective KB’s of a FC,
which is principal component of a robust ICS. The devel-
oped SCO toolkit implements mechanisms for creating,
configuring, and transmitting control parameters in the
form of control signals received from KB of FC without
destroying the low executive level. The use of soft com-
puting and developed ICS design technologies reduces the
impact of expert judgment in the training and configura-
tion of ICS.
From computer science viewpoint, QA of QFI model
plays the role of the information algorithmic and SW-plat-
form support for design of self-organization process.
Quantum computing, which ensures robustness of ICS,
introduces the property of self-organization, allows CO
to function effectively in conditions of a lack of a priori
information. The use of this kind of computing can reduce
the influence of the hardware (for example, reducing the
number of sensors) on the effectiveness of the control
system. Technologies for remote configuration and trans-
mission of KB allow the CO to receive KB from the SCO
unit or from other CO, which makes it possible to manage
structurally new objects, such as teams of robots, multi-
agent systems, complex automated production, etc. In
addition, this technology gives CO ability to update and
adapt KB for a specific control situation, including contin-
gency.
The experiments described above demonstrate that the
inclusion of QFI module in ICS based on SCO with soft
computing made it possible to provide a complete solu-
tion to the positioning problem in standard control situa-
tions and in the conditions of external unforeseen control
situations. The QFI module significantly improved the
accuracy index of the positioning of the functional device
under strong internal disturbances, the performance index
improved 5 times, and the relative error in the positioning
of the links decreased. It is important to note that we must
to organize the separation of control during the design of
ICS for complex CO’s (precisely such CO’s appear in the
experiments described above). This is necessary for the
organization of coordination management without signifi-
cantly increasing the complexity of the control system.
However, such a management decomposition often leads
to a mismatch of work and a decrease in the quality of
management. In the example of multi-agent interaction
QFI module, using data from stereo vision module, suc-
cessfully generates a generalized robust control signal.
Moreover, control robustness is achieved even with a
22 Distributed under creative commons license 4.0
limited set of sensors and in the presence of internal and
external disturbing influences.
Appendix 1: Emotional Learning and Its utilization
in Control Engineering [56]. The Limbic System, as part
of the mammalian creatures' brain, is mainly in charge
of the emotional processes. The Limbic System locat-
ed in the cerebral cortex consists mainly of following
components: Amygdala, Orbitofrontal Cortex, Thal-
amus, Sensory Cortex, Hypothalamus, Hippocampus
and some other less important areas. In this section, we
try to describe briefly these main components and their
tasks. Figure A1.1 illustrates the anatomy of the main
components of Limbic System. The first sign of affective
conditioning of the system appears in Amygdala which
is a small almond-shaped in sub-cortical area. This com-
ponent is placed in a way to communicate with all other
Sensory Cortices and areas within the Limbic System.
The Amygdala connections to/from other components
are illustrated in Figure A1.2. The studies show that a
stimulus and its emotional consequences are associated
in the Amygdala area. In this region, highly analyzed
stimuli in the Sensory Cortices, as well as coarsely cat-
egorized stimuli in the Thalamus are associated with an
emotional value.
In a reciprocal connection, the Orbitofrontal Cortex,
as of another component of the brain system, interacts
with the Amygdala. The main interrelated function of
this component is: Working Memory, Preparatory Set and
Inhibitory Control. The current and recent past events
are represented in the Working Memory. The Preparatory
Set is the priming of other structures in anticipation of
impending action. Inhibitory Control is the selective sup-
pression of areas that may be inappropriate in the current
situation. More specifically, the Orbitofrontal Cortex takes
action in omission of the expected reward or punishment
and control the extinction of the learning in the Amyg-
dala. Another component in this area is Thalamus which
lies next to the basal ganglia. It is a non-homogeneous
sub-cortical structure and a way-station between cortical
structures and sub-cortical. Moreover, various parts of the
Thalamus also relay the majority of sensory information
from the peripheral sensory systems to the Sensory Cor-
tices. Particularly, the Thalamic Sensory Inputs going to
the Amygdala are believed to mediate inherently emotion-
ally charged stimuli as well as coarsely resolved stimuli
in general. The Thalamus signal going to the Amygdala
evades the processes involved in the Sensory Cortex and
other components of the system. Therefore, Amygdala re-
ceives a non-optimal but fast stimulus from the Thalamus
which among the input stimuli is often known as a charac-
teristic signal.
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Figure A1.2. Connections of the Amygdala with other

components of the Limbic System
The main issue in using the emotional learning for differ-
ent applications is defining the sensory and emotional sig-
nals in such a way that properly represent the state and ob-
jectives of the system. Some of researchers have developed
intelligent systems based on BELBIC which in this section
some of the designed applications are briefly introduced
and the results of simulation are demonstrated in some
applications. Rouhani and co-workers used BELBIC in a
neuro-fuzzy model of microheat exchanger. First, a locally
linear learning algorithm called Locally Linear Mode Tree
(LoLiMoT) was applied to build the neuro-fuzzy model.
Then, the BELBIC based on PID control was adopted for
the micro-heat exchanger plant. The performance of pre-
sented controller was compared with classic PID controller.
Figure A1.3 and Figure A1.4 show the closed-loop system
response using BELBIC and PID controller respectively.
As shown the performance of the system using BELBIC is
Figure A1.1. The major brain structures associated with much better than that of PID controller.
the Limbic System
The next component is the Sensory Cortex close to the
Thalamus which receives its input from the latter one. In
fact, Sensory Cortex processes the information from the
sensory areas. The Sensory Cortex sends highly analyzed
input to the Amygdala and Orbitofrontal. Generally, the
mammalians use these areas of their Limbic System for
higher perceptual processing. Below the Thalamus, lies
another component named Hypothalamus which is ap-
parently in charge of regulation of the endocrine system,
the autonomous nervous system and primary behavioral Figure A1.3. Closed-loop system response using BELBIC
surviving states. The lateral region of Hypothalamus is with LoLiMoT identifier
connected to various regions of the Amygdala and vice
versa. The connections are believed to have a major role
in motivational control of the Structures within the Hy-
pothalamus. Furthermore, one of the most complex and
twisting components of the Limbic System is Hippocam-
pus which is located in the same area as the Amygdala.
Its main role is the mapping of the environment based
on environmental cue. The Hippocampus has other func-
tions such as spatial navigation, laying down of the long-
term memory and formation of the contextual represen-
tations. Figure A1.4. Closed-loop system response using PID with
LoLiMoT identifier

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440 23

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Jafarzadeh et al. proposed an intelligent autopilot for a
2-DoF helicopter model based on BELBIC. The majority of
previous systems were based on the linearization model or
through several linearization techniques for helicopter that
made the proposed controls unreliable. The designed model
used a BELBIC controller and feedback linearization tech-
nique to a nonlinear model of a helicopter. In this method, the
states of the system have been separated into two parts, and
each part has been controlled by one of the control inputs.
The performance of the two mentioned controllers simulated
the in Simulink. The simulation results of controller system
by BELBIC controller and feedback linearization controller
have been demonstrated in the Figure A1.5, and in Figure
A1.6, the control inputs of the system have been shown.
Figure A1.5. Height (left) and Collective rotor blade
angle (right) of helicopter (Solid: set point, Dashed: feed-
back linearization, Dotted: BELBIC)
Figure A1.6. First (left) and second (right) control input
of helicopter (Solid: feedback linearization, Dotted: BEL-
BIC)
It can be seen from these simulations that the tracking
performance of BELBIC controller for the height is better
than Feedback linearization controller, but in the sense
of steady state the performance of both controllers is sat-
isfactory. However, stability guarantee is an important
drawback for this controller.
An intelligent control based on BELBIC has been in-
troduced for speed and flux control of an induction motor.
It was a novel and simple model of induction motor drives
control which controlled motor speed and flux accurately,
without needing to use any conventional controllers and
independent of motor parameters. In order to evaluate
this emotional controller, digital computer simulations
have been performed using Matlab/Simulink. The results
showed that the emotional controller had some gains,
which gave good freedom for choosing desired responses
24 Distributed under creative commons license 4.0
in terms of overshot, settling time, steady state error and
smoothness. These made the controller effective and flex-
ible in high performance applications. Moreover, Simple
structure, fast auto learning and high tracking potency of
BELBIC have been made to present a new control plant
that is independent of motor parameters and controls
speed and flux simultaneously.
As a demonstration of the performance of the BELBIC
controller in real-world applications beyond numerical
simulation, it will be implemented on a KUKA Light-
weight Robot (LWR), a 7DoF KUKA LWR4+ [57]. Bio-in-
spired by the human arm and with a payload of 7 kg and
its 7 axes, all equipped with internal position as well as
force-/torque-sensors, this redundant robot offers a range
of features which are essential for the considered appli-
cation. The seven revolute joints of the industrial robot
are driven by brushless motors via harmonic drives. The
working envelope is described in Figure A1.7.
Figure A1.7. Axes-nomination of the KUKA LWR4+
(KUKA, 2012) (left); Description of the work envelope of
the KUKA LWR4+ (right)
The scientific work [57] is dedicated to the tasks of
namely path following and position control. The tracking
of complex freeform-trajectories by robotic manipulators
is essential to many manufacturing processes like grind-
ing, welding, polishing or gluing. Besides pick-and-place
operations, path following is the most common type of au-
tomation tasks. With conventional controllers satisfactory
performance is obtained for basic constrained motions and
therefore they are still widely used in industry. The use of
these conventional control schemes is however restricted
to robotic manipulators with well-known dynamic and
kinematic parameters following rather simple continuous
paths in a disturbance-free environment. The desire to
extend position control to robotic manipulators with un-
known parameters following discontinuous freeform paths
in the presence of disturbances explains the interest in the
trajectory tracking control problem.
As a verification of the performance of the suggested
control scheme, the controller is evaluated through simula-
tion in a Matlab/Simulink-environment. For the simulation
a two-link planar robotic manipulator with revolute joints.
Excellent simulation-results were obtained. Robot-external
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
disturbances are added as time-dependent 2-dimensional
function. Internal uncertainties are added in the form of
maximal +/-10% deviations from the dynamic parame-
ters-values. Switching constraints are implemented in the
form of a desired trajectory switching between different
curved and straight segments as shown in Figure A1.8. The
desired continuous trajectory is not differentiable.
Figure A1.8. Trajectory to be tracked by the tool on the
workpiece surface in the simulated application; the arrows
indicate the movement-direction
Table A1.1 shows the performance of the three considered
controllers for trajectory tracking of discontinuous freeform
paths, exhibited on the example shown in Figure A1.8.
Table A1.1. Absolute maximal, minimal and mean posi-
tion errors for both manipulator-links using PID-, Com-
puted Joint Torque (CJT)-control and BELBIC-SMC
In order to demonstrate the efficiency of the suggested
controller in real-world applications beyond numerical
simulation, an experimental validation is performed. The
first experiment is a goal-reaching task. The task is about
moving the robot’s joints consecutively to a specified goal
position. Table A1.2 shows that the BELBIC-SMC-concept
outperforms PID- and Computed Joint Torque-control.
Table A1.2. Absolute maximal, minimal and mean posi-
tion errors for all manipulator-links using PID-, Comput-
ed Joint Torque (CJT)-control and BELBIC-SMC when
reaching a desired goal position
Distributed under creative commons license 4.0
The simulation verifies the tracking performance of
the BELBIC-SMC-controller for the chosen case. The
BELBIC-SMC-controller outperforms both, the PID-
and the Computed Joint Torque-controller. The free-
dom-shape and especially the discontinuity of the path
deteriorate the tracking performance of the conventional
controllers.
Appendix 2: quantum computing in deep convolu-
tional neural networks [58]. The growing importance of
deep learning in industry and in our society will require
extreme computational power as the dataset sizes and the
complexity of these algorithms are expected to increase.
Quantum computers are a good candidate to answer this
challenge. The recent progress in the physical realization
of quantum processors and the advances in quantum al-
gorithms increase more than ever the need to understand
their capabilities and limits. In particular, the field of
quantum machine learning has witnessed many innovative
algorithms that offer speedups over their classical coun-
terparts. Quantum deep learning, the problem of creating
quantum circuit that enhance the operations of neural net-
works, has been studied in several works. It however re-
mains a challenging problem as it is difficult to implement
non linearities with quantum unitaries. CNN are a type of
deep learning architecture well suited for visual recogni-
tion, signal processing and time series. In this work we
propose a quantum algorithm to perform a complete con-
volutional neural network (QCNN) that offers potential
speedups over classical CNNs.
The adaptation of the CNNs to the quantum setting
implies some modifications that could alter the efficien-
cy of the learning or classifying phases. In work [58] were
presented some experiments to show that such modified
CNNs can converge correctly, as the original ones. The
experiment, using the PyTorch library, consists of train-
ing classically a small convolutional neural network for
which was added a “quantum” sampling after each con-
volution. In the following results, we can see that r quan-
tum CNN is able to learn and classify visual data from
the widely used MNIST dataset. This dataset is made
of 60.000 training images and 10.000 testing images of
handwritten digits. Each image is a 28x28 grayscale pix-
els between 0 and 255 (8 bits encoding), before normal-
ization.
Let’s first observe the “quantum” effects on an image
of the dataset. In particular, the effect of the capped non
linearity, the introduction of noise and the quantum sam-
pling (see Figure A2.1).
DOI: https://doi.org/10.30564/aia.v2i2.1440 25
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

results of the QCNN when applied on the test set (10.000

images). They distinguish to use cases: in Table 4 the
QCNN has been trained quantumly as described in this pa-
per, whereas in Table 5 they first have trained the classical
CNN, then transferred the weights to the QCNN only for
the classification. This second use case has a global run-
ning time worst than the first one, but we see it as another
concrete application: quantum machine learning could
be used only for faster classification from a classically
Figure A2.1. Effects of the QCNN on a 28x28 input im- generated model, which could be the case for high rate
age
classification task (e.g. for autonomous systems, classifi-
Note: From left to right: original image, image after ap- cation over many simultaneous inputs). Are presented the
plying a capReLu activation function with a cap C at 2.0, test loss and accuracy for different values of the sampling
introduction of a strong noise during amplitude estimation ratio σ, the amplitude estimation error ϵ, and for the back-
with = 0.5, quantum sampling with ratio σ = 0.4 that sam- propagation noise δ in the first case. The cap C is fixed at
ples the highest values in priority. The useful information 10. These values must be compared to the classical CNN
tends to be conserved in this example. The side gray scale classification metrics, for which the loss is 0.129 and the
indicates the value of each pixel. Note that during the accuracy is 96.1%. Note that researchers used a relatively
QCNN layer, a convolution is supposed to happen before small CNN and hence the accuracy is just over 96%, low-
the last image but we chose not to perform it for better vi- er than the best possible accuracy with larger CNN.
sualization.
I. Kerenidis et al. present the full simulation of our Table A2.1. QCNN trained with quantum backpropaga-
quantum CNN. In the following, they use a simple net- tion on MNIST dataset. With C = 10 fixed
work made of 2 convolution layers, and compare quantum
CNN to the classical one. The first and second layers are
respectively made of 5 and 10 kernels, both of size 7x7. A
three-layer fully connected network is applied at the end
and a softmax activation function is applied on the last
layer to detect the predicted outcome over 10 classes (the
ten possible digits). Note that researchers didn’t introduce
pooling, being equivalent between quantum and classical
algorithms and not improving the results on our CNN. The
objective of the learning phase is to minimize the loss func-
tion, defined by the negative log likelihood of the classifi-
cation on the training set. The optimizer used was a built-
in Stochastic Gradient Descent. Using PyTorch, researchers
have been able to implement the following quantum effects
(the first three points are shown in Figure A2.1): Table A2.2. QCNN created from a classical CNN trained
(1) The addition of a noise, to simulate the approxima- on MNIST dataset. With δ = 0.01 and C = 10 fixed
tion of amplitude estimation during the forward quantum
convolution layer, by adding gaussian noise.
(2) A modification of the non-linearity: a ReLu function
that becomes constant above the value T (the cap).
(3) A sampling procedure to apply on a tensor with a
probability distribution proportional to the tensor itself,
reproducing the quantum sampling with ratio σ.
(4) The addition of a noise during the gradient descent,
to simulate the quantum backpropagation, by adding a
gaussian noise centered on 0 with standard deviation δ,
multiplied by the norm of the gradient.
In the following researchers report the classification
QCNN is able to learn despite the introduction of noise,

26 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1440

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
tensor sampling and other modifications. In particular it
shows that only a fraction of the information is meaning-
ful for the neural network, and that the quantum algorithm
captures this information in priority. This learning can be
more or less efficient depending on the choice of the key
parameters. For reasonable values of these parameters, the
QCNN is able to converge during the training phase. It
can then classify correctly on both training and testing set,
indicating that it does not overfit the data. The learning
curves sometimes present a late start before the conver-
gence initializes, in particular for small sampling ratio.
This late start can be due to the random initialization of
the kernel weights, that performs a meaningless convo-
lution, a case where the quantum sampling of the output
is of no interest. However it is very interesting to see that
despite this late start, the kernel can start converging once
they have found a good combination. Overall, it is possi-
ble that the QCNN presents some behaviors that do not
have a classical equivalent. Understanding their potential
effects, positive or negative, is an open question, all the
more so as the effects of the classical CNN’s hyperparam-
eters are already a topic an active research.
In work [59] demonstrates the potential of Quan-
tum-Classical Convolutional Neural Networks (QCCNN)
by applying it to the Tetris dataset. Researchers create a
Tetris image dataset that consists of 1000 grey-scale im-
ages with shape 3×3, in which each grey-scale image is a
simulated Tetris brick.
Figure A2.2. (a) Hybrid quantum-classical Convolutional
Neural Network (QCCNN); (b) Details of of parametric
quantum circuit design, which is made of interlaced sin-
gle-qubit layer and two-qubit layers
Remark. On Figure A2.2 the input demonstrated here is
a two-dimensional array, which is sent to a quantum con-
volutional layer of 6 filters. Each filter takes a 2x2 window,
translating it into a separable 4-qubit quantum state, and
evolves this state with a parametric quantum circuit. After
Distributed under creative commons license 4.0
that a correlational measurement is made on the output
quantum state and a scalar is obtained. Gathering the scalar
outputs, the final output of the quantum convolutional layer
is a 3-dimensional array. Then a pooling layer is used to
reduce the dimensionality of the data. This process could be
repeated and finally ends with a fully connected layer. The
single-qubit layer consists of Ry gates, each containing one
tunable parameter. The two-qubit layer consists of CNOT
gates on nearest-neighbour pairs of qubits
Concretely, the foreground pixels are represented by
random floating numbers ranging from 0.7 to1, whereas
the background are small floating numbers ranging from
0 to 0.1. There are 5 labels, namely S, O, I, T and L, each
of which represents a type of Tetris bricks. The dataset is
further processed by randomly splitting into a training set
and a testing set that contain 80% and 20% of the images,
respectively. This QCCNN was compared with CNN with
two particular structures, namely one with a single convo-
lutional layer and another with two convolutional layers. To
see the performances with a different number of labels, re-
searchers create another dataset by only picking the two la-
bels S, T out of the original training and testing data. For the
single-layer structure, they use a single (quantum) convo-
lutional layer with 5 filters with no padding, plus a pooling
layer also with no padding. For the two-layer structure, they
use two (quantum) convolutional layers with 2 and 3 filters
respectively, plus a pooling layer with padding 1. The win-
dow shape for all the layers is 2 × 2, and the stride value s =
1. Therefore, the number of qubits fed to the quantum filter
is 4, and the depth of the parametric quantum circuit is set
as 4. During 1000 iterations, it was computed the accuracy
on the testing data and store the values of the loss function,
which is chosen as mean square loss. In Figure A2.3 (a, c),
researchers plot the accuracy and loss values for the 2-label
case. While in Figure A2.3(b, d) they plot the accuracy and
loss values for the 5-label case.
Figure A2.3. Accuracy and loss as a function of the num-
ber of iterations
Note: In all the figures the blue line represents the result of one-layer
DOI: https://doi.org/10.30564/aia.v2i2.1440 27
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
CNN and the black line represents the result of two-layer CNN, the blue
dashed line represents the result of one-layer QCCNN and the black
dashed line represents the result of two-layer QCCNN. The results are
averaged over 10 random simulations. (a) Accuracy in case of 2 labels. (b)
Accuracy in case of 5 labels. (c) Loss in case of 2 labels. (d) Loss in case
of 5 labels.
We can see that QCCNN can reach almost 100% ac-
curacy for both the two structures they have used, and it
can reach much lower loss values for both cases compared
to its classical counterpart. Benefiting from the high-di-
mensional nature of the quantum system, the advantages
of QCCNN become more transparent when the number
of labels increases from 2 to 5. We can also see that the
5-label case takes more iterations to converge than the
2-label case, and that QCCNN with a two-layer structure
converges faster than the single-layer structure, especially
in the 5-label case, which indicates that for complex prob-
lems, better performance could be achieved by deeper
architectures. In summary, researchers present a hybrid
quantum-classical Convolutional Neural Network which
could be used to solve real world problems with current
quantum computers. As a quantum machine learning al-
gorithm inspired by classical CNN, QCCNN keeps the
features of CNN such as the nonlinearity, locality of the
convolutional layer, as well as extensibility to deep struc-
tures. Moreover, the generalized feature map with a para-
metric quantum circuit is able to explore the correlations
of neighbouring data points in a exponentially large linear
space, hopefully allowing this algorithm to capture the
patterns in the dataset more efficiently or precisely with a
quantum computer.
In [60], authors have discussed the implementation of
the fast Fourier transform (FFT) as a quantum circuit.
The quantum version of the FFT (QFFT) is defined as
a transformation of a tensor product of quantum states.
The QFFT has been constructed by a combination of
several fundamental arithmetic operators such as an
adder, subtractor and shift operators which have been
implemented into the quantum circuit of the QFFT
without generating any garbage bits. One of the ad-
vantages of the QFFT is due to its high versatility: the
QFFT are applicable to all the problems that can be
solved by the conventional FFT. The frequency domain
filtering of digital images is one of the possible appli-
cations of the QFFT. The major advantage of using the
QFFT lies in its quantum superposition: multiple im-
ages are processed simultaneously (see Figure A2.4 as
a conceptual image of quantum parallelism). It is even
superior to the QFT when the number of images is suf-
ficiently large. Thus, the QFFT will be a great help in
the industries including security, healthcare, and mar-
keting research, where numerous images are involved.
28 Distributed under creative commons license 4.0
In particular, we can create a high pass filter for edge
detection (Figure A2.4).
Figure A2.4. A conceptual image of the high pass filter
applied to quantum multiple images
There are several issues remaining to be solved. The
first one is how to effectively generate a quantum super-
position of multiple data sets. Though, in principle, one
can generate it using the Hadamard gate and the Pauli-X
gate, it is still a difficult problem to superpose arbitrary
data sets with arbitrary probability amplitudes. The second
is how to use the resultant multiple data sets obtained by
performing the QFFT. The QFFT sustains all the informa-
tion of Fourier coefficients until the moment the quantum
state is measured. If the quantum state that contains the
Fourier coefficients of multiple data sets was passed on to
some quantum device directly and there were some prop-
er techniques to handle it, it would play a key role in the
field of quantum machine learning.
References
[1] Sergei V. Ulyanov, Kazuo Yamafuji, Valery G. Gra-
detsky, Andrea Pagni. Expert fuzzy-neuro controller
design for wall climbing robot for decontamination
of nuclear-power station[J]. 1995, 7(1): 75-85.
[2] Ulyanov S.V., Yamafuji K., Fukuda T. Development
of intelligent mobile robots for service use and mo-
bile automation systems including wall climbing
robots: Pt. 1: Fundamental design principles and
motion models[J]. Intern. J. of Intelligent Mecha-
tronics, 1995, 1(3): 111-143.
[3] Takayuki Tanaka, Junji Ohwi, Ludmila V. Litvintse-
va, Kazuo Yamafuji, Sergei V. Ulyanov. Intelligent
control of a mobile robot for service use in office
buildings and its soft computing algorithms[J].
Journal of Robotics and Mechatronics, 1996, 8(6):
538-552.
[4] Ulyanov S. V. Quantum fast algorithm computa-
tional intelligence Pt I: SW/HW smart toolkit [J].
Artificial Intelligence Advances, 2019, 1(1): 18-43.
[5] T. Tanaka, J. Ohwi, L.V. Litvintseva, K. Yamafuji,
S.V. Ulyanov. Soft computing algorithms for intel-
ligent control of a mobile robot for service use Pt
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
II: Path planning, navigation and technology opera-
tions[J]. Soft Computing, 1997, 1: 99-106.
[6] Ulyanov S. V., Litvintseva L. V., Sorokin S. V.
Robastnost’ baz znanij samoorganizujuschihsya
intellectual’nih regilyatorov v nepredvidennih sit-
uatsiyah upravlenija[J]. Fuzzy Systems and Soft
Computing, 2008, 3(4): 49-73.
[7] R. Mohammadi, C. Lucas and B. N. Araabi, A novel
controller for a power system based BELBIC, Pro-
ceedings of World Automation Congress, 2004, 18:
409-420.
[8] Litvintseva L. V., Ulyanov I. S., Ulyanov S. V., Uly-
anov S. S. Quantum Fuzzy Inference for Knowledge
Base Design in Robust Intelligent Controllers[J].
Journal of Computer and Systems Sciences Interna-
tional, 2007, 46(6): 908-961.
[9] Ulyanov S.V. System and method for control us-
ing quantum soft computing)[P]. US Patent No
7,383,235 B1, 2003; EP PCT 1 083 520 A2, 2001;
Efficient simulation system of quantum algorithm
gates on classical computer based on fast algo-
rithm[P]. US Patent No 883 2006/0224547 A1,
2006.
[10] Ulyanov, S. V. Intelligent Robust Control System
Based on Quantum KB-Self-organization: Quantum
Soft Computing and Kansei / Affective Engineering
Technologies[J]. Springer International Publishing,
2014: 37-48.
[11] Ohwi J., Ulyanov S.V., Yamafuji K. GA in contin-
uous space and fuzzy classifier system for opening
a door with a manipulator of mobile robot: New
Benchmark of evolutionary intelligent comput-
ing[J]. Robotics and Mechatronics, 1996, 8(3): 297-
301.
[12] Fujii S., Ulyanov S.V. A model for motorcycle rider
operation based on genetic algorithm [J]. Yamaha
Motor Technical Review, 2004.
[13] Hagiwara T., Ulyanov S.V. An application of a
smart control suspension system for a passenger car
based on soft computing[J]. Yamaha Motor Techni-
cal Review, 2003.
[14] Ulyanov S.V. Intelligent self-organized robust
control design based on quantum/soft computing
technologies and Kansei Engineering[J]. Computer
Science J. of Moldova, 2013, 21, 2(62): 242 - 279.
[15] Ulyanov S.V. Self-organizing quantum robust con-
trol methods and systems for situations with uncer-
tainty and risk [P]. Patent US 8788450 B2, 2014.
[16] Litvintseva L. V., Ulyanov S. I., Ulyanov S. V.,
Ulyanov S. S. Kvantoviy nechetkiy vyvod dlya
sozdanija baz znaniy v robastnih intellektualnyh
regulyatorah[J]. Izvestiya RAN. Teoriya I sistemy
Distributed under creative commons license 4.0
upravlenija, 2007, 6: 71-126.
[17] Litvintseva L. V., Ulyanov S. V. Intellektualnije
sistemy upravlenija. Kvantovije vychislenija I algo-
ritm samoorganizatsii[J]. Izvestiya RAN. Teoriya I
sistemy upravlenija, 2009, 6: 102-141.
[18] Nielsen M.A., Chang I.L. Quantum Computation
and Quantum Information. Cambridge Univ. Press.
UK, 2000.
[19] Ulyanov S. V., Sorokin S. V., Litvintseva L. V. Opti-
mizator baz znaniy na osnove kvantovyh vychislenij
dlya proektirovanija samoorganizujuschihsja nech-
etkih reguljatorov: programmnyj instrumentarij[J].
System analyze in sience and education, 2012, 1.
[20] A.J. van der Schaft. Theory of port-Hamiltonian
systems (Lectures 1, 2, and 3), Network Modeling
and Control of Physical Systems. DISC, 2005.
[21] Ulyanov S.V. “Self-organized control system,” US
patent, No. 6, 411, 944, 1997; see also “Self-organi-
zation fuzzy chaos intelligent controller for robotic
unicycle: A new benchmark in AI control,” Pro-
ceedings 5th Intelligent System Symposium, Tokyo,
Sep. 29 -30, 1995: 41-46.
[22] Haddad W. M., Chellaboina V., Nersesov S. G.
Thermodynamics: A Dynamical Systems Approach,
Princeton Series in Applied Mathematics Princeton.
NJ: Princeton University Press, 2005.
[23] Ulyanov S.V. System and method for stochastic
simulation of nonlinear dynamic systems with high
degree of freedom for soft computing application.
US patent, No 0039555 A1, Pub. Date: Feb. 26,
2004.
[24] Demetriusa L. Manke T. Robustness and network
evolution - an entropic principle. Physica, 2005,
A346: 682-696.
[25] Ulyanov S.V., Litvintseva L.V. Design of self-orga-
nized intelligent control system based on quantum
fuzzy inference: Intelligent system of systems en-
gineering approach[R]. Proc. of IEEE int. conf. on
system, man and cybernetics (SMC’2005). Hawai,
USA, 4.
[26] Reshetnikov A.G., Ulyanov S.V. Knowledge extract
method from physically measured teaching signal:
knowledge base design for fuzzy controller. System
analyze in sience and education, 2013, 1.
[27] Zurek W.H. Probabilities from entanglement, Born’s
rule from envariance. Phys. Review, 2005, A71(5).
[28] Goldenblat I.I., Ulyanov S.V. Introdaction to rela-
tivity theory and its applications to a new technique.
Moscow, Fizmatgiz Publ., 1979.
[29] Petrov B.N., Ulanov G.M., Ulyanov S.V. Control
problems of relativistic and quantum dynamic sys-
tems. Moscow, Nauka Publ., 1982.
DOI: https://doi.org/10.30564/aia.v2i2.1440 29
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
[30] Li Y., Ang K.H., Chong G.C.Y. Patents, software
and hardware for PID control: an overview and
analysis of the current art. IEEE Control Syst. Mag,
2006, 26(1): 42-54.
[31] Nielsen M.A., Chuang I.L. Quantum Computation
and Quantum Information. - Cambridge Univ. Press,
UK, 2000.
[32] Marinescu D.C., Marinescu G.M. Approaching
quantum computing. Pearson Prentice Hall, USA,
2005.
[33] Ulyanov S.V., Litvintseva L.V., Ulyanov S.S. et all.
Quantum information and quantum computational
intelligence: Backgrounds and applied toolkit of
information design technologies. Milan: Note del
Polo (Ricerca), Universita degli Studi di Milano,
2005, 78-86.
[34] Ulyanov S.V., Litvintseva L.V., Ulyanov S.S. et all.
Self-organization principle and robust wise control
design based on quantum fuzzy inference[R]. Proc.
of Internat. Conf. ICSCCW’2005, Antalya. Turkey,
2005.
[35] Litvintseva L.V., Ulyanov S.V., Takahashi K. et all.
Design of self-organized robust wise control sys-
tems based on quantum fuzzy inference[R] Computer Vision and Pattern Recognition (CVPR),
Proc. of World Automation Congress (WAC’2006):
Soft computing with Industrial Applications (ISS-
CI’2006). Budapest, Hungary, 2006, 5.
[36] Litvintseva L.V., Ulyanov S.V., Ulyanov S.S. Pos-
troenije robastnih baz znanij nechetkih regulyatorov
dlya intellektual’nogo upravlenija suschestvenno
nelinejnymi dinamicheskimy sistemamy[J]. Izvesti-
ya RAN. Teoriya I sistemy upravlenija, 2006(5).
[37] Nikolaeva A. V., Ulyanov S. V. Proektirovanije
intellectual’noj sistemy upravlenija robotom manip-
ulyatorom. Ch. 3: Modelirovanije I fizicheskij ek-
speriment na osnove tehnologii kvantovyh vychis-
lenij[J]. System analysis in sience and education,
2013, 1.
[38] Ulyanov S. V., Litvintseva L.V., Mishin A. A., So-
rokin S. V. Kvantovyj optimizator robastnyh baz
znanij dlya proektirovanija intellektual’nyh system
upravlenija: RF patent №2011616869. Pub. Date
5.09.2011 (Russia).
[39] Tarasov V. B. Ot mnogoagentnyh system k intellek-
tul’nym organisatsyjam. Moscow, 2002: 352.
[40] Lebedev B. K., Lebedev V. B. Planirovanije na os-
nove roevogo intellekta I geneticheskoj evolutsyi
[J]. Izvestija Juzhnogo federal’nogo universiteta,
2009, 93(4).
[41] Ulyanov S. V., Reshetnikov A. G. Synergetics of
information-cognitive interaction in intelligent ro-
botic systems with remote knowledge exchange[J].
30 Distributed under creative commons license 4.0
Programmnije produkti i sistemy, 2017, 4: 593-600.
[42] Nemkov R. M. Development of neural network al-
gorithms for invariant pattern recognition: disserta-
tion of the candidate of technical sciences: 05.13.18.
Stavropol’, 2015: p.
[43] Haykin S.. Neural networks: full course. Moscow,
2008: 113+281-330.
[44] Gonzalez R. Digital image processing. Moscow,
2012: p.
[45] Ranzato M. A., Jarrett K., Kavukcuoglu K., LeCun
Y. What is the best multi-stage architecture for ob-
ject recognition? In ICCV, 2009.
[46] Richard S. Sutton. Reinforcement training. Mos-
cow, 2011: 399.
[47] Ackley D. H., Hinton G. E., Sejnowski T. J. A learn-
ing algorithm for Boltzmann machines. Cognitive
Science, 1985, 9: 147-169.
[48] Bengio Y. Learning deep architectures for AI. Foun-
dations and Trends in Machine Learning, 2009,
2(1): 1-127.
[49] Ciresan D., Meier U., Schmidhuber J. Multicolumn
Deep Neural Networks for Image Classification[R].
In Proceedings of the 2012 IEEE Conference on
CVPR '12, pp.3642-3649, Washington, DC, USA,
2012. IEEE Computer Society.
[50] Ranzato M., Huang F., Boureau Y., LeCun Y. Un-
supervised learning of invariant feature hierarchies
with applications to object recognition[R]. In Pro-
ceedings of the Computer Vision and Pattern Rec-
ognition Conference (CVPR’07), IEEE Press, 2007.
[51] Gill F. Practical optimization. Moscow, 1985: 509.
[52] Izmajlov A. F. Numerical optimization methods.
Moscow, 2005: 304.
[53] Tarkhov D. A. Neural network models and algo-
rithms. Moscow, 2014: 352.
[54] Sergey Ulyanov, Andrey Reshetnikov, Nikita Ry-
abov. Deep machine learning and pattern/face rec-
ognition based on quantum neural networks and
quantum genetic algorithm[R]. Distributed Comput-
ing and Gridtechnologies in Science and Education.
Book of abstract of the 8th International Conference
Dubna, 10 - 14 September, 2018: 38.
[55] Ulyanov S. V., Petrov S. P. Kvantovoe raspozna-
vanije lits i kvantovaja visual’naja kriptografija:
modeli i algoritmi[J]. System analyze in sience and
education, 2012(1).
[56] Zahra Beheshti, Siti Zaiton Mohd Hashim. A Re-
view of Emotional Learning and It’s Utilization in
Control Engineering. Int. J. Advance. Soft Comput.
Appl., 2010, 2(2): 191-208.
[57] Klecker S. Robotic Trajectory Tracking: Position-
DOI: https://doi.org/10.30564/aia.v2i2.1440
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
and Force-Control: PhD dissertation (Sciences de
l’ingénieur). Université du Luxembourg, Luxem-
bourg, 2019: 122.
[58] Kerenidis I., Landman J., Prakash A. Quantum Al-
gorithms for Deep Convolutional Neural Networks.
Universit´e Paris Diderot, Paris, France, 2019: 40.
[59] Junhua Liu, Kwan Hui Lim, Kristin L. Wood, Wei
Huang, Chu Guo, Heliang Huang. Hybrid Quan-
Distributed under creative commons license 4.0
tum-Classical Convolutional Neural Networks.
Quantum Intelligence Lab, Supremacy Future Tech-
nologies, Guangzhou, China, 2019: 5.
[60] Ryo Asaka, Kazumitsu Sakai, Ryoko Yahagi. Quan-
tum Circuit for the Fast Fourier Transform. De-
partment of Physics, Tokyo University of Science,
Kagurazaka 1-3, Shinjuku-ku, Tokyo, Japan, 2019:
16.
DOI: https://doi.org/10.30564/aia.v2i2.1440 31
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Artificial Intelligence Advances

https://ojs.bilpublishing.com/index.php/aia

ARTICLE
Intelligent Robust Control of Redundant Smart Robotic Arm Pt II:
Quantum Computing KB Optimizer Supremacy
Alena V. Nikolaeva1,2　Sergey V. Ulyanov3,4*
1. Dubna State University, Russia
2. INESYS LLC (EFKO GROUP), Russia
3. Physical and Mathematical Sciences; JINR, Russia
4. LIT JINR, INESYS LLC (EFKO GROUP), Russia

ARTICLE INFO ABSTRACT

Article history This article is a continuation of the work “Intelligent robust control of
Received: 7 November 2019 redundant smart robotic arm Pt I: Soft computing KB optimizer - deep
machine learning IT”. In the first part of the paper, we examined control
Accepted: 24 April 2020 systems with constant coefficients of the conventional PID controller
Published Online: 30 June 2020 (based on genetic algorithm) and intelligent control systems based on soft
computing technologies. For demonstration, MatLab / Simulink models
Keywords: and a test benchmark of the robot manipulator demonstrated. Advantages
Quantum computing supremacy and limitations of intelligent control systems based on soft computing
technology discussed. Intelligent main element of the control system
Quantum-classical correlation based on soft computing is a fuzzy controller with a knowledge base in
Knowledge base it. In the first part of the article, two ways to implement fuzzy controllers
Fuzzy controller showed. First way applyied one controller for all links of the manipulator
and showed the best performance. However, such an implementation is
Quantum fuzzy inference
not possible in complex control objects, such as a manipulator with seven
degrees of freedom (7DOF). The second way use of separated control
when an independent fuzzy controller controls each link. The control
decomposition due to a slight decrease in the quality of management has
greatly simplified the processes of creating and placing knowledge bases.
In this Pt II, to eliminate the mismatch of the work of separate indepen-
dent fuzzy controllers, methods for organizing coordination control based
on quantum computing technologies to create robust intelligent control
systems for robotic manipulators with 3DOF and 7DOF described. Quan-
tum supremacy of developed end-to-end IT design of robust intelligent
control systems demonstrated.

*Corresponding Author:
Sergey V. Ulyanov,
Physical and Mathematical Sciences; JINR, Russia; LIT JINR, INESYS LLC (EFKO GROUP), Russia;
Email: ulyanovsv@mail.ru

32 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

1. Intelligent Control System of Manipulator
with Seven Degrees of Freedom
1.1 Intelligent Control System Based on Quan-
tum Computing Knowledge Base Optimizer
(QCOptKBTM)
(6) Decoding. The calculation of the output value of the
corresponding signal.
(7) Denormalization. Output value calculation.
The connection diagram of the QFI block in the Intel-
ligent Control System (ICS) is shown in Figure 1 where:
QREF is the reference signal, Q ′ is the measured variable,

S
uch unconventional phenomena as entangled states, E is the control error, s(t) is a limitation of the control ac-
teleportation, superdense coding appear in the tion, mc(t),mMS(t) are noises in the control channels and in
quantum logic of computational intelligence and the measuring system, U is the control action, d(t) is a de-
quantum information theory, due to the failure to fulfill lay in the Measuring System (MS), fo(t) is external action
the classical law of distributivity. on the CO, Q is an adjustable value.
Quantum computing technologies for creating self-or-
Intelligent superstructure
ganizing Knowledge Bases (KB) of intelligent regulators KBO based on
soft computing
are considered in [1,2]. In [1] a new type of quantum search
FC1
algorithm described on the generalized space of the KB

ck
FC2 QFI

edba
designed for soft computing, and the quantum fuzzy al- ...

nt fe
FCi
gorithm for designing a generalized robust control signal

llige
applied. The Quantum Fuzzy Inference (QFI) [3] imple-

Inte
Quantum FC
ments the self-organization of the KBs, based on the
physical laws of the theory of quantum computing [4, 5] and Qref Ε PID- U Q
CO
the application of four operators: superposition, quantum Controller
correlation, interference, and measurement. Operators su- Q' s(t) mC (t) d(t) fCO (t)
perposition, quantum correlation and interference are uni-
tary, invertible quantum operators, and the measurement MS

operator is classical non-invertible. Lower executive level

QFI algorithm [3] consists of the following sequence
steps:
(1)  Normalization. Bringing input signals to normal-
ized states.
1
(2)  Coding. All normalized states 0 =   and
0
0
1 =   constructed for the current input variables at the
1
coding step.
Next, the probability amplitudes of quantum states 0
and 1 are calculated from the corresponding histograms.
(3) Choosing a correlation type. Selection of the type of
quantum correlation for constructing control output signals.
We will consider three types of miscellaneous cor-
relations: spatial, spatio-temporal and temporal [6], and in
addition to this, correlations of various numbers of fuzzy
controllers (FC) types. The dependence of the output sig-
nals determined by the correlation of the input coefficients
sets, where each set is an entangled state [7].
(4) Superposition. The construction of a superposition
of entangled states.
(5) Measurement of an intelligent quantum state. The
choice of the state with the maximum probability ampli-
tude [8, 9].
mMS (t)
Figure 1. The structure of the ISC based on soft and quan-
tum computing
The design of QFI is produced using the computational
intelligence toolkit as Knowledge Base Optimizer (KBO)
based on quantum computing [10].
Let us consider the method of embedding QFI block in
the ICS structure based on KBO on soft computing.
1.2 Coordination Management Organization
Seven FC are implemented in the selected configuration
of the ICS structure, each of FCs independently controls
one of the seven links.
As a standard control situation for the i FC, a typical
control situation appears, in the conditions of which a
teaching signal TSi received.
Unexpected control situations divided into external and
internal. External unexpected situations include disturbing
action, such as a forced change in the position of the CO,
a change in the reference signals, etc. Internal control sit-
uations include changes in the parameters of the ICS com-
ponents, for example restrictions, noise, and signal delays.
ICS based on soft computing (Figure 1), may contain
information about seven control situations (standard or
unexpected) for each of the links.

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 33

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Let us consider an example when, in the designed ICS for a specific CO (a 7DOF robot manipulator) and de-
based on KBO on soft computing with separate control, scribed in [12]:
FC1 contains information about the standard situation 1 (1) Position Task Solution in known control situations
for link 1, FC2 contains information about the standard PTSKCS.
situation 1 for link 2, etc. (2) Position Task Solution in the external considered
Example. The QFI block in the ICS model based on control situations PTSACCS1.
KBO on Soft Computing implements the self-organization (3) Position Task Solution in the internal considered
of the FC KB in learning situations. In this case, it is pos- control situations PTSACCS2.
sible to combine information about the standard situation (4) Performance IT.
1 for all links, and to extract additional information (by (5) Relative overshoot value σ .
methods of quantum computing and quantum information (6) Relative error in link positioning å .
theory) from the reaction of the designed KBs to imple- (7) Time of one iteration t.
ment robust control in regular and unexpected control sit- (8) The complexity of implementing control P.
uations (which are not included in existing designed KB). (9) Full Control Behavior FCB.
The KBs, whose responses used to design robust control,
in unexpected situations may not be robust. 1.4 The Quality of an Intelligent Control System
The connection diagram of the QFI block in the intelli- with QFI
gent superstructure of the control system shown in Figure Control system models of 7DOF robot manipulator in
2, on which blocks designed using soft and quantum-com- MatLab/Simulink developed to research the quality of
puting technologies highlighted. control systems.
K1′ = [K P′ 1 K D′ 1 K I′1 ] K1=[KP1 KD1 KI1]
The initial ICS on KBO on soft and quantum comput-
TS1
FC1
KB1
GA1
q1
ing contains seven FCs, each of which can correlate with
K 2′ = K P′ 2 K D′ 2 K I′ 2 ]


QFI
K2=[KP2 KD2 KI2]
FC2 TS2 q2 one or more NR systems.
GA2
... KB2
Consider six combinations:
K 7′ = [K P′ 7 K D′ 7 K I′ 7 ] K7=[KP7 KD7 KI7]
FC7 TS7
GA7
q7 (1) The correlation of 2 adjacent fuzzy KBs (the infor-
KB7

Quantum computing mation FC i and i+1 is used to control the i-th link of the
Soft computing technologies
technologies
manipulator);
Intelligent superstructure on KBO obased on soft computing and a generalizing QFI block on
KBO based on quantum computing (2) the correlation of 3 adjacent fuzzy KBs (the infor-
mation FC i, i+1 and i+2 is used to control the i-th link of
Figure 2. Intelligent add-on structure of the ICS based on
KBO on quantum and soft computing the manipulator, as shown on Figure 3);
(3) the correlation of 4 adjacent fuzzy KBs;
The ICS using the QFI called the ICS based on quan- (4) the correlation of 5 adjacent fuzzy KBs;
tum computing KBO. (5) the correlation of 6 adjacent fuzzy KBs;
(6) the correlation of all 7 fuzzy KBs.
1.3 Test Procedure and Quality Criteria
For selected six combination spatial, spatio-temporal
A series of experiments in standard and unexpected con- and temporal correlations are considered.
trol situations carried out to test the robustness of control
system models. The following configuration is taken as the
1
Quantum
FC1 correlation Link 1
initial position of the manipulator: q1 = 0 deg; q2 = 0 deg; q3 KB1 operator

2
Quantum
= 0 deg; q4 = 0 deg; q5 = 0 deg; q6 = 0 deg; q7 = 0 deg. FC2
KB2
correlation
operator
Link 2

Emergencies divided into external and internal. Exter- FC3

Quantum
correlation 3 Link 3
KB3
nal situations include a forced change of provisions and a operator

4
Quantum
FC4 correlation Link 4
change in initial conditions; unexpected situations include KB4 operator

5
Quantum
limitation of control actions, the introduction of noise into FC5
KB5 correlation
operator
Link 5

the control channels and errors in the MS.

6
Quantum
FC6 correlation Link 6
KB6
A system of quality criteria that takes into account operator

7
Quantum
FC7 correlation Link 7
methods for assessing transients of automatic control the- KB7
operator

ory [11] was introduced to evaluate and compare the test 7DOF Robot
manipulator
results of the considered types of control systems (control Soft computing Quantum computing

systems with constant coefficients and ICS for KBO on Figure 3. The application of the correlation of three
Soft and Quantum computing). The methods were adapted neighboring FC

34 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Comparison of the ICS operation based on Quantum Spatial

KBO for spatial, spatio-temporal and temporal correla- correlation

tions shown in Figures 4-9 for selected combinations of Spatio-temporal

correlation
correlated FC.
Temporal
correlation

Performance

overshoot value
Relative

Behavior
iteration

Full Control
in link positioning

Time of one
Relative error

implementing control
The complexity of
control situations
Internal considered
control situations
External considered
situations
Known control
Spatial correlation

Spatio-temporal
correlation

Temporal Figure 8. The result of the work of the ICS on quantum

computing KBO using the correlation of 6 FC
link positioning
Relative error in

correlation
Time of one iteration

Behavior
Full Control
situations
Known control

control situations

control situations

Performance

implementing control
External considered

Internal considered

The complexity of
value
Relative overshoot

Spatial
correlation

Spatio-temporal
Figure 4. The result of the work of the ICS on quantum correlation
computing KBO using the correlation of 2 FC
Temporal
correlation

Performance

overshoot value
Relative

Behavior
iteration

Full Control
in link positioning

Time of one
Relative error

implementing control
The complexity of
control situations
Internal considered
control situations
External considered
situations
Known control
Spatial correlation

Spatio-temporal
correlation

Temporal
correlation
Figure 9. The result of the work of the ICS on quantum
link positioning
Relative error in

Time of one iteration

Behavior
Full Control
situations
Known control

control situations

control situations

Performance

implementing control
External considered

Internal considered

The complexity of
value
Relative overshoot

computing KBO using the correlation of 7 FC

We conclude from the results:
(1) for IСS based on quantum KBO with the correla-
Figure 5. The result of the work of the ICS on quantum tion of 2 FCs, the spatial correlation is the best (Figure 4);
computing KBO using the correlation of 3 FC (2) for IСS based on quantum KBO with the correla-
tion of 3 FCs, the spatio-temporal correlation is the best
Spatial correlation (Figure 5);
(3) for IСS based on quantum KBO with the correla-
Spatio-temporal
correlation tion of 4 FCs, the spatial correlation is the best (Figure 6);
Temporal (4) for IСS based on quantum KBO with the correla-
correlation
tion of 5 FCs, the spatial correlation is the best (Figure 7);
link positioning
Relative error in

Time of one iteration

Behavior
Full Control
situations
Known control

control situations

control situations

Performance

implementing control
External considered

Internal considered

The complexity of
value
Relative overshoot

(5) for IСS based on quantum KBO with the correla-

tion of 6 FCs, the spatial correlation is the best (Figure 8);
(6) for IСS based on quantum KBO with the correla-
Figure 6. The result of the work of the ICS on quantum tion of 7 FCs, the spatial correlation is the best (Figure 9);
computing KBO using the correlation of 4 FC Next, we compare the work of the IСS based on quan-
tum KBO using spatial correlation with the work of the
Spatial correlation IСS based on soft computing KBO (Table 1).
Application of QFI allows the following results:
Spatio-temporal
correlation (1) to solve the problem of positioning in regular situa-
Temporal tions using either correlation type;
correlation
(2) to improve the results of the positioning task in
link positioning
Relative error in

Time of one iteration

Behavior
Full Control
situations
Known control

control situations

control situations

Performance

implementing control
External considered

Internal considered

The complexity of
value
Relative overshoot

conditions of external unexpected situations; results do

not improve in internal unexpected situations;
(3) to increased performance criterion by 5 times;
Figure 7. The result of the work of the ICS on quantum (4) the overall control result is improved; the best result
computing KBO using the correlation of 5 FC is achieved using the spatial correlation of 7 FCs (Fig-
ure 10 shows a comparison with ICS on soft computing).

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 35

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Table 1. Comparison ICS based on quantum computing and ICS based on soft computing

ICS based on quantum KBO

ICS based
Control situation models on soft Spatial cor- Spatio-tempo- Spatial cor- Spatial cor- Spatial cor- Spatial cor-
KBO relation of 2 ral correlation relation of 4 relation of 5 relation of 6 relation of 7
FC of 3 FC FC FC FC FC
1 Known control situations PTSKCS 0,923 1,000 1,000 1,000 1,000 1,000 1,000
External considered control situations PT-
2 0,744 0,821 0,821 0,795 0,769 0,795 0,821
SACCS1
Internal considered control situations
3 0,923 0,846 0,885 0,885 0,923 0,923 0,923
PTSACCS2
4 Performance IT 0,092 0,477 0,461 0,472 0,476 0,465 0,459
5 Relative overshoot value σ 0,969 0,973 0,968 0,974 0,974 0,974 0,971
6 Relative error in link positioning å 0,911 0,962 0,947 0,962 0,962 0,962 0,960
7 Time of one iteration t 0,973 0,961 0,962 0,961 0,961 0,961 0,960
The complexity of implementing control
8 0,946 0,957 0,953 0,973 0,977 0,982 0,986
P
9 Full Control Behavior FCB 0,721 0,816 0,818 0,816 0,817 0,821 0,826

ICS based on
Quantum computing

ICS based on Soft

computing
overshoot value

Behavior FCB
Relative

Full Control
iteration
Time of one

implementing control
control situations

The complexity of
Internal considered
situations
Known control

control situations
External considered

Relative error
in link positioning
Performance

Figure 10. Comparison of ICS based on soft KBO and

ICS based on quantum KBO using spatial correlation of 7
FC
Figures 11-12 demonstrate the operation of the manip-
ulator when using the ICS based on soft KBO and the ICS (a) (b)
based on quantum KBO in the conditions of a standard
control situation and an external unexpected control situa- Figure 12. The movement of the robot manipulator in an
tion (the initial position has been changed). external unexpected control situation: under control of
ICS based on soft computing (a); ISU on ICS based on
quantum computing (b)
From Figures 11-12 we can conclude that ICS based
on quantum computing KBO under the conditions of the
standard and external unexpected control situations solves
the problem of precise positioning of a 7DOF robot ma-
nipulator, in contrast ICS based on soft KBO.
A comparison of the phase portraits of the ICS based
on the KBO on soft computing and the ICS based on the
KBO on quantum computing is shown in Figures 13 and
14 for the considered standard and external unexpected
control situations.
(a) (b)

Figure 11. The movement of the robot manipulator in a

standard control situation: under control of ICS based on
soft computing (a); ISU on ICS based on quantum com-
puting (b)

36 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

 PTS= 1, if q1ref − q1 + q2 ref − q2 + ... + q7 ref − q7 ≤ 2 deg

 ,
 PTS = 0, else

where q1ref , q2 ref , ... , q7 ref are the desired positions of

Figure 13. Changing the position of the manipulator links
in a standard control situation: a comparison of ICS based
on soft computing KBO and ICS based on quantum com-
puting KBO
the links, q1 , q2 , ... , q7 are the current links positions.
However, under conditions of strong interference, it is
almost impossible to reach accurate positioning of each
link. The criterion Performance IT can be excluded from
consideration (time will always be equal to the number of
allowed iterations).
We formulate the general problem: to position the func-
tional device of the manipulator as accurately as possible.
We introduce a new quality criterion: a metric error in the
positioning of the manipulator end effector EEP (End Ef-
fector Positioning):

(X − X ) + (Yref − Y ) + ( Z ref − Z ) ,
2 2 2
EEP
= ref

where { X ,Y , Z} are the coordinates of the reached

point, {X ref , Yref , Z ref } are the coordinates of the desired
point.

 N EEPi 
∑ ,
i =1 EEP max
Figure 14. Changing the position of the manipulator links EEPPERFORMANCE ≡ 
in an external control situation: a comparison of the ICS N
based on soft computing KBO and the ICS based on quan-
tum computting KBO where N is the number of experiments, EEP max is the
maximum metric error from the estimated series of exper-
The application of ICS based on quantum computing
iments.
in solving of positioning problems in internal unexpected
Updated system of quality criteria for evaluating the
situations did not help to improve the results compared
operation of control systems under conditions of strong
to ICS based on soft computing. Let us consider in more
internal interference:
detail internal unexpected situations and increase internal
(1) Metric positioning error of the manipulator end ef-
noises.
fector EEP.
2. The Quality of Work of an ICS based on (2) Relative overshoot value σ .
QFI under Conditions of Strong Internal In- (3) Relative error in link positioning ε .
(4) Time of one iteration t.
terference
(5) The complexity of implementing control P.
Earlier, the following internal unexpected situations with (6) Full Control Behavior FCB.
the introduction of interference considered:
(1) Introducing noise into the control channels (noise FCB ≡ w1 ⋅ P [ EEP ] + w2 ⋅ P [σ ] + w3 ⋅ P [ε ] + w4 ⋅ P [t ] + w5 ⋅ P ,
amplitude = 1);
(2) Making errors in the MS (1,5 degrees). where w = [ 0, 4 0,1 0,1 0, 2 0, 2] is the weight’s vec-
tor.
Let us consider the behavior of ICS under conditions
Consider an unexpected situation with the introduction
of strong interference. Previously, the solution to the po-
of a strong interference in the MS.
sitioning problem was determined by the most accurate
positioning of manipulator links: 2.1 The Quality of Control Systems in Conditions
of Strong Disturbing Interference in the MS
Previously, an internal unexpected situation was consid-

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 37

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

ered with the introduction of random noise with an ampli- 0..30% 79

tude = 1,5 into the MS. 31..60% 92
61..90% 38
Increase the amplitude of the introduced random noise
91..120% 23
to 5 degrees. The number of experiments where the error
121..150% 11
We carry out the following series of experiments: for decreased by N%
150..200% 12
13 points of the manipulator’s working space, 30 random 200..250% 9
noises generated. The generated noises introduced into the 250..300% 6
300 and more % 22
MS of ICS based on the soft KBO with separate control
The result has not changed. 11
and into the ICS based on the quantum KBO.
In Figure 15 shows a comparison of metric errors in
0,907 0,956
0,941
positioning the functional device of the manipulator based 0,656 0,847
on soft and quantum computing for 380 experiments. 0,395
Metric error according to the results of 380 experi-
0,888
ments, when using QFI, it is less than when using only
0,642 ICS based on
soft computing (Table 2). 0,359
0,974
0,807 quantum KBO
0,811
Let us compare the work of the ICS based on soft KBO ICS based on

end effector EEP

error of the manipulator
Metric positioning

value σ
Relative overshoot
soft KBO

positioning ε
Relative error in link
and the ICS based on quantum KBO in accordance with

iteration t
Time of one

control P
The complexity
of implementing

the introduced system of quality criteria Figure 16.
From Figure 16 can concluded that the ICS based on
quantum KBO by all criteria (except for the time of one
iteration) is better to the ICS based on soft KBO.
Behavior FCB
Full Control
Figure 16. The quality of the ICS based on soft and quan-
tum computing in the conditions of strong interference in
the MS
Example. Let us consider an example of the influence
of noise in a control channel of ICS based on soft and
quantum computing. In Figure 17 shows an example of
noise in a control channel. In Figure 18 shows a compari-
son of the control signals during the operation of the ICS
based on soft (a) and quantum (b) computing. In Figure 19
shows the position of the manipulator when operating the
ICS based on soft (a) and quantum computing (b).

Noise q1 Noise q2
5 5
Figure 15. Metric positioning error of the manipulator end
effector [m]: FC is result of ICS based on soft computing; 0 0

QFC is result of ICS based on quantum computing -5

0 100 200 300
-5
0 100 200 300
Noise q3 Noise q4
5 5
Table 2. Comparison of the results of ICS based on KBO 0 0
on quantum and soft computing -5 -5
0 100 200 300 0 100 200 300
Total number of experiments 390 Noise q5 Noise q6
5 5
The number of experiments in which ICS based on quantum com-
87
puting worsen the result than when using only soft computing: 0 0
0..30% 61 -5 -5
31..60% 17 0 100 200 300 0 100 200 300
The number of experiments where the error Noise q7
61..90% 5
increased by N% 5
91..120% 2
0
121..150% 2
The number of experiments in which ICS based on quantum com- -5
303 0 100 200 300
puting improve the result than using only soft computing:

Figure 17. Noise in the control channel

38 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

U1 U2
The error in positioning an end effector when using
40 8
only soft computing does not exceed 0,3 meters with a
30 6

20 4
length of one link of 1 m.
10 2
We will generate noise at which the metric error of po-
0
0 100 200 300
0
0 100 200 300
sitioning the functional device during the operation of the
ICS based on the KBO on soft computing will be more
U3 U4
than 0, 5 m and, with the same noise, we will evaluate the
40 40
behavior of the ICS based on quantum computing. To do
30 30

20 20
this, increase the amplitude of the noise introduced into
10 10
the control channels to 100.
0
0 100 200 300
0
0 100 200 300
Five random noises generated for 13 workspace points.
Generated noises introduced into the control channels of the
U5 U6
ICS based on KBO on soft computing with separate control
20 40
and the ICS based on KBO on quantum computing.
15 30

10 20
Error = 1,918 m Error = 0,184 m
5 10

0 0
0 100 200 300 0 100 200 300

U7
30

20 Desired
point Desired point
Reached Reached
10 point point

0
0 100 200 300

(a)
U1 U2
50 8

40
6
30
4
20
2
10

0 0
0 100 200 300 0 100 200 300

U3 U4
40 30

30
20

20

10
10

0 0
0 100 200 300 0 100 200 300

U5 U6
20 40

15 30

10 20

5 10

0 0
0 100 200 300 0 100 200 300

U7
50

40

30

20

10

0
0 100 200 300
(a) (b)
(b)
Figure 18. Control signals generated by ICS based on soft Figure 19. The position of the manipulator in space when
(a) and quantum (b) computing operating ICS based on soft (a) and quantum computing (b)

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 39

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Figure 20 shows a comparison of metric errors in posi-

Noise q1 Noise q2
tioning the manipulator functional device of the ICS based 50 50
on KBO on soft and quantum computing for 65 experi- 0 0
ments. -50 -50
0 100 200 300 0 100 200 300
Noise q3 Noise q4
50 50

0 0
-50 -50
0 100 200 300 0 100 200 300
Noise q5 Noise q6
50 50
0 0

-50 -50
0 100 200 300 0 100 200 300
Noise q7
Figure 20. Functional device metric positioning error [m]: 50
FC is the result of ICS operation based on soft comput- 0
ing, QFC is the result of ICS operation based on quantum -50
0 100 200 300
computing
According to the results of 65 experiments, we con- Figure 22. Noise in the control channel
clude that the metric error obtained when using quantum
U1 U2
computing is much smaller than when using only soft 80 60

computing. 60
40
Let us compare the work of the ICS based on KBO on 40
soft and quantum computing in accordance with the intro- 20
20

duced system of quality criteria (Figure 21) 0 0

0 100 200 300 0 100 200 300

0,904 0,926 0,941

0,745 0,859
U3 U4
100 60
0,494
0,767
40
0,683 ICS based on
0,414 50
0,976 quantum KBO
0,817 20
0,775
ICS based on
end effector EEP
error of the manipulator
Metric positioning

value σ
Relative overshoot

soft KBO
positioning ε
Relative error in link

iteration t
Time of one

0 0
control P
The complexity
of implementing

0 100 200 300 0 100 200 300

Behavior FCB
Full Control

U5 U6
60 60
Figure 21. The quality of the ICM based on KBO on soft
and quantum computing in the conditions of strong noise 40 40

in the control channels

20 20
It can be seen from Figure 21 that ICS based on KBO
0 0
on quantum computing by all criteria (with the exception 0 100 200 300 0 100 200 300

one iteration time) is superior to the ICS than based on

KBO on soft computing. U7
Example. Let us consider an example of the influence 80

of noise in control channel. In Figure 22 is an example of 60

noise in a control channel. In Figure 23 shows a compari- 40

son of control signals. In Figure 24 shows the position of 20

the manipulator when operating ICS based on soft (a) and 0

0 100 200 300
quantum computing (b).
(a)

40 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

U1 U2
80 60

60
40
40
20
20

0 0
0 100 200 300 0 100 200 300

U3 U4
80 60

60
40
40
20
20

0 0
0 100 200 300 0 100 200 300

U5 U6 (a) (b)
60 60
Figure 24. The position of the manipulator in space when
40 40 operating ICS based on soft (a) and quantum computing (b)

20 20
In the considered unexpected situations with the intro-
duction of strong noise into the control channels, the ICS
0 0 based on KBO on quantum computing showed an advan-
0 100 200 300 0 100 200 300
tage over ICS based on KBO on soft computing.
80
U7
Consider an unexpected situation with the introduction
of a strong disturbing effect in MS.
60

40 2.2. The Quality of Control Systems in Conditions

20
of Strong Disturbing Influences in the Measure-
ment System
0
0 100 200 300
Previously, an internal unexpected situation was considered
(b)
with the introduction of random noise with an amplitude = 1.5
Figure 23. Control signals generated by ICS based on soft into MS. Increase the amplitude of the introduced random
(a) and quantum (b) computing noise to 5 degrees, 30 random noises generated for 13 points
of the manipulator workspace. The generated noise intro-
duced into MS of the ICS based on KBO on soft computing
Error = 2,992 m Error = 0,197 m
with separate control and ICS on quantum computing.
In Figure 25 shows a comparison of the metric errors
in positioning the manipulator functional device of ICS
based on KBO on soft and quantum computing for 380
Desired Reached
point Desired
experiments.
Reached point
point point

Figure 25. Functional device metric positioning error [m]:

FC is the result of ICS operation based on soft comput-
ing, QFC is the result of ICS operation based on quantum
computing

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 41

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

According to the results of 380 experiments, we con-

clude that the metric error obtained when using quantum Noise q1 Noise q2
5 5
computing is much smaller than when using only soft
0 0
computing (Table 3).
-5 -5
It can see from Figure 26 that ICS based on KBO on 0 100 200 300 0 100 200 300
Noise q3 Noise q4
quantum computing by all criteria (with the exception one 5 5
iteration time) is superior to the ICS based on KBO on 0 0

soft computing. -5 -5
0 100 200 300 0 100 200 300
Example. Let us consider an example of the influence Noise q5 Noise q6
5 5
of noise in MS. In Figure 27 is an example of noise in
0 0
MS. In Figure 28 shows a comparison of control signals.
-5 -5
In Figure 29 shows the position of the manipulator when 0 100 200 300 0 100 200 300
Noise q7
operating ICS based on soft (a) and quantum computing 5
(b). 0
-5
0 100 200 300
Table 3. Comparison of the results of ICS based on KBO
on quantum and soft computing
Figure 27. Noise in the control channel
Total number of experiments 390
The number of experiments in which ICS based on quantum com-
105
puting worsen the result than when using only soft computing:
0..30% 80
31..60% 21
The number of experiments where the error
61..90% 2
increased by N%
91..120% 0
121..150% 1
The number of experiments in which ICS based on quantum com-
285
puting improve the result than using only soft computing:
0..30% 131
31..60% 79
61..90% 24
91..120% 16
The number of experiments where the error
121..150% 9
decreased by N%
150..200% 4
200..250% 4
250..300% 2
300 and more % 4
The result has not changed. 11

Let us compare the work of the ICS based on KBO on

soft and quantum computing in accordance with the intro-
duced system of quality criteria (Figure 26)

0,907 0,941 0,956

0,656 0,847

0,395
0,888
0,642 ICS based on
0,359
0,974 quantum KBO
0,807
0,811
ICS based on
end effector EEP
error of the manipulator
Metric positioning

value σ
Relative overshoot

soft KBO
positioning ε
Relative error in link

iteration t
Time of one

control P
The complexity
of implementing

Behavior FCB
Full Control

(a)

Figure 26. The quality of the ICM based on KBO on soft

and quantum computing in the conditions of strong noise
in MS

42 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

(a) (b)

Figure 29. The position of the manipulator in space when

(b)
Figure 28. Measured signals during operation ICS based
on soft (a) and quantum (b) computing
operating ICS based on soft (a) and quantum computing
(b)
The use of ICS based on KBO on quantum computing
allows us slightly improve the performance compared to
ICS based on KBO on soft computing, but the improve-
ment is insignificant.
In general, the positioning error of a functional device
when using only soft computing does not exceed 0,3 m
with a single link length of 1 m.
In Figure 30a comparison of metric errors of position-
ing the functional device manipulator of the ICS based on
soft and quantum KBO for 65 experiments is presented.

Reached
point
Reached
Error = point
Error = Desired
point 0,262 m Desired
2,519 m
point

Figure 30. Functional device metric positioning error [m]:

FC is the result of ICS operation based on soft comput-
ing, QFC is the result of ICS operation based on quantum
computing
We conclude from the results of 65 experiments: the
metric error obtained when using quantum computing is
much smaller than when using only soft computing.
Let us compare the work of the ICS based on soft and
quantum KBO in accordance with the introduced system

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 43

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

of quality criteria (Figure 31).

0,882 0,933 0,968

0,809
0,504

0,258
0,517
0,330 ICS based on
0,062
0,977 quantum KBO
0,780
0,597
ICS based on
end effector EEP
error of the manipulator
Metric positioning

value σ

soft KBO
Relative overshoot

positioning ε
Relative error in link

iteration t
Time of one

control P
The complexity
of implementing

Behavior FCB
Full Control

Figure 31. The quality of the ICS based on soft and quan-

tum computing in the conditions of strong interference in
the MS
From Figure 31 it can be seen that the ICS based on
quantum KBO by all criteria (except for the time of one
iteration) is better to the ICS based on soft computing.
Example. Let us consider an example of the interfer-
ence in a MS on the operation of an ICS in soft and quan-
tum computing. In Figure 32 shows an example of noise
in a MS.

Noise q1 Noise q2 (a)

5 5
0 0

-5 -5
0 100 200 300 0 100 200 300
Noise q3 Noise q4
5 5

0 0
-5 -5
0 100 200 300 0 100 200 300
Noise q5 Noise q6
5 5
0 0
-5 -5
0 100 200 300 0 100 200 300
Noise q7
5
0
-5
0 100 200 300

Figure 32. Noise in the MS

In Figure 33 shows a comparison of the measured sig-
nals during the operation of the ICS based on soft (a) and
quantum (b) computing. In Figure 34 shows the position
of the manipulator when operating the ICS based on soft
(a) and quantum computing (b).

(b)
Figure 33. Measured signals during ICS based on soft (a)
and quantum (b) computing

44 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

external unexpected control situations;

(3) to improve the accuracy of the positioning of the
functional device in conditions of strong internal interfer-
ence;
(4) to improve the Performance criterion;
(5) to reduce relative link positioning error.
Next, we consider the application of QFI in the control
system of a simpler CO as 3DOF robot manipulator.

3. Control Systems for a Manipulator with

3DOF based on Quantum Computing
In this section, 3DOF manipulator control systems con-
sidered both at the simulation level and at the physical test
benchmark.

3.1. Management Task

Three FCs implemented in the selected configuration of
the ICS structure; each FC independently controls one of
the 3links.
In Figure 35 shows a structural diagram of ICS by
3DOF manipulator based on KBO on soft computing with
separated control.

K3 =[KP3 KD3 K I 3 ] TS3

FC3 q3
KB3
GA3
K2 =[KP2 KD2 K I 2 ] TS2
FC2 q2
KB2
GA2
K1 =[KP1 KD1 K I 1 ] TS1
FC1 q1
KB1
GA1

Intelligent superstructure based on KBO on soft computing

Q ref E PID- U = [u1 u2 u3] Q = [q1 q2 q3]

regulator CO
Q'

(a) (b) MS

Figure 34. The position of the manipulator in space when
operating ICS based on soft (a) and quantum computing
(b)
In the considered unexpected situations with the intro-
duction of strong interference into the MS, the ICS based
on quantum KBO improves quality than when using only
soft computing.
Thus, we considered the method of organizing coordi-
nation control using quantum computing technologies on
the example of a complex CO a 7DOF robot manipulator.
We have demonstrated at the simulation model (using the
MatLab/Simulink package) that the inclusion of generalize
QFI unit into the ICS based on soft computing allowed:
(1) to provide a complete solution to the problem of
positioning in standard situations;
(2) to improve positioning accuracy in conditions of
Figure 35. ICS with separated control based on KBO on
soft computing for 3DOF manipulator
In Figure 35 K is the matrix of proportional, differ-
ential and integral coefficients of the PID controller
K Pi , K Di ,K Ii , i = 1,3 , i is the number of the corresponding
link of the robot of the manipulator, TSi is the training sig-
nal , GAi is a GA that generates a teaching signal for the
formation of the i-th KB.
As a standard control situation for the FCi, a typical
control situation appears in the conditions of which a
teaching signal TSi received. Unexpected control situa-
tions divided into external and internal. Designed ICS
based on KBO on soft computing with separated divided
control (Figure 35) contain information about three situa-
tions of control (standard or unexpected) for each of three

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 45

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

links. based on quantum computing using temporal (2), spatial

In the designed ICS based on KBO on soft computing (3) and spatio-temporal (4) correlations when testing Mat-
with separated control FC1 contains information about lab/Simulink models
the standard situation 1 (KB1) for link 1, FC2 contains From the comparison results, we conclude that the
information about the standard situation 1 for link 2 (KB2), addition of the QFI unit to ICS with separated control the
and FC3 contains information about the standard situation control task is solved for both standard and unexpected
1 for link 3 (KB3). control situations; the performance increases; the accura-
The scheme for extracting hidden information about cy of link positioning improves; and the implementation
the relationships between existing FCs designed using soft complexity of control decreases. The complexity of the
computing technologies for three manipulator links with implementation control depends on the dynamics of the
KBs obtained for standard control situations using QFI control signal; we consider the robustness of the generated
unit shown in Figure 36. control laws below.
KB1 for link1 Because of QFI block apply in ICS, all quality criteria
Standard control situation 1
Coordination control are improved as a result of eliminating the mismatch of
of links 1-3 for: the work of separated independent KBs by organizing
KB2 for link2 Standard situations coordination control. Moreover, if for MatLab/Simulink
QFI
Standard control situation 2 1,2, ..., N
Unexpected
models the best indicator of full control behavior provided
situations 1,2, ..., N when using spatio-temporal correlation, and then manip-
KB3 for link3
Standard control situation 3 ulator Testbench determines the optimal choice of spatial
Figure 36. Methodology for extracting hidden informa- correlation.
tion of the relationships of the KB Next, we will consider the ICS based on KBO on quan-
tum computing only with the use of spatial correlation.
We will add the QFI block to ICS based on KBO on In Figure 38 compares the operation of ICS based on
soft computing, which can realize the self-organization of KBO on soft computing with separated control in the
the KBs. conditions of an external unexpected control situation in
comparison with the ICS based on quantum computing.
3.2. Modeling and Experiment: Control Quality
The forced displacement of the second link is an unex-
The performance of the considered ICSs evaluated based pected situation in this case.
on the results of MatLab / Simulink modeling and the re-
sults of a series of experiments on CO Testbench.
Let us compare the work of ICS based on KBO on
quantum computing using spatial, spatio-temporal and
temporal correlations and ICS based on soft computing
with separated control. Spatial (3)

In Figures 37 and 38, a comparison of ICS is given for Temporal (2)

Spatio-temporal (4)
MatLab/Simulink models and manipulator Testbench ac-
ICS with separated control (1)
cording to the introduced system of quality criteria.
Position Task
Solution in standard
control situations
Position Task
Solution in
unexpected control
situations
Performance

Relative
overshoot value

Relative error in
link positioning

One iteration
time

Implementation
complexity

Full Control
Behavior

Figure 38. Comparison of the results of ICS based on

KBO on soft computing with separate control (1) and ICS
Spatio-temporal (4) based on quantum computing using temporal (2), spatial
Spatial (3)
(3) and spatio-temporal (4) correlations when testing ma-
Temporal (2)
nipulator Testbench
ICS with separated control (1)
Position Task
Solution in standard
control situations
Position Task
Solution in
unexpected control
situations
Performance

From Figure 39 it can seen that in the considered unex-

Relative
overshoot value

Relative error in
link positioning

One iteration
time

Implementation
complexity

pected control situation, ICS based on KBO on quantum

Full Control
Behavior

computing decides with the positioning problem with a

given accuracy, in contrast to ICS based on KBO on soft
Figure 37. Comparison of the results of ICS based on computing with separated control.
KBO on soft computing with separate control (1) and ICS

46 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Consider internal unexpected control situations. Let us

compare ICS based on KBO on soft computing with one
FC and ICS based on KBO on quantum computing in the
conditions of changing restrictions of the control channel.
In Figures 41 and 42 ICS is compared respectively for
MatLab / Simulink models and for manipulator Testbench,
taking into account internal and external unexpected con-
trol situations.

ICS based on KBO on

quantum computing
(a)

Position Task
Solution in standard
control situations
ICS based on KBO on soft

Position Task
Solution in
unexpected control
situations
computing with one FC

Performance

Relative
overshoot value

Relative error in
link positioning

One iteration
time

Implementation
complexity

Full Control
Behavior
Figure 41. Comparison of the results of ICS based on
KBO on soft computing with one FC and ICS based on
quantum computing when testing Matlab / Simulink mod-
els

ICS based on KBO on

(b) quantum computing
Position Task
Solution in standard
control situations

ICS based on KBO on soft

Position Task
Solution in
unexpected control
situations

Performance

computing with one FC

Relative
overshoot value

Relative error in
link positioning

Figure 39. ICS based on KBO on soft computing with

One iteration
time

Implementation
complexity

Full Control
Behavior
separated control in unexpected control situation (a); ICS
based on KBO on quantum computing (b)
The inability of ICS based on soft computing to solve Figure 42. Comparison of the results of ICS based on
the problem of exact position control in Figure 40 also KBO on soft computing with one FC and ICS based on
illustrated. FC responsible for managing the second link quantum computing when testing manipulator Testbench
for the allotted time was not able to rehabilitate after a
It can see from the comparison results that ICS based
powerful external impact. Because of which the position-
on KBO on soft computing and ICS based on quantum
ing error of the second link was more than 50 degrees, the
computing solves the positioning problems for standard
control goal has not achieved and the control system as a
and external unexpected situations. ICS based on soft
whole was not robust.
computing with one FC not always coping with internal
unexpected control situations. Full Control Behavior for
External impact on the second link
ICS on quantum computing is higher, both for MatLab/
Current position
Simulink models and for the manipulator Testbench.
Reference signal
Let us demonstrate the operation of ICS based on KBO
on soft computing with one FC in the conditions of an
internal unexpected control situation (Figure 43) in com-
parison with ICS based on KBO on quantum computing.
Figure 40. The change in the position of the second link
under the control of ICS based on KBO on soft computing
with separated control

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 47

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Desired position
Desired position
Reached Reached
position
position
Initial
position
Initial
position
(a)
Figure 43. ICS based on KBO on soft computing with
one FC in the conditions of an internal unexpected control
situation (a); ICS based on quantum computing (b)
From Figure 43 it can seen that ICS based on KBO on
quantum computing provides a better solution quality than
ICS based on soft computing with one FC.
Consider the control signals generated by the consid-
ered types of ICS based on soft and quantum computing.
In Figure 44 QFC are the control signals of ICS based
on KBO on quantum computing, FC are the control sig-
nals generated by ICS on soft computing with one FC, FC
Decomposition are the control signals generated by ICS
on soft computing with separated control.
First link
Second link
Reaction to external
influences
Third link
(a)
Figure 44. Control signals generated by different types of
ICS in unexpected control situations
From Figure 44 (a) it can seen that the control signals
generated by ICS based on KBO on quantum computing
under external disturbing influences are similar to those
generated by ICS on soft computing with one FC. How-
ever, the amplitude of the control signal at the time of
external influence is significantly lower for ICS based on
quantum computing. The control signal formed by ICS
based on soft computing with separated control have be-
come overshoot.
From Figure 44 (b) it can seen that the control signal gen-
erated by ICS based on KBO on soft computing with one FC
for the second link has a local instability section, while ICS
based on quantum computing generates robust control sig-
nals under conditions of external disturbing influences.
48 Distributed under creative commons license 4.0
Thus, the minimum consumption of useful resource in
the formation of control signals ensured when using ICS
based on KBO on quantum computing.
4. Conclusion
Using two COs of varying complexity (3DOF and 7DOF
manipulators) as an example, the advantages and limita-
tions of using soft and quantum computing technologies
in ICS were demonstrated. On the example of 3DOF
(b) manipulator, the minimal difference between the results
of the physical Manipulator Testbench and the MatLab/
Simulink model demonstrated.
Figure 45 shows a generalized comparison of ICS
based on KBO on soft and quantum computing for 3DOF
and 7DOF manipulators for standard and unexpected con-
trol situations of the examples considered in this article
(Figures 11, 13, 19, 24, 39, 40).
First link
Second link
Local instability
Third link
Figure 45. Comparison of ICS based on soft and quantum
computing for 3DOF and 7DOF
Further research focused on the development and anal-
(b) ysis of the physical Testbench of Manipulator with 7 DOF,
as well as its integration with the mobile platform.
Appendix 1. Requirements to QFI-model De-
sign and Its Features in Quantum Algorithm
Control of Knowledge Base Self-organization
A1.1 Main Proposals and Features of QFI Model
Main proposals and features of the developed swarm
QFI-model in the solution of intelligent control problems
are as following:
A. Main proposals
(1) The digital value’s set of control signals produced
by responses of FC outputs are considered as swarm parti-
cles along of classical control trajectories with individual
marked intelligent agents;
(2) Communication between particle swarm trajectories
through a quantum link is introduced;
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
(3) Intelligent agents are used different types of quan-
tum correlations (as behavior control coordinator with
quantum computation by communication) and information
transport (value information);
(4) The (hidden) quantum value information extracted
from classical states of control signal classical trajectories
(with minimum entropy in “intelligent states” of designed
robust control signals).
B. Features
(1) Developed QFI model is based on thermodynamic
and information-theoretic measures of intelligent agent in-
teractions in communication space between macro- and mi-
cro-levels (the entanglement-assisted correlations in an active
system represented by a collection of intelligent agents);
(2) From computer science viewpoint, QA of QFI
model plays the role of the information-algorithmic and
SW-platform support for design of self-organization pro-
cess;
(3) Physically, QFI supports optimally a new developed
thermodynamic trade-off of control performance (between
stability, controllability and robustness) in self-organiza-
tion KB process.
From quantum information theory viewpoint, QFI
reduces the redundant information in classical control
signals using four facts[4] from quantum information for
data compression in quantum information processing:1)
efficient quantum data compression; 2) coupling (separa-
tion) of information in the quantum state in the form of
classical and quantum components; 3) amount of total,
classical, and quantum correlation; and 4) hidden (observ-
able) classical correlation in the quantum state.
The gate structure of QFI model with self-organization
KB properties that includes all of these QA features (see,
below Figure A1) based on above-mentioned proposals.
Let us discuss the following question.
Q. What is a difference between our approach and Nat-
ural (or man-made) models of self-organization?
A. Main differences and features are as followings:
(1) In our approach a self-organization process is de-
scribed as a logical algorithmic process of value informa-
tion extraction from hidden layers (possibilities) in classi-
cal control laws using quantum decision-making logic of
QFI-models based on main facts of quantum information,
quantum computing and QA’s theories;
(2) Structure of QFI includes all of natural elements of
self-organization (templating, self-assembly, and self-or-
ganization structure) with corresponding quantum opera-
tors (superposition of initial states, selection of quantum
correlation types and classes, quantum oracles, interfer-
ence, and measurements);
(3) QFI is a new quantum search algorithm (belonging
Distributed under creative commons license 4.0
to so called QPB-class) that can solve classical algorith-
mically unsolved problems;
(4) In QFI the self-organization principle is realized
using the on-line responses in a dynamic behavior of clas-
sical FC’s on new control errors in unpredicted control
situations for the design of robust intelligent control;
(5) Model of QFI supports the thermodynamic inter-
relations between stability, controllability and robustness
for design of self-organization processes.
Specific features of QFI applications in design of ro-
bust KB. Let us stress the fundamentally important spe-
cific feature of operation of the QA (in the QFI model) in
the design process of robust laws for the coefficient gain
schedules of fuzzy PID controllers based on the individual
KB that designed on Soft Computing Optimizer (SCO)
with soft computing.
A1.2 Quantum Information Resources in QFI Al-
gorithm
In this section we introduce briefly the particularities of
quantum computing and quantum information theory that
are used in the quantum block, QFI supporting a self-or-
ganizing capability of FC in robust ICS. According to
described below algorithm the input to the QFI gate is
considered as a superposed quantum state K1 (t ) ⊗ K 2 (t ) ,
where K1,2 (t ) are the outputs from fuzzy controllers FC1
and FC2 designed by SCO (see, below Figure A1) for the
given control task in different control situations (for ex-
ample, in the presence of different stochastic noises).
A1.2.1 Quantum Hidden Information Extraction
in QFI
Using the four facts from quantum information theory
QFI extracts the hidden quantum value information from
classical KB1 and KB2 (see Figure A1).
(a)
DOI: https://doi.org/10.30564/aia.v2i2.1395 49
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
(b)
Figure A1. Example of information extraction in QFI
In this case between KB1 and KB2 (from quantum in-
formation theory of viewpoint) we organize a communica-
tion channel using quantum correlations that is impossible
in classical communication theory. The algorithm of su-
perposition calculation is presented below and described
in details in [4].
We discuss for simplicity the situation in which an ar-
bitrary amount of correlation is unlocked with a one-way
message.
Example. Let us consider the communication process
between two KBs as communication between two players
n computing optimizer).
A and B (see Figure A1) and let d = 2 . According to
the law of quantum mechanics, initially we must prepare
a quantum state description by density matrix ρ from
two classical states (KB1 and KB2). The initial state ρ is
shared between subsystems held by A (KB1) and B (KB2),
with respective dimensions d,
1 d −1 1
=ρ
2d =∑∑
k 0=t 0
(k k ⊗t t ) ⊗ (U
A t k k U t†
Here U0=I and U1 changes the computational basis to a
conjugate basis i U1=
k 1 d ∀i, k .
In this case, B chooses k randomly from d states in
two possible random bases, while A has complete knowl-
edge on his state. The state (A1) can arise from follow-
ing scenario. A picks a random n-bit string k and sends
B k or H ⊗n k depending on whether the random bit
t = 0 or 1 . Player A can send t to player B to unlock the
correlation later. Experimentally, Hadamard transform, H
and measurement on single qubits are sufficient to prepare
50 Distributed under creative commons license 4.0
the state (A1), and later extract the unlocked correlation in
ρ ′ . The initial correlation is small, i.e. I Cl(l ) ( ρ ) = 1 log d .
2
The final amount of information after the complete
measurement MA in one-way communication is ad hoc,
I Cl (=
ρ ′ ) I Cl(l ) =
( ρ ) log d + 1 , i.e., the amount of accessi-
ble information increase. This phenomenon is impossible
classically. However, states exhibiting this behaviour need
not be entangled and corresponding communication can
be organized using Hadamard transform.
Therefore, using the Hadamard transformation and a
new type of quantum correlation as the communication
between a few KB’s it is possible to increase initial in-
formation by unconventional quantum correlation (as the
quantum cognitive process of a value hidden information
extraction in on-line, see, e.g. Figure A1, b). In present
section we consider a simplified case of QFI when with
the Hadamard transform is organized an unlocked correla-
tion in superposition of two KB’s; instead of the difficult
defined entanglement operation an equivalent quantum
oracle is modelled that can estimates an “intelligent state”
with the maximum of amplitude probability in corre-
sponding superposition of classical states (minimum en-
tropy) principle relative to extracted quantum knowledge
[5]
. Interference operator extracts this maximum of ampli-
tude probability with a classical measurement.
Figure A2 shows the algorithm for coding, searching
and extracting the value information from KB’s of fuzzy
PID controllers designed by SCO and QCO (quantum
) B
.(A1)
Figure A2. The structure of QFI gate
Optimal drawing process of value information from
a few KBs that are designed by soft computing is based
on following four facts from quantum information theo-
ry[4]: (i) the effective quantum data compression; (ii) the
splitting of classical and quantum parts of information in
quantum state; (iii) the total correlations in quantum state
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

are “mixture” of classical and quantum correlations; and

(iv) the exiting of hidden (locking) classical correlation in
quantum state.
This quantum control algorithm uses these four Facts
from quantum information theory in following way: (1)
compression of classical information by coding in com-
putational basis {0 ,1} and forming the quantum corre-
lation between different computational bases (Fact 1); (2)
separating and splitting total information and correlations
on “classical” and “quantum“ parts using Hadamard trans-
form (Facts 2 and 3); (3) extract unlocking information
and residual redundant information by measuring the clas-
sical correlation in quantum state (Fact 4) using criteria of
maximal corresponding amplitude probability. These facts
are the informational resources of QFI background. Using
these facts, it is possible to extract an additional amount
of quantum value information from smart KBs produced
by SCO for design a wise control using compression and
rejection procedures of the redundant information in a
classical control signal.
Below we discuss the application of this quantum con-
trol algorithm in QFI structure.
A1.3 Structures of Robust ICS and Information
Design Technology of Quantum KB Self-organi-
zation
The kernel of the above-mentioned FC design toolkit is
a so-called SCO implementing advanced soft computing
ideas. SCO is considered as a new flexible tool for design
of optimal structure and robust KBs of FC based on a
chain of genetic algorithms (GAs) with information-ther-
modynamic criteria for KB optimization and advanced
error BP-algorithm for KB refinement. Input to SCO can
be some measured or simulated data (called as TS) about
the modelling system. For TS design (or for GA fitness
evaluation) we use stochastic simulation system based on
the control object model.
Figure A3 illustrates as an example the structure and
main ideas of self-organized control system consisting
of two FC’s coupling in one QFI chain that supplies a
self-organizing capability. CO may be represented in
physical form or in the form of mathematical model. We
will use a mathematical model of CO described in Matlab
/ Simulink. The kernel of the above-mentioned FC design
tools is a so-called SCO implementing advanced soft
computing ideas.
Figure A3. Structure of robust ICS based on QFI
Robustness of control signal is the background for sup-
port the reliability of control accuracy in uncertainty envi-
ronments. The effectiveness of the developed QFI model
is illustrated for important case, the application to design
of robust control system in unpredicted control situations.
The main technical purpose of QFI is to supply a self-or-
ganization capability for many (sometimes unpredicted)
control situations based on a few KBs. QFI produces a ro-
bust optimal control signal for the current control situation
using a reducing procedure and compression of redundant
information in KB’s of individual FCs. Process of rejection
and compression of redundant information in KB’s uses the
laws of quantum information theory. Decreasing of redun-
dant information in KB-FC increases the robustness of con-
trol without loss of important control quality as reliability
of control accuracy. As a result, a few KB-FC with QFI can
be adapted to unexpected change of external environments
and to uncertainty in initial information.
Let us discuss in detail the design process of robust KB
in unpredicted situations.
A1.3.1 KB self-organization Quantum Algorithm
of FC’s Based on QFI
We use real value of a current input control signal to de-
sign normalized state 0 . To define probability ampli-
tude a0 we will use simulation results of controlled object
behavior in teaching conditions. In this case by using con-
trol signal values, we can construct histograms of control
signals and then taking integral we can receive probability
distribution function and calculate α 0 = P0 . Then we
can find α=
1 1 − P0 . After that it is possible to define
state 1 as shown on Figure A4 below [13].

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 51

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

k PKB1 ,KB2 (ti )k PKB1 ,KB2 (ti − ∆t ) → k Pnew (ti ) ⋅ gainP

2. QFI based
on temporal k DKB1 ,KB2 (ti )k DKB1 ,KB2 (ti − ∆t ) → k Dnew (ti ) ⋅ gainD
correlations
k IKB1 ,KB2 (ti )k IKB1 ,KB2 (ti − ∆t ) → k Inew (ti ) ⋅ gainI
KB1 KB1 KB 2 KB 2 new
3. QFI based k P (ti ) k D (ti − ∆t ) k P (ti − ∆t ) k D (ti ) → k P (ti ) ⋅ gainP
on spatio-tem- KB1
k (ti )k IKB1 (ti − ∆t )k DKB 2 (ti − ∆t )k IKB 2 (ti ) → k Dnew (ti ) ⋅ gainD
poral correla- D
tions k IKB1 (ti )k PKB1 (ti − ∆t )k IKB 2 (ti − ∆t )k PKB 2 (ti ) → k Inew (ti ) ⋅ gainI

Step 3: Superposition and entanglement. According to

the chosen quantum correlation type construct superpo-
sition of entangled states as shown on general Figure A5,
Figure A4. Example of control signal and corresponding where H is the Hadamard transform operator.
probability distribution function Step 4: Denormalization
For QA design of QFI it is needed to apply the addi- Calculate final (denormalized) output result as follows:
tional operations to partial KBs outputs that drawing and
k Pnew ( t ) ⋅ gainP , k Doutput =
k Poutput = k Dnew ( t ) ⋅ gainD , k Ioutput =
k Inew ( t ) ⋅ gainI .
aggregate the value information from different KBs. Soft
computing tool does not contain corresponding necessary
Step 4a: Find robust QFI scaling gains { gainP , gainD , gainI }
operations. The necessary unitary reversible operations are based on GA and a chosen fitness function.
called superposition, entanglement (quantum correlation)
and interference that physically are operators of quantum
computing.
Consider main steps of developed QFI process that is
considered as a QA.
Step 1: Coding
(1) Preparation of all normalized states 0 and 1 for

current values of control signal {k P ( t ) , k D ( t ) , k I ( t )} (in-

i i i

dex i means a number of KB) with respect to the chosen

knowledge bases and corresponding probability distribu-
tions, including:
(2) calculation of probability amplitudes a0, a1 of states
0 and 1 from histograms;
(3) by using a1 calculation of normalized value of state
1 .
Step 2: Choose quantum correlation type for prepa-
ration of entangled state. In the Table A1 investigat-
ed types of quantum correlations are shown. Take,
for example, the following quantum correlation type:
Figure A5. The algorithm of superposition calculation
{k ( t ) , k ( t ) , k ( t ) , k ( t )} → k ( t ) , where 1 and 2 are
1
P
1
D
2
P
2
D
new
P
In proposed QFI we investigated the proposed types of
indexes of KB. quantum QFI correlations shown in Table A1 where the
Then a quantum state a1a2 a3 a4 = k P ( t ) k D ( t ) k P ( t ) k D ( t )
1 1 2 2
correlations are given with 2KB, but in general case a few
is considered as correlated (entangled) state. of KBs may be; ti is a current temporal point and Δt is a
Table A1. Types of quantum correlations correlation parameter.
Let us discuss the particularities of quantum computing
1. QFI based
k PKB1 ,KB2 (ti )k DKB1 ,KB2 (ti ) → k Pnew (ti ) ⋅ gainP that are used in the quantum block QFI (Figure A2) sup-
on spatial k DKB1 ,KB2 (ti )k IKB1 ,KB2 → k Dnew (ti ) ⋅ gainD porting a self-organizing capability of a fuzzy controller.
correlations
k IKB1 ,KB2 (ti )k PKB1 ,KB2 (ti ) → k Inew (ti ) ⋅ gainI Optimal drawing process of value information from a few
of KBs as above-mentioned is based on the following four
facts from quantum information theory:
(1) the effective quantum data compression (Fact1);

52 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

(2) the splitting of classical and quantum parts of infor-

mation in quantum state (Fact 2);
(3) the total correlations in quantum state are “mixture”
of classical and quantum correlations (Fact 3);
(4) existing of hidden (locking) classical correlation
in quantum state using criteria of maximal corresponding
probability amplitude (Fact 4).
These facts are the informational resources of QFI
background. Using these facts, it is possible to extract
the value information from KB1 and KB2. In this case
between KB1 and KB2 (from quantum information theo-
ry point of view) we organize a communication channel
using quantum correlations that is impossible in classical Figure A6. First internal layer of QC Optimizer
communication. In QFI algorithm with the Hadamard
transform an unlocked correlation in superposition of
states is organized. The entanglement operation is mod-
elled as a quantum oracle that can estimate a maximum
of amplitude probability in corresponding superposition
of entangled states. Interference operator extracts this
maximum of amplitudes probability with a classical meas-
urement. Thus, from two FC-KBs (produced by SCO for
design a smart control) we can produce a wise control
by using compression and rejection procedures of the
redundant information in a classical control signal. This
completes the particularities of quantum computing and
quantum information theory that are used in the quantum
block supporting a self-organizing capability of FC.

A1.4 Robust FC Design Toolkit: SC and QC Opti-

mizers for Quantum Controller’s Design Figure A7. Second internal layer of QC Optimizer

To realize QFI process we developed new tools called “QC
Optimizer” that are the next generation of SCO tools.
A1.4.1 QC Optimizer Toolkit
QC Optimizer Toolkit is based on Quantum & Soft Com-
puting and includes the following:
(1) Soft computing and stochastic fuzzy simulation
with information-thermodynamic criteria for robust KBs
design in the case of a few teaching control situations;
(2) QFI-Model and its application to a self-organization
process based on two or more KBs for robust control in
the case of unpredicted control situations.
Internal structure of QC Optimizer is shown on Fig-
ures A6 and A7.
Remark. On Figure A6, the first internal layer of QC
Optimizer is shown (inputs/output). On Fig/ A7, the quan-
tum block realizing QFI process based on three KB is
described. The “delay time = 20 (sec)” on the Figure A7,
corresponds to the parameter “Δt” given in temporal
quantum correlations description (see Table A1); the knob
named “correlation parameters” call other block (see
Figure A7) where a chosen type of quantum correlations
(Table A1) is described.
On Figure A8 description of temporal quantum correla-
tions is shown. Here “kp1_r” means state 0 for kp(t) of
FC1 (or KB1); “kp1_r_t” means state 0 for kp(t+Δt) of
FC1 (or KB1); “kp1_v” means state 1 for kp(t) of FC1 (or
KB1); “kp1_v_t” means state 1 for kp(t+Δt) of FC1 (or
KB1); and so on for other FC2 (KB2) and FC3(KB3).

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395 53

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Using unconventional computational intelligence

toolkit, we propose a solution of such kind of generaliza-
tion problems by introducing a self-organization design
process of robust KB-FC that supported by QFI based on
Quantum Soft Computing ideas.
The main technical purpose of QFI is to supply a
self-organization capability for many (sometimes unpre-
dicted) control situations based on a few KBs. QFI pro-
duces robust optimal control signal for the current control
situation using a reducing procedure and compression of
Figure A8. Internal structure of “correlation parameters” redundant information in KB’s of individual FCs. Process
block of rejection and compression of redundant information
in KB’s uses the laws of quantum information theory.
A1.4.2 Design of Intelligent Robust Control Sys-
Decreasing of redundant information in KB-FC increases
tems for Complex Dynamic Systems Capable to
the robustness of control without loss of important control
Work in Unpredicted Control Situations
quality as reliability of control accuracy. As a result, a few
Describe now key points of Quantum & Soft Computing KB-FC with QFI can be adapted to unexpected change of
Application in Control Engineering as follows: external environments and to uncertainty in initial infor-
(1) PID Gain coefficient schedule (control law) is de- mation.
scribed in the form of KB of a Fuzzy Inference System At the second stage of design with application of the
(realized in FC); QFI model, we do not need yet to form new production
(2) GA with complicated Fitness Function is used for rules. It is sufficient only to receive in on-line the response
KB-FC forming; of production rules in the employed FC to the current
(3) KB-FC tuning is based on Fuzzy Neural Networks unpredicted control situation in the form of the output
using error BP-algorithm; control signals of the coefficient gains schedule in the
(4) Optimization of KB-FC is based on SC optimizer fuzzy PID controller. In this case, to provide the operation
tools (Step 1 technology); of the QFI model, the knowledge of particular production
(5) Quantum control algorithm of self-organization is rules fired in the KB is not required, which gives a big ad-
developed based on the QFI-model; vantage, which is expressed the form of an opportunity of
(6) QFI-model realized for the KB self-organization to designing control processes with the required robustness
a new unpredicted control situation is based on QC opti- level in on-line.
mizer tools (Step 2 technology). Note that the achievement of the required robustness
In this Appendix 1 we are introduced briefly the partic- level in an unpredicted control situation essentially de-
ularities of quantum computing and quantum information pends in a number of cases on the quality and quantity of
theory that are used in the quantum block QFI (Figure A9) the employed individual KB’s.
supporting a self-organizing capability of FC in robust ICS. Thus, the QA in the QFI model is a physical prototype
of production rules, implements a virtual robust KB for a
QFI input/output
Delay time (if temporal QFI):
t_corr/sampletime
fuzzy PID controller in a program way (for the current un-
QFI kernel menu
predicted control situation), and is a problem-independent
Templating
Multiple KB set
toolkit. The presented facts give an opportunity to use
K pFC 0
K pnew experimental data of the teaching signal without designing
K dFC 0
a mathematical model of the CO. This approach offers the
Multiple KB SC Optimizer

K dnew Probability
KiFC 0
QFI (Self-Organization)

function
Kinew challenge of QFI using in problems of CO with weakly
Self-
Assembly
formalized (ill-defined) structure and a large dimension of
K pFC 2
Robust QFC the phase space of controlled parameters.
K dFC 2
Appendix 2: Quantum Computing Optimizer of
KiFC 2 Knowledge Base (QCOptKBTM)
Scale coefficients:
max K p ( d ,i ) ⋅Q_A_params Quantum Computing Optimizer of KB (QCOptimizer)
Description of correlation type
is a software toolkit designed for modeling, creation and
Figure A9. QFI-process by using QC Optimizer (QFI optimization of complex control structures, including
kernel) fuzzy and quantum control modules. Working with

54 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1395

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
QCOptimizer one can:
(1) create a model of complex control system,
consisting from different blocks connected at any fashion;
(2) analyze performance of modeled control system
and optimize it using different teaching signals.
A2.1 Learning Signal for QCOptimizer
In order to perform different optimization algorithms
QCOptimizer requires learning signal, which presents
samples of input values and corresponding output values.
QCOptimizer is able to read signal data from Matlab v.4
and v.5 files and from text files.
Text files are processed based on locale data, which
defines symbols for decimal point, thousands of separators
and so on. By default, QCOptimizer uses windows set-
tings for these parameters. If those settings do not match
signal file format they can be changed at any moment.
Once changed, locale parameters are saved in model and
will be used for future processing of data. Locale setting
affects reading and writing of text data files and model
files.
Using Quantum Computing Optimizer.
A2.2 Starting with Quantum Computing Opti-
mizer
Main module of Quantum Computing Optimizer is a
Win32 application, called qcowin.exe. It can be used in
Windows 95, 98, NT, Me, 2000 and XP environments. In
order to start the program, you should type “QCOptimizer”
in command prompt or double-click QCOptimizer icon in
windows explorer. During process of model optimization
QCOptimizer creates temporary files in current directo-
ry. Those files have “mms” extensions. Before starting
QCOptimizer check that current directory does not con-
tain files with same extension.
A2.2.1 SCOptimizer Program Window
Main program window of QCOptimizer is shown on Fig-
ure A10.
QCOptimizer window is mostly occupied with model
display. It displays current control system model as a set
of connected blocks. Each block corresponds to some
data-processing module in control system. Lines show
how inputs and outputs of modules are connected. Model
display allows access to each module configuration and
allows manipulation of modules and connections.
Distributed under creative commons license 4.0
Figure A10. QCOptimizer program window
A2.2.2 QCOptimizer Toolbar
QCOptimizer toolbar is located at the upper part of the
main window. Buttons on the toolbar provide access for
model-editing and optimization functions. They will be
described in detail in next sections.
A2.2.3 QCOptimizer Menu
QCOptimizer also has standard window menu, displayed
on the top of program window. To choose a command
with the mouse or keyboard, first select the menu and then
choose command you want. Each underlined character in
menu and command names corresponds to the key you
can press to select a menu or to choose a command.
To choose a command by using the mouse you should
first click the menu name containing the command you
want and then click the command name in the pop-up
menu.
To choose a command by using the keyboard first press
the ALT key. The File menu appears selected. Use left and
right arrow keys to select required menu and then press
down arrow or ENTER to open pop-up menu. Use up and
down arrows to select desired command and press ENTER
to activate it. You can also press one of the underlined
characters to quickly select desired menu or command.
A2.2.3.1 FILE Menu
QCOptimizer allows project files to be managed by the
means of commands in the FILE Menu.
The following items are available in this menu:
New: closes actual model and starts creation of new
model.
Open: open existing model from disk file.
Import: import SCOptimizer file to the current model.
Save: save actual model to the file.
DOI: https://doi.org/10.30564/aia.v2i2.1395 55
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Close: close current model without saving.
Signal Manager: activate learning signal manager.
Number Format: set number format conventions for
teaching signal and model files.
Exit: closes QCOptimizer.
A2.2.3.2 VIEW Menu
View menu can be used to open signal view windows:
Open signal view: displays new signal window.
A2.2.3.3 HELP Menu
Help menu currently have only one command About,
which displays version information of QCOptimizer.
A2.2.4 Dialog Boxes
When you choose a command having options, QCOp-
timizer shows a dialog box. A dialog box may contain
fields in which you can enter text, numbers or select some
items. Typical dialog box will have OK button, which you
should press after filling all fields in order to activate the
command, and CANCEL button, which will abort com-
mand execution. You can also use ENTER key on the key-
board to activate the same function as OK button and ESC
as CANCEL.
If an operation requires many parameters Wizard-style
dialog box will be used. Wizard box consists of a sequence
of dialog boxes that will guide you through the steps of an
operation. Three buttons will be available for navigation
of the wizard box. NEXT>> button tells wizard that you
have successfully completed filling of current page and
want to switch to the next page or perform the command
if no additional parameters are required. <<BACK button
should be used to return to one of the previous pages, if
you want to change some parameters. CANCEL button
aborts command execution.
A2.2.5 QCOptimizer Files
SCOptimizer saves project in special files: model files
(with *.sc2 extension). Model file is used to store all mod-
el data, including model blocks, their properties, connec-
tions and positions, as well as most optimization and other
settings.
QCOptimizer can save genetic algorithm optimization
state to disk file, so you can continue optimization process
interrupted some time ago. Those files have *.st extension
and will be called state file.
If you wish to copy QCOptimizer files to new location
(for example to another computer) you should copy model
file, teaching signal files and state files, if it is required. If
your teaching signal file was originally located in directo-
ry other than one with project file, than QCOptimizer may
be unable to load it automatically in new location. In this
56 Distributed under creative commons license 4.0
situation use Teaching Signal Manager to point QCOpti-
mizer to new location of teaching signal file.
A2.3 Working with QCOptimizer
A2.3.1 Creating New Model
Select File/New from menu to create new model. An emp-
ty model will be created and you can than add various
blocks to it.
A2.3.2 Loading Model from File
If you have a file with previously saved model you should
select File/Open command. Standard windows File Open
dialog box will appear. Select file with you model and
press Open button.
A2.3.3 Saving Model files
You can use File/Save command to save you model files.
Standard windows File Save dialog box will appear. Se-
lect file with you model and press Save button.
We recommend saving you model regularly, in order
to avoid loss of data due to possible program/computer
failure or undesired model change. Note that most
operations of QCOptimizer do not have “undo” option, so
the only way to restore to previous state is to load model
from saved file.
A2.3.4. Importing SCOptimizer Files
You can import SCOptimizer model files (.sco exten-
sion) in order to use fuzzy control systems developed in
SCOptimzier in QCOptimizer projects. Select File/Import
command to start model input. Standard windows File
Open dialog box will appear. Select file with SCOptimizer
model and press Open button.
Import options dialog will than appear as shown in Fig-
ure A11.
Figure A11. Import options dialog
Check option checkboxes as required and press OK to
import model. The options are following:
Convert input blocks to scale: by default imported
SCOptimizer model includes new input blocks that are
used to feed input data to fuzzy control system. If this
option is selected the model will be imported with scale
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
blocks instead, which will use input normalization param-
eters to scale signal. This option can be useful when creat-
ing models, which includes several fuzzy control modules
using the same input.
Convert output blocks to scale: same as before, but ap-
plied to fuzzy control output.
Load active database only: allow import of only one
fuzzy rule database from SCOptimizer model, the active
one. By default, all databases are imported and active da-
tabase is connected to output blocks.
A2.3.5 Changing Locale Settings
This operation is available from File/Number Format
menu and from New Model creation Wizard. After selec-
tion of this operation QCOptimizer will show you dialog
box, containing current settings (Figure A12).
Figure A12. Number Format dialog box
Every string in this list shows one of the locale param-
eters. Those parameters have following meaning:
List separator
List separator is a symbol or string used to separate
several successive numbers in list.
Decimal separator
Separates integer and fractional part of numbers.
Negative sign
Defines negative numbers.
Positive sign
Shows that number is positive. SCOptimizer won’t
print this string before positive numbers, but will accept
numbers with this sign in data files.
Grouping method
Define how digits of integer part of numbers will be
grouped. This is a semicolon-separated list of numbers,
defining number of digits in each group, from right to left.
Trailing 0 means “use previous value for all following
groups”. SCOptimizer will insert the thousand separators
in every place defined by grouping method, but will ac-
cept thousand separators in any position in input files.
Thousand’s separator
Separates groups of digits, defined by Grouping meth-
od.
Distributed under creative commons license 4.0
You can change any parameter by entering desired
symbol(s) to the corresponding field. There are also
two buttons, which you can use to switch to one of the
predefined configurations:
(1) System: switch to system defaults, as defined by
Windows locale settings.
(2) Matlab: switch to format, used by Matlab for text
files. Following settings are default for the Matlab:
① List separator: “;”
② Decimal separator: “.”
③ Negative sign: “-”
④ Positive sign: “+”
⑤ Grouping method: “3;”
⑥ Thousand’s separator: “ ”
When entering locale parameters be careful not to:
(3) set different parameters to the same symbol
(4) make list separator, decimal separator or negative
sign an empty string
(5) do not set any parameter to one of the following
symbols: (,),{,},:,’.
Breaking these rules may cause program not to save
and/or load files correctly.
After you set all options to desired characters press
OK to apply changes or CANCEL if you want to reset you
changes.
A2.3.6 Working with Model
You can edit model using model view in the main QCOp-
timizer window. It displays model as a set of processing
modules, called blocks, and connections between them.
You can add new blocks, delete existing, create and de-
stroy links between blocks and access block properties.
Each block is displayed as a rectangle with block name
written above of it. All blocks have some number of input
and output ports, which allow signals to be passed to and
from blocks. Input ports are drawn as squares on the left
edge of the block, and output ports are drawn as circles on
the right edge. The size of the block depends on number
of input and output blocks.
Small numbers near input and output ports show port
dimensions. For regular signals it is ‘1’, for fuzzified
signal it is number of membership functions used for
fuzzification+1. Only ports with equal dimensions can be
connected. Dimension of the port depend on each module
type and configuration.
To access various edit functions QCOptimizer toolbar
buttons are used:
(1) Select tool (Table A2): use this button to enable
select mode (default). Click on the block to select it. Se-
lected block is displayed with altered colors (depend end-
ing on windows color settings). You can select multiple
DOI: https://doi.org/10.30564/aia.v2i2.1395 57
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
blocks by holding CTRL key and clicking on blocks to be
selected. To deselect all blocks click on a free space.
Selected block[s] can be moved across display. To move,
click and hold left mouse on selected block, and drag it to
desired location. Release mouse button to stop moving.
You can also access block properties in this mode.
Double click left mouse button over block to display cor-
responding option dialog box.
(2) Move window tool (Table A2): select this button
to move display window if the model does not fit inside
screen. Click toolbar button and then click and hold left
mouse button on model view and drag mouse to change
display position. Release mouse to stop moving.
(3) Create block (Table A2): click this button to create
new block. After you click it menu with list of available
block types will be displayed. Select required block type
from this menu. Depending on block type dialog box may
appear, where you should enter initial block settings. After
finishing with block settings, you will see new block dis-
played in model view. Drag block to it’s desired location
and click mouse to finish block creation.
(4) Create link tool (Table A2): click this button to create
link between blocks. Then click to ports (one input and one
output, in any order) to create link between ports. Any output
port can have as many connections as required. Input ports
can have only one connection. If you select input port which
already have connection it will be reconnected as instructed.
(5) Delete block tool (Table A2): To delete block click
this button and then block to be deleted.
(6) Delete link tool (Table A2): To delete link click this
button and then input port, which should be disconnected.
You cannot delete links by clicking on output ports.
Table A2. The icons
Select tool
Move window tool
Create block
Create link tool
Delete block tool
Delete link tool
A2.3.7 Learning Signal Manager
If you want to use learning signal for optimization it should
be loaded and assigned for model variables before use.
QCOptimizer may load teaching signal from Matlab v.4
and v.5 files and from text files. File type is autodetected.
If the file does not look like Matlab file than it assumed to
be a text file.
58 Distributed under creative commons license 4.0
Matlab file should contain an array of real numbers
with number of columns equal to sum of number of input
and output variables of the model. If version file contains
several variables then first one will be loaded.
Text files are processed using locale settings. Please set
locale settings as described in section 4.2.5 before loading
text files. If locale settings do not match the file format
than signal will be loaded incorrectly and no error mes-
sage will typically be displayed. Text file should contain
learning signal data separated by any separators. This will
include text files produced by Matlab, CSV files and prac-
tically any other file of this kind.
To manage learning signals, select File/Signal Manag-
er command from QCOptimizer menu. Signal manager
window will be displayed on Figure A13.
Figure A13. Signal manager dialog box
To load new signal file press Load signal button. Stan-
dard windows file open dialog will be displayed. Select
file with signal and click Open to load it. Loaded files will
appear in list box labeled Loaded signals.
Click on a file name in this list box and it signal to vari-
able assignments will be displayed below. Those define
how loaded signal and model variables are related. By
default, table show “not connected” to indicate that sig-
nal variable is not assigned to model variable. To set the
assignment (or change it) click Change button. Another
dialog will appear as shown on Figure A14.
Figure A14. Variable assignment dialog box
In this dialog select block and output port name corre-
sponding to loaded teaching signal variable. Press OK to
save assignment. To assign signal to model input select
“Model Input/Output” as block name, then select corre-
sponding input or output in second list.
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Loaded and assigned signals can be used during opti-
mization, if teaching signal is selected as a source of fit-
ness function in optimization manager parameters. In this
case for model evaluation loaded signals will be presented
on model input and block outputs will be compared to as-
signed signals.
File with learning signal can be unloaded by selecting
it in the list and clicking Unload button. Unloaded signals
will not be used during optimization.
A2.3.8. Available blocks
Blocks are building parts of a control system models. Fol-
lowing block types are supported:
(1) Model input: feeds input data to model;
(2) Model output: used to pass calculated output from
model to external world;
(3) Fuzzification: performs fuzzification of signal;
(4) Inference: fuzzy rule database, performing Sugeno
inference;
(5) Constant: generates constant signal;
(6) Quantum generalization: perform quantum general-
ization algorithm;
(7) File: injects data from external file to model;
(8) Sum: calculates weighted sum of its inputs;
(9) Scale: perform scale and offset on signal.
Blocks are described in the following sections.
A2.3.8.1 Model Input Blocks
Input blocks are used to pass incoming data to model.
They do not have input ports and receive data from exter-
nal connection or assigned learning signal file.
Input blocks are numbered from 0, and the index of
block is shown after block name. Vector signals from ex-
ternal connection are separated to components and send to
input blocks according to their index. Block index depend
on the order in which blocks where created and can only
be changed by deleting blocks with lower indexes.
Input blocks also support signal normalization. To alter
signal normalization parameters double click block. Fol-
lowing dialog will be displayed on Figure A15.
Figure A15. Input block parameter dialog box
Distributed under creative commons license 4.0
In this dialog you can change block name (initially
equals block index) and normalization parameters. Signal
range corresponding to current not-normalized signal are
shown below.
Press OK to accept changes and close window. Press
Apply to change block parameters without closing
window. Press Cancel to close window without applying
modifications.
A2.3.8.2 Model Output Blocks
Output blocks are used to pass results of calculation to
external world. Data arriving at output blocks is then sent
over external link or compared to assigned learning sig-
nals.
Output blocks are numbered from 0, and the index of
block is shown after block name. Output signal sent over
external connection are compiled from signals of output
blocks in order of their indexes. Block index depend on
the order in which blocks where created and can only be
changed by deleting blocks with lower indexes.
Output blocks support signal denormalization. To alter
signal normalization parameters double click block. Fol-
lowing dialog will be displayed on Figure 16.
Figure A16. Input block parameter dialog box
In this dialog you can change block name (initially
equals block index) and normalization parameters. Signal
range corresponding to current not-normalized signal are
shown below.
Note that output block perform demoralization on the
signal, so that directly connected input and output blocks
with equal parameters will leave signal intact. Input block
offset signal to offset parameter and than scale it, output
block perform reverse operation.
Press OK to accept changes and close window. Press
Apply to change block parameters without closing
window. Press Cancel to close window without applying
DOI: https://doi.org/10.30564/aia.v2i2.1395 59
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
modifications.
A2.3.8.3 Fuzzification Blocks
Fuzzification blocks perform fuzzification of input signal.
During block creation dialog box will appear where you
should enter number of fuzzy membership functions and
block name (Figure A17).
Figure A17. Fuzzification block creation dialog
Once created fuzzification block parameters can be set us-
ing properties dialog (Figure A18).
Figure A18. Fuzzification block parameter dialog
There is the name of the block atop of the dialog.
Then there is display membership functions (MF’s) of
the variable.
Graphical window display distribution functions of
MF’s. You can change appearance of this window by the
pop-up menu, activated by the right-click of the mouse in
the window. Menu items are the following:
Track cursor: Use this feature to see the margins of al-
pha-levels. When mouse cursor is on the y-axis then lines
representing alpha level will be drawn, as well as color
lines which will show margins of this level for all MF’s.
When cursor is somewhere else on the window then verti-
cal lines at the position of cursor and horizontal lines from
intersections of this line with MF’s will be drawn.
Display MF supports: Display supports of MF’s using
colored vertical lines.
Display signal interval: Display margins of signal
change interval with vertical lines.
Color shapes: Draw functions using filled color figures
(default).
60 Distributed under creative commons license 4.0
Color lines: Draw functions using color lines.
B&W lines: Draw functions using black lines.
Save Image: Save current image to file (Windows BMP
format).
You can use this window to change MF distribution
parameters. Move mouse to the x coordinate of the modal
value or support margin of one of the MF’s. Colored line
will appear showing selected parameter. Press and hold
left mouse button. Move mouse left or right to change
the parameter. New shape of the MF will be drawn using
outline method. Release left mouse button when you are
satisfied with the shape of MF.
The list in the very bottom of the page display member-
ship functions and their parameters. First column of the
list is MF name, next distribution type followed by distri-
bution parameters.
You can change those parameters by double clicking
list items. If you do, the dialog box shown on Figure A19
will appear.
Figure A19. Membership function parameters dialog
This dialog box display all parameters of the member-
ship function. Parameters can be changed by entering new
data into corresponding fields. Please note that some pa-
rameters may not be available for different distributions.
For example, Descending normal distribution require only
Modal Value parameter and Right Fuzzy Coefficient pa-
rameter. Fields, corresponding to the unused parameters,
are grayed and can not be changed.
When you are done with this dialog press OK to apply
you changes or Cancel to return without modifying the
variable.
You can also use this dialog to delete membership func-
tions of input variables. If you wish to do it press Delete
button.
Add MF button allow you to manually add new mem-
bership function to the current variable. After you press
this button the same dialog as was used for editing new
variable will appear and you will be able to enter param-
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
eters of newly added membership function. If you press
OK MF with those parameters will be added, if you press
CANCEL it won’t. Since rule database structure is high-
ly dependant on number of MF’s, rule database will be
cleared and recreated if you add a MF.
A2.3.8.4 Inference Blocks
Inference block support fuzzy inference with Sugeno
model. When the block is being created following dialog
will appear, where you should enter initial block parame-
ters (Figure A20)
Figure A20. Inference block creation page
To view or edit inference block properties double click
the block. Dialog box as shown on Figure A21 will ap-
pear.
Figure A21. Inference block parameter dialog
This dialog displays following parameters:
Rule database type: type of the database used in model.
Can be Complete rule database or LBRW Rule database.
Distributed under creative commons license 4.0
Maximal number of rules: Maximal number of rules for
current model.
Total rules in db: Number of rules stored in the data-
base. For complete database this is always equal to max-
imal number of rules. For LBRW database it can be less
then maximal.
Show rules for output: Rule database is displayed for
one of the outputs. This list selects output variable for
which database will be displayed.
Selected rule: Displays textual representation of select-
ed rule, if any.
Rule Database editor display database as network with
four layers. First layer is input variable layer. Each circle
in this layer represents input variable. Second layer is in-
put MF layer. Circles of this layer represent membership
functions of variables. Circles in the third layer represent
rules of the database. Number written inside this circle is
a rule number of the rule in the database. Last layer is the
output layer. For Mamdani model output layer is com-
posed of circles, corresponding to membership functions
of selected output variable. For Sugeno models output lay-
er displays numerical parameters of the rule.
Database structure is shown with lines, connecting
different layers. Each node in the rule level is linked with
those MF’s in input MF layer, which are included in the
if-part of the rule. It is also linked with output MF or nu-
merical parameter of the then-part.
Since all rules of the model won’t fit on display, scroll
bar is implemented which scrolls nodes of the rule level.
You can select a rule from the database by clicking on
the node of the rule level. Textual representation of the
rule will be shown in the Selected rule field and you will
be able to edit or delete this rule. When you select a rule
in the database it activation level is displayed as red lines
on the Rule Activation Level page.
If you wish to delete the rule, select it and press delete
button. Rule will be removed from the database. Note that
rules from complete database can not be deleted.
By pressing Edit button, you can change selected rule.
The dialog window as shown on Figure A22 will appear.
Figure A22. Add/Edit Rule dialog for Mamdani model
DOI: https://doi.org/10.30564/aia.v2i2.1395 61
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
The left part of the dialog represents if-part of the rule,
and the right part corresponds to the then-part. You can
change parameters of any part by selecting items from the
list and changing values in the drop-down box below the
list.
For Sugeno 0 model this dialog will have a slightly dif-
ferent appearance (Figure A23).
Figure A23. Add/Edit Rule dialog for Sugeno 0 model
To change output parameter for Sugeno 0 model select
corresponding line from the list, enter new value in the
text field below and press set.
You can add rule to the database by pressing Add but-
ton. The same dialog as used for rule editing will appear.
Change values in the if-part and then-part as desired and
press OK to add rule. If the rule with selected if-part al-
ready exists in the database it will be replaced with new
rule.
A2.3.8.5. Constant Blocks
Constant blocks can be used to create constant signals in
the model. During block creation the dialog box as shown
on Figure A24 will appear.
Figure A24. Constant block parameter dialog
Enter number of block outputs desired and configura-
tion of each output, as follows:
(1) Value is constant value
(2) Minimal and Maximal values are range of value
62 Distributed under creative commons license 4.0
search for block optimization
(3) Cyclic is indicate that value have cyclic property
which is used during optimization
Number of block outputs cannot be changed after the
block is created. Other parameters can be changed by acti-
vating parameter dialog by double clicking on the block.
A2.3.8.6 Quantum Generalization Blocks
During creation of quantum generalization block you
should enter initial parameters in the dialog box shown on
Figure A25.
Figure A25. Quantum block creation dialog
Enter block name, number of inputs and outputs,
size of correlation vector. Correlation vector size can be
changed later, while number of inputs and outputs cannot.
When done press OK to create block.
Configuration of block can be checked and changed by
double clicking on the block. The dialog box as shown on
Figure A26 will appear.
Figure A26. Quantum block parameters dialog
Main page of parameter dialog show configuration of
inputs and outputs. Table in the input section lists follow-
ing configurable parameters for each input:
(1) Name: input name, to be used in correlation config-
uration;
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
(2) Record enable: check to enable histogram recording
for selected channel;
(3) Decode: uncheck to skip this input during decoding
(required by QPID algorithm).
To the left of table there is a histogram review window.
It shows histogram for the channel selected in the drop-
down list above it. Clear All button can be used to erase
all recorded histograms for all channels.
Delay time control in the top part of the section deter-
mine amount of delay introduced for temporal correlation.
In the output section there is a table with output param-
eters:
(1) Name: output name, to be used in correlation con-
figuration;
(2) Scale coef.: output is scaled to this constant.
Output scale coefficients can be optimized. Optimiza-
tion settings can be accessed in dialog box activated by
Optimization Parameters button (Figure A27).
Figure A27. Quantum block optimization parameters
dialog
The dialog allow selection of parameter search range,
default value, number of bits for binary GA chromosomes
and periodic property, if any.
Quantum correlation can be configured by pressing
Correlation parameters button (Figure A28).
Figure A28. Quantum correlation parameters dialog
Input field at the top part of window allow changing
correlation vector size. Table below lists current correla-
Distributed under creative commons license 4.0
tion vectors. Each to columns of the table correspond to
real (one with “_r” appended) and virtual (“_v”) parts of a
vector for each output.
Entries in the table show which value will be used at
which position of a vector. Those values can be changed
by clicking on them and selecting new value from menu.
Name of the component is composed of input name (or
input number if name is not specified), and modifier, as
follows:
(1) _r: real component is used;
(2) _r_t: real delayed component used;
(3) _v: virtual component used;
(4) _v_d: virtual delayed.
Quantum algorithm requires the histogram to be load-
ed for each channel. Histograms can be recorded during
block operation or by processing file from disk. In the first
case check Record enable checkbox in input table of cor-
responding inputs and run signal through the model. Data
appeared on inputs corresponding to checked table lines
will be recorded in histogram. To load histogram press
“Load from file button” (Figure A29).
Figure A29. Histogram load dialog
In the dialog appear first press Load button and select
file to be used. File name and number of rows and col-
umns will be displayed. Then select if the data used will
be taken from row or column (depending on file dimen-
sions and format). Select file row/column index and index
of input where histogram should be loaded. If required
enter normalization parameters. Press Make histogram to
load specified data. You can than select another indexes
and load histograms to other inputs by pressing Make his-
togram again. Press Cancel when done with loading.
A2.3.8.7 File Blocks
File block can be used to insert data from external file to
the model. Block support reading of Matlab v.5 and v.6
and text files. File type is detected automatically.
During of block creation the dialog box as shown on
Figure A30 will appear.
DOI: https://doi.org/10.30564/aia.v2i2.1395 63
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Figure A30. File block parameter dialog
Select file you wish to use by clicking on “…” button
and selecting file name using standard windows file open
dialog. Data from file can be normalized before sending to
block output. Normalization parameters should be entered
to the corresponding fields of the dialog. Press OK to cre-
ate block.
During operation file block output one record from the
file for every model calculation step. Current record index
is shown in block properties dialog and can be reset to
zero by pressing Rewind button. Block automatically re-
starts from first record when all records from file are read.
A2.3.8.8 Sum Blocks
Sum block calculate weighted average of its inputs.
During creation of a block dialog window will appear
where you should select required number of inputs and
initial weights (Figure A31).
Figure A31. Sum block parameter dialog
Enter desired number of inputs in input count field,
weight factors in scale column in input configuration table
and press OK to create the block.
You cannot change number of inputs of sum block, but
input weights can be changed. To do it double click the
block. The same dialog window will appear, where you
can set weight factors as desired.
A2.3.8.9 Scale Blocks
Scale blocks transform input signal by scaling and off-
setting it. To set scale and offset parameters double click
64 Distributed under creative commons license 4.0
the block (Figure A32).
Figure A32. Scale block parameter dialog
Set desired parameters in dialog and press OK or Apply
to change block parameters.
A2.3.9 External Link
QCOptimizer support data exchange with external pro-
grams, which can be used to pass data from external pro-
gram as input to QCOptimizer model and return resulting
model output. This link uses the same interface as one
available for external inference in SCOptimizer, and can
be accessed from SCLib library functions.
If current fitness calculation mode is set to “teaching
signal” external link is not used and cannot be activated.
Change fitness calculation mode to Matlab/external signal
to enable external link access.
To connect via external link QCOptimizer and anoth-
er program should use the same link name. Default link
name is SC2. Link name in SCOptimizer can be changed
by pressing toolbar button. Dialog box will appear
where you can enter desired link name (Figure A33).
Figure A33. Link name configuration dialog
Current link name is also displayed in the status line in
the left-bottom corner of main QCOptimizer window pre-
ceding with “IPC:”.
To enable external link press toolbar button. To indi-
cate that QCOptimizer is ready for external connection an
‘*’ symbol will be shown after link name in status line. To
disable external connection press , ‘*” will disappear.
Always enable external link first from QCOptimizer
and than in external program. Close link in reverse order –
first in external program, then in QCOptimizer.
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
A2.3.10 Viewing Data Passing through the Model
You can view signals passing through the model in the
graphical representation. To view signal select View/Open
signal view menu item. Signal view window will appear,
showing signals on model input. To change signal shown
in window click right mouse button above window. In the
menu appear select block (or model input/output) and than
block output.
You can open as many data view windows as you like,
by selecting the same menu item several times.
A2.3.11 Testing Model on Input Signal from File
You can make QCOptimizer perform calculations using
input signal recorded in file. To do so you should:
(1) Load signal using Teaching Signal Manager and
associate it with model inputs.
(2) Select “teaching signal” fitness mode calculation in
optimization configuration.
Once those two actions are done, you should see “file”
displayed in the status line typically used to show external
link name. Press toolbar button to perform a full run
through data available in loaded files. You can view re-
sulting signals using signal view windows, or you can use
this mode to record histograms in quantum block.
A2.3.12 Optimization
QCOptimizer support various optimization algorithms
and modes. It can perform optimization of single block, or
several blocks at a time, including optimization with dif-
ferent algorithms, running at the same time and evaluated
over same link.
A2.3.12.1 Selecting Optimization Options
To set optimization settings press toolbar button (Fig-
ure A34).
Figure A34. Optimization settings
Distributed under creative commons license 4.0
First select desired fitness calculation mode:
(1) Teaching signal: model is evaluated using teaching
signal loaded in Teaching Signal Manager and fitness is
computed by comparing actual data from model with data
selected as teaching signal.
(2) Fitness calculation in matlab: during optimization
QCOptimizer set model to the state being evaluated and
calls matlab function specified in configuration. Function
should evaluate model quality (typically using external
link to do some calculations) and return fitness value. To
work in this mode matlab should be available and one
should create three functions and set it names to following
parameters:
① Initialize session command: called by QCOptimizer
when optimization is started. Should perform initializa-
tion, for example load some simulink model.
② Fitness calculation command: called by QCOpti-
mizer to calculate fitness for current state.
③ Close session command: called when optimization
is finished. May free resources allocated for calculations.
(3) Fitness calculation from external signal. Signal
coming from external link is used for model input and
fitness calculation. Concrete fitness function is configured
during optimization start, but here you can enter initial
conditions and calculation time. Initial conditions are set
in Initial state table where you can enter desired initial
value for some variables and possible delta. In example
shown on figure above optimization will start when output
1 of quantum block 432 will equal 1±2. Once initial con-
dition is met QCOptimizer will wait for delay time set in
wait before averaging field and than calculate fitness func-
tion as average over averaging time selected.
In this window you can also select what QCOptimizer
will do if it detects that the same model state evaluation
is requested during single optimization process. If do not
check for duplicate chromosomes is selected than dupli-
cate detection will be disabled and all states will be eval-
uated. You can greatly improve optimization time by se-
lecting single evaluation mode. In this case you may also
want to set hash table size used to store processed states.
A2.3.12.2 Starting Optimization
To start optimization press button. Optimization start
dialog will appear (Figure A35).
DOI: https://doi.org/10.30564/aia.v2i2.1395 65
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Figure A35. Optimization start dialog
First select optimization algorithm from list. Blocks
available for optimization are shown in optimization target
lists. Select blocks to be optimized by clicking on them.
Below optimization target list there is a fitness func-
tion configuration table. In “fitness calculation in Matlab”
mode those settings are not used. For NSGA algorithm
you can use more than one fitness function. In the table
select desired singal, optimization mode (for minimum,
for maximum, or to match a value).
You can start several instances of the same algorithm
operating over different blocks by setting parallel start
option to desired number. Several pages will appear where
you can select targets for all algorithms.
Once you press OK you will be prompted to enter al-
gorithm parameters (depending on algorithm selection).
When you confirm algorithm settings optimization will
start.
A2.3.12.3 Optimization Progress/Evaluation
When optimization is active optimization status dialog is
shown on Figure A36.
Figure A36. Optimization progress dialog
66 Distributed under creative commons license 4.0
For each optimization currently active it displays a
line, which shows current status, generation and fitness
achieved. You can abort one of algorithms by selecting it
and pressing Abort button. Pressing Abort All will abort
all active algorithms. Algorithms may be aborted with
some delay, if there is a calculation which cannot be
stopped immediately.
Press Pause calculations to pause all calculations.
Current evaluations will be restarted from beginning after
calculation is resumed.
When calculation is over (or after each generation if
you select corresponding option in algorithm configura-
tion) the dialog will appear showing state of current popu-
lation (Figure A37).
Figure A37. Optimization progress dialog
The table lists all parameters for all chromosomes, as
well as evaluation, fitness and other algorithm-dependent
parameters. Top line shows best chromosome available,
next lists chromosomes of current generation. You can
save this data to text (CSV) file by pressing Save to CSV
button. Select chromosome and press Try solution to set
model state to chromosome selected, so that you can eval-
uate it manually.
If this dialog is shown after optimization is over, you
should select a chromosome which will be used as opti-
mization result (by default best one shown in top line is
used). When you are finished this chromosome will be
applied to model.
References
[1] Ulyanov, S. V. System and method for control using
quantum soft computing [P]. US Patent No 7,383,235
B1, 2003; EP PCT 1 083 520 A2, 2001; Efficient
simulation system of quantum algorithm gates on
classical computer based on fast algorithm [P]. US
Patent No 2006/0224547 A1, 2006.
[2] Ulyanov, S. V. Quantum fast algorithm computation-
al intelligence PT I: SW / HW smart toolkit [J]. Arti-
ficial Intelligence Advances, 2019, 1(1): 18-43.
DOI: https://doi.org/10.30564/aia.v2i2.1395
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
DOI: https://doi.org/10.30564/aia.v1i1.619
[3] Grover, L. K. A fast quantum mechanical algorithm-
for database search[P]. US Patent US 6,317,766
B1, 2001.
[4] Litvintseva, L. V. et al. Quantum fuzzy conclusion
for creating knowledge bases in robust intelligent
regulators[J]. RAN News. Theory and control sys-
tems, 2007, 6: 71–126. (in Russian).
[5] Litvintseva, L. V., Ulyanov, S. V. Intelligent control
systems. 1. Quantum computing and self-organiza-
tion algorithm[J]. RAN News. Theory and control
systems, 2009, 6: 102–141. (in Russian).
[6] Ulyanov, S. V., Sorokin, S. V., Litvintseva, L. V.
Knowledge Base Optimizer based on quantum com-
puting for designing self-organizing fuzzy control-
lers: software tools[J]. System analysis in science
and education, 2012, 1: 1–27. (in Russian).
[7] Nielsen, M. A., Chuang, I. L. Quantum computation
and quantum information[M]. UK, Cambridge, Uni-
versity Press, 2000: 700.
[8] Marinescu, D. C., Marinescu G. M. Approaching
quantum computing[M]. New Jersey: Pearson Pren-
tice Hall, 2005: 400.
[9] Nikolaeva, A. V., Ulyanov, S. V. Design of an intel-
ligent robot control system by a manipulator. Part 3:
Modeling and physical experiment based on quantum
computing technologies[J]. System analysis in sci-
Distributed under creative commons license 4.0
ence and education: a network scientific publication,
2013, 1: 1-25. (in Russian).
[10] Ulyanov, S. V., Litvintseva, L. V., Mishin, A.A.,
Sorokin, S. V. Certificate of state registration of
computer programs No. 2011619257. Optimizer of
robust knowledge bases for the design of intelligent
control systems on soft computing[P]. Application
No. 2011617532 dated 11.10.2011 RF Registered in
the Register of computer programs on December 1,
2011. (in Russian).
[11] Mikhailov, V. S. Control Theory[M]. K: Higher
School, 1988: 312. (in Russian).
[12] Nikolaeva, A. V., Ulyanov, S. V., Barkhatova, I. A.,
Nozdrachev, A. V.Intelligent control system for re-
dundant robotic arm with seven degrees of freedom
based on soft computing[J]. System analysis in sci-
ence and education: a network scientific publication,
2014, 2: 48-55p. (in Russian).
[13] Ulyanov, S. V. Self-Organized Intelligent Robust
Control Based on Quantum Fuzzy Inference [J] Re-
cent Advances in Robust Control - Novel Approach-
es and Design Methods, ISBN: 978-953-307-339-2,
InTech, 2011. Available from:
http://www.intechopen.com/books/recent-advanc-
es-in-robust-controlnovel-approaches-and-de-
sign-methods/self-organized-intelligent-robust-con-
trol-based-on-quantum-fuzzy-inference.
DOI: https://doi.org/10.30564/aia.v2i2.1395 67
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Artificial Intelligence Advances

https://ojs.bilpublishing.com/index.php/aia

ARTICLE
Robotic Smart Prosthesis Arm with BCI and Kansei / Kawaii / Affective
Engineering Approach. Pt I: Quantum Soft Computing Supremacy
Alexey V. Nemchaninov1　Alena V. Nikolaeva1　Sergey V. Ulyanov1*　Andrey G. Reshetnikov1,2
1. INESYS LLC (EFKO GROUP), Russia
2. Dubna State University, Russia

ARTICLE INFO ABSTRACT

Article history A description of the design stage and results of the development of the
Received: 22 November 2019 conceptual structure of a robotic prosthesis arm is given. As a result, a
prototype of man-made smart prosthesis on a 3D printer as well as a foun-
Accepted: 3 June 2020 dation for computational intelligence presented. The application of soft
Published Online: 30 June 2020 computing technology (the first step of IT) allows to extract knowledge
directly from the physical signal of the electroencephalogram, as well as
Keywords: to form knowledge-based intelligent robust control of the lower perform-
Robotic prosthetic arm ing level taking into account the assessment of the patient’s emotional
state. The possibilities of applying quantum soft computing technologies
Cognitive computational intelligence (the second step of IT) in the processes of robust filtering of electroen-
Brain-computer-device neurointerface cephalogram signals for the formation of mental commands of robotic
Mental commands prosthetic arm discussed. Quantum supremacy benchmark of intelligent
control simulation demonstrated.
Quantum soft computing
Fuzzy cognitive controller
Quantum supremacy benchmark

　
1. Introduction ligent information technologies in the field of intelligent

T
he development of robotic human limbs prostheses
and the production of human-like electromechani-
cal devices - anthropomorphic robots are receiving
more, but not sufficient, attention in both the scientific,
technical and socio-economic plants [1-4]. It should also
be noted that in the strategy for the development of arti-
ficial intelligence (AI) in the Russian Federation [5] in the
Healthcare section, the application of AI in such an im-
portant socio-technical domain as intelligent prosthetics
and smart cognitive control systems, as well as rehabilita-
tion of disabled people, is not indicated at all.
Principles of the development and application of intel-
control using cognitive technologies (“brain-computer-de-
vice” interface - BCI) are considered. Neurointerface of
this type (manmade interface) allow to restore and expand
the capabilities of a person with physical (for example,
disability with loss of limbs) or mental disorders in var-
ious activities (for example, autistic children or patients
with impaired mental activity - dementia). Cognitive
interfaces provide the ability to communicate, evaluate
emotions, transfer and control devices with mental com-
mands.
The process of developing a cognitive intelligent simu-
lator toolkit discussed. The application of developed tool-

*Corresponding Author:
Sergey V. Ulyanov,
INESYS LLC (EFKO GROUP), Russia;
Email: ulyanovsv@mail.ru

68 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
kit provides developers with the ability to control robotic
devices at the lower (executive) level (using the so-called
“mental commands”), and at the upper level - the intelligent
level (to develop cognitive intelligent control technologies
with the possibility of application in applied tasks).
Let us consider briefly main principles, peculiarities
and features of cognitive intelligent control applied in
biomechanical products and presented the description
of a hierarchical intelligent control system based on
QSCOptKBTM (knowledge base optimizer on quantum soft
computing).
2. Cognitive Intelligent Control: Design Prin-
ciples and Features of Biomechanical Product
Sources of technological and breakthrough innovations in
these areas are following: (1) new technologies for creating
intelligent materials; (2) technologies for creating an intelli-
gent software product integrated into devices and applied at
all stages of interaction with devices; (3) new human-ma-
chine interfaces, the principles of which are based primarily
on the method of reading the activity of the functioning of
the brain and nerve endings of the body [2,5].
So, the first direction allows a person to restore (and in
the future, to exceed) the functional state of limbs dam-
aged as a result of any injuries due to the creation of more
advanced alloys, material structures, nano coatings. The
second, innovative direction is associated with the cre-
ation of sophisticated software that allows the biomechan-
ical device to learn and adapt to individual physiological
and psychological qualities and characteristics of the
human operator. Through the application of deep machine
learning and medical recommendation systems with a
deep knowledge representation, it is possible to recognize
complex human commands, read and recognize the emo-
tional state of the operator [6 - 8]. At the same time, a certain
computing basis in the form of embedded end-to-end
technologies of cognitive computing and computational
intelligence must correspond to software of this level. The
development of the third direction based on the sources of
new human-machine interfaces that can effectively com-
plement and expand human information capabilities. Such
interfaces include infrared - spectrometers, electroenceph-
alographs, magnetoencephalographs, cognitive helmets
and equipment for virtual and complementary (augmented)
3D - and 4D - reality, invasive and non-invasive sensors
and beacons, for example, mounted on the wrist, or other
parts of the human body [9 - 15].
2.1 Related Works
Many researches known in which patients routinely
Distributed under creative commons license 4.0
apply such interfaces to solve everyday problems and
control various devices. Interfaces are actively used for
rehabilitation and diagnostic procedures, helping to im-
prove interaction with the human environment, including
with robotic devices [10-14,16,17]. Technologies are actively
involved quantum end-to-end technologies in EEG data
processing and educational processes at state levels [18].
Research of this kind has been funded by states since the
early 70’s. There are a number of research collaborations
on the creation and development of man - machine inter-
faces associated with all three areas [10-14,19,20] etc. In par-
ticular, research in this area can divided into the following
groups:
(1) Recognition research - development of devices for
the diagnosis, modeling, simplification and reduction of
threats to the interaction of the brain with the system.
(2) Simulation of the brain mechanism - the use of neu-
ral network effects and the phenomena of the functioning
of the brain in applied problems of information technolo-
gy, for example, analysis and synthesis of information.
(3) Restorative medicine - restoration of behavioral
cognitive functions lost as result of damage to the brain or
body [8].
(4) Elaboration - development of brain-computer sys-
tems in the feedback loop to accelerate and improve the
functional behavior of the system [1, 21].
The development of these researches made it possi-
ble to create new technologies for the neural interface to
detect fundamental and interregional brain functions in
online, as well as to develop complex mathematical algo-
rithms to model brain activity and the resulting behavioral
functions and reactions.
2.2 Architecture of the Limbic System
The Limbic System, as part of the mammalian creatures’
brain, is mainly in charge of the emotional processes. (see
Appendix 1). The Limbic System located in the cerebral
cortex consists mainly of following components: Amygda-
la, Orbitofrontal Cortex, Thalamus, Sensory Cortex, Hy-
pothalamus, Hippocampus and some other less important
areas. We try to describe briefly these main components
and their tasks.
Figure 1 illustrates the anatomy of the main compo-
nents of Limbic System [22].
The first sign of affective conditioning of the system
appears in Amygdala which is a small almond-shaped in
sub-cortical area. This component placed in a way to com-
municate with all other Sensory Cortices and areas within
the Limbic System.
DOI: https://doi.org/10.30564/aia.v2i2.1567 69
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Figure 1. The major brain structures associated with the
Limbic System [22]
The Amygdala connections to other components illus-
trated in Figure 2 [23].
Figure 2. Connections of the Amygdala with other com-
ponents of the Limbic System [23]
The studies show that a stimulus and its emotional
consequences are associated in the Amygdala area. In this
region, highly analyzed stimuli in the Sensory Cortices, as
well as coarsely categorized stimuli in the Thalamus are
associated with an emotional value. Furthermore, one of
the most complex and twisting components of the Lim-
bic System is Hippocampus which is located in the same
area as the Amygdala. Its main role is the mapping of the
environment based on environmental cue. The Hippocam-
pus has other functions such as spatial navigation, laying
70 Distributed under creative commons license 4.0
down of the long-term memory and formation of the con-
textual representations there are other components, which
have specific role in the Limbic System. To that extent
components such as Basal Ganglia, Globus Pallidus, Sub-
stantia Nigra, Subthalamic Nucleus and Periamygdaloid
Cortex could be mentioned. Since in this paper, biological
description of the Limbic System is not under focus, it has
been tried to avoid detailed and comprehensive explana-
tion of each component. We deal with the key characteris-
tics components in the System.
A computational model developed that mimics Amyg-
dala, Orbitofrontal Cortex, Thalamus, Sensory Input Cor-
tex and generally those parts of the brain thought respon-
sible for processing emotions.
Figure 3 shows the computational model of emotional
learning [24].
Figure 3. Graphical depiction of the Brain Emotional
Learning (BEL) process [24]
The model divided into two parts: The Amygdala and
the Orbitofrontal cortex. The Amygdala part receives in-
puts from the Thalamus and from cortical areas, while the
Orbitofrontal obtains inputs from the cortical areas and
the Amygdala. The system also receives a reinforcing sig-
nal (Primary Reward) which been left unspecified, as it is
still uncertain from where it comes.
2.3 Brain Emotional Learning Based Intelligent
Controller
Based on the cognitively motivated open loop model,
BELBIC- Brain Emotional Learning Based Intelligent
Controller- was introduced by Lucas et al [24, 25]. The in-
telligent controller has been utilized by several industrial
applications and control purposes. The model of Figure 3
illustrated as control blocks in Figure 4 [25].
The BELBIC is essentially an action generation mech-
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

anism based on sensory inputs and emotional cues (Reward amputation, send signals (red arrows) or the cortical re-
signals) (see Appendix 1). organization (red star in the brain) generates the phantom
limp pain. Other sensations that can felt involve tingling,
cramping, heat, and cold.

Figure 4. Basic block structure of emotional controller [25]

Figure 5 demonstrates a reasonable candidate for em-
bedding the BELBIC model within a typical feedback
control block diagram [25]. The implemented functions in Figure 7. Working of neural prosthetics using a brain-ma-
emotional cue and sensory input blocks defined for each chine interface (a). Phantom limb pain depiction (b)
application [22 - 25].
Depending on the type of amputation, (see Table 1)
the most suitable type of prostheses selected (see Figure
8a). The choice of prosthesis design determined by the
position of phantom pains (see Figure 7b). On Figure 8b
presented an artificial limb which replaces an arm missing
above the elbow. The complexity of the artificial limb de-
pends on the level of amputation.
Figure 5. Control system configuration using BELBIC [25]
Table 1. Presents Classification of Prosthetic as per ampu-
Example. A locally linear learning algorithm called Lo-
tation
cally Linear Mode Tree (LoLiMoT) applied to build the
neuro - fuzzy model. Then, the BELBIC based on PID con- N Type of Amputation Type of Prostheses
1 Shoulder disarticulation From Shoulder
trol adopted for the micro-heat exchanger plant. The perfor-
Below Elbow
mance of presented controller compared with classic PID - 2 Elbow disarticulation
Above Elbow
controller. Figures 6a and 6b show the closed-loop system 3 Wrist disarticulation Below Elbow
response using BELBIC and PID - controller respectively. 4 Trans carpal disarticulation Below Elbow
5 Finger amputations Below Elbow

Automated Prosthetic arms considered as biomedical de-

vices and developing the same is interdisciplinary activity,
i.e. combination of mechanisms and electronics. The selec-
tion of prosthetic arm depends upon type of the disarticula-
tion the patient has undergone and the patients need.

Figure 6. Closed - loop system response using PID with

LoLiMoT identifier [26] (a).Closed-loop system response
using BELBIC with LoLiMoT identifier [26] (b)
As shown the performance of the system using BEL-
BIC is much better than that of PID controller.
Example. Figure 7a demonstrates afferent somatosen-
sory signal that taken from the prosthetic device and is
fed into the brain, from where the motor signal is sent
back to the prosthetic limb. The nerve endings (located
at the red circle). Figure 7b, still present at the site of the Figure 8. Amputation level (a). Transhumeral Prosthesis (b)

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567 71

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Figure 9 shows some of the ways to determine funda-
mental and interregional brain functions online: behavior-
al functions and reaction control of robotic limb prosthet-
ics using different BCI.
Figure 9. Detect fundamental and interregional brain
functions in on line: behavioral functions and response
control of robotic prosthetics limb with BCI
Figure 10 shows the generalized structure of an intel-
ligent cognitive control system with feedback based on
deep machine learning using artificial neural networks
with an optimal structure, taking into account existing
approaches to the cognitive control of a robotic prosthetic
arm [27 - 29].
Figure 10. The structure of cognitive intelligent control
system
In the considered part of the work cycle, the existing
hardware and software research basis and information
technology of sophisticated class of cognitive intelligent
control system (see Figure 11) for supporting the design
and operation of a new class of devices are presented.
72 Distributed under creative commons license 4.0
Figure 11. Structure of Quantum cognitive intelligent
control system
Remark. According [17] the application of quantum neu-
ral network in filtering processing of EEG signals increase
the accuracy of the filtering processing. The architecture
of recurrent quantum neural network (RQNN) model is
based on the principles of quantum mechanics with the
Schrodinger wave equation playing a major part. This
approach enables the online estimation of a time-vary-
ing probability density function that allows estimating
and removing the noise from the raw EEG signal and
demonstrate quantum supremacy in signal data process-
ing. Applications of sophisticated intelligence toolkit
(QSCOptKBTM) include universal approximating func-
tions of RQNN with optimal structure for quantum deep
machine learning and achieved quantum soft computing
supremacy [21] (see below Figures 23 and 24).
Structure of cognitive intelligent control system on
Figure 11 includes two controllers: fuzzy and cognitive
controllers. Design of fuzzy controller and of cognitive
controller KBs is achieved with SCOptKBTM toolkit [30, 31].
In this case the responses of fuzzy and cognitive control-
lers in general case with imperfect KB are inputs for box
“quantum fuzzy inference (QFI)” and the output of QFI is
robust KB of self-organized controller for forming in on
line time dependent control laws of coefficient gain sched-
ule for conventional controller of robotic prosthetic arm.
3. Cognitive Intelligent Control of the Pros-
thetic Arm: Quantum Soft Computing Ap-
proach
Let us consider the redistribution problem of the level of
responsibility between the cognitive and fuzzy controller.
The basis of prostheses and robotic manipulators is spatial
mechanisms with many degrees of freedom. According-
ly, when designing a cognitive intelligent controller for
a prosthesis, the intelligent control system of the robotic
arm can take as the basis. In this example, three fuzzy
controllers are implemented in the structure of the in-
telligent control system of the robot manipulator with 3
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Degrees of Freedom (DoF), each of which controls one of

the three links independently (see Figure 12).
Among unforeseen situations, the management distin-
guish external and internal. To external unforeseen situa-
tion it is customary to include disturbing influences - such
as a forced change in the position of the control object at
the beginning or in the process of working, changing the
reference signals, etc. Internal control situations include
changes to the parameters of the components of the con-
trol system (limits, noise, and signal delays). In details,
intelligent control system on soft computing described in
[30 - 32]
.
To prevent the negative effects of unforeseen situa-
tions in control, a quantum fuzzy inference (QFI) unit
integrated in the control system of the robot manipulator,
with the application of which KB self-organization in FC
is realized. In this case, it is possible not only to combine
information about the standard situation for three links,
but also to extract additional information (by methods of
quantum computing and quantum information theory)
from the response of the designed KB for the implemen-
tation of robust control in regular and unforeseen control
situations (which are not included in existing designed
KB).
Figure 13. The methodology of extraction of hidden
information from relationships between KB, designed for
learning control situations
As result of inclusion SCO on soft computing with
divided controls of the designed block Quantum fuzzy
inference, get a new type of control system - Intelligent
Control System Based on Quantum Computing.
The quantum fuzzy inference (see Figure 14) algorithm
performs the following sequence of steps:
Step1. Coding.
Step 2. Selection with quantum genetic algorithm (QGA)
the type of quantum correlation for constructing control
output signals. It is considered spatial, spatio - temporal and
temporal correlations (all three types are mixed)
Spatial. Output dependency K P ( t ) , K D ( t ) , K I (t )
i _ new i _ new i _ new
is determined by the correlation of the following sets of
input coefficients:

{K ( t ) , K ( t ) , K ( t ) , K ( t ) , K ( t ) , K ( t )} → K ( t )
1
P
2
P
3
P
1
D
2
D
3
D
new
P

{K ( t ) , K ( t ) , K ( t ) , K ( t ) , K ( t ) , K ( t )} → K ( t ) ,
1
D
2
D
3
D
1
I
2
I
3
I
new
D

{K ( t ) , K ( t ) , K ( t ) , K ( t ) , K ( t ) , K ( t )} → K ( t )
1
I
2
I
3
I
1
P
2
P
3
P
new
I

Where each set is an entangled state:

a1a2 a3 a4 a5 a6 = K P1 ( t ) , K P2 ( t ) , K P3 ( t ) , K D1 ( t ) , K D2 ( t ) , K D3 ( t ) .

Figure 12. Intelligent control system with separate con-

trol for Soft Computing Optimizer of KB for manipulator
control with 3DoF
Remark. On Figure 12: K - matrix of proportional, dif-
ferential and integral coefficients PID-regulator, Kpi, Kdi,
Kii , i=1,3 , i - the number of link of the robot manipulator,
TSi - training signal, GAi - a genetic algorithm that gener-
ates a learning signal for design a knowledge base.
Moreover, the knowledge bases themselves, whose
responses are used to design robust control, in emergen-
cies may not be robust (imperfect KBs). The scheme for
extracting hidden information about the relationships
between existing FC (designed using soft computing
technologies) for the three links of the manipulator with
knowledge base obtained for regular control situations us-
ing the quantum fuzzy sensing unit shown on Figure 13 [32]. Figure 14. Quantum fuzzy inference algorithm sequence
of steps

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567 73

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Spatio - temporal: solves the problem of the exact positioning of the manipu-
lator robot in contrast to the intelligent control systems on
{K ( t ) , K ( t − ∆t ) , K ( t ) , K ( t − ∆t ) , K ( t ) , K ( t − ∆t )} → K ( t )
1
P
1
D
2
P
2
D
3
P
3
D
new
P the SCO on soft computing.
{K ( t ) , K ( t − ∆t ) , K ( t ) , K ( t − ∆t ) , K ( t ) , K ( t − ∆t )} → K ( t ).
1
D
1
I
2
D
2
I
3
D
3
I
new
D
Let us demonstrate the operation of the intelligent con-
trol systems on the SCO on soft computing with divided
{K ( t ) , K ( t − ∆t ) , K ( t ) , K ( t − ∆t ) , K ( t ) , K ( t − ∆t )} → K ( t )
1
I
1
P
2
I
2
P
3
I
3
P
new
I
control in the conditions of an external unforeseen control
Temporal: situation (Figure 17) in comparison with the intelligent
control systems in the SCO on quantum computing. As an
{K ( t ) , K ( t ) , K ( t ) , K ( t − ∆t ) , K ( t − ∆t ) , K ( t − ∆t )} → K ( t )
1
P
2
P
3
P
1
P
2
P
3
P
new
P
unforeseen situation is the forced displacement of the sec-
{K ( t ) , K ( t ) , K ( t ) , K ( t − ∆t ) , K ( t − ∆t ) , K ( t − ∆t )} → K ( t ).
1 2 3 1 2 3 new ond link of 3DOF manipulator.
D D D D D D D

{K ( t ) , K ( t ) , K ( t ) , K ( t − ∆t ) , K ( t − ∆t ) , K ( t − ∆t )} → K ( t )
1
I
2
I
3
I
1
I
2
I
3
I
new
I

Step 3. Building a superposition of entangled states.

Step 4. Intelligent quantum state measurement.
Step 5. Decoding.
Step 6. Denormalization.
The overall assessment of the quality of control is high-
er in the case of applying the intelligent control systems
on the QCO on quantum computing (for all considered
types of correlations) compared to the intelligent control
systems on the SCO on soft computing with conventional
control. Which is a consequence of introducing into the
structure of the intelligent control systems an additional
quantum fuzzy inference link organizing coordination
management. Moreover, if for a robot manipulator with 3
DoF as result of testing MatLab / Simulink models, it had
the best performance intelligent control systems, using
spatio - temporal correlation, then physical testing deter- Figure 15. The movement of manipulator in a standard
control situation: under control of ICS based on SCO
mined the use of spatial correlation to be the most opti-
with soft computing (a). ICS based on SCO with quantum
mal. For a robot manipulator with 7 DoF, in most cases,
computing (b)
spatial correlation was also optimal.
For a robotic arm as robot with 3 DoF, the overall as-
sessment of control quality is improved when using intel-
ligent control systems on QCO with quantum computing
compared to using intelligent control systems on SCO on
soft computing with one FC. Intelligent control systems
on SCO with soft computing with one FC, it is able to
solve the positioning problem in the conditions of external
unforeseen situations, but it does not always cope with the
occurrence of internal unforeseen situations.
Example. On Figures 15 and 16 demonstrated the oper-
ation of the manipulator Intelligent Control System (ICS)
when using the intelligent control systems on the SCO (on
soft computing) and the intelligent control systems on the
QCO (on quantum computing) in the conditions of a reg-
ular control situation and an external unforeseen control Figure 16. The movement of manipulator in an external
situation (the initial position is changed). unforeseen situation: under control of ICS based on SCO
Quantum supremacy. As can be seen from Figures 15 with soft calculations (a). ICS based on SCO with quan-
and 16, the intelligent control systems on the QCO (on tum computing (b)
quantum computing) under the conditions of the consid- From Figure 17 it is seen that in the considered un-
ered standard and external unforeseen control situations

74 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

foreseen situation, the control of the intelligent control
systems on the SCO on quantum computing copes with
the task of positioning with a given accuracy, in contrast
to the intelligent control systems on the SCO on soft com-
puting with a divided control.
FC is created and the amount of memory in the system in
which the FC is located.
Moreover, for complex control objects, such as a robot
manipulator with 7DoF, the organization of a single FC is
impossible.
In general, the possibility of decomposing the control
(dividing one knowledge base into several identical inde-
pendent knowledge bases) and organizing coordination
management by introducing the quantum fuzzy factor link
significantly increases the possible number of input vari-
ables and thereby expands the possibilities of accounting
for the parameters of the system and the control object.

3.1 Information-thermodynamic Analysis of Cog-

Figure 17. The work of the intelligent control systems on
the SCO on soft computing with divided control in an un-
foreseen control situation (a); intelligent control systems
operation on SCO on quantum computing (b)
nitive Intelligent Control System
The distribution (trade-off) of control qualities determined
by the following system of equations:

The inability of the intelligent control system on SCO dV n

on soft computing to solve the problem of exact positional dt

(
= ∑qi ⋅ ϕ q, t , k ( SO − ( SТc + SCc ) ) , u )
 
i =1

control is also illustrated in Figure 18. FC, responsible Stability Controllability
for managing the second link for the allotted time, was (1)
not able to “rehabilitate” after a powerful external impact,
as result of which the positioning error of the second 
( (
+ ( SO − ( SТc + SCc ) ) ⋅ SO − SТc + SCc ≤ 0 ))
Robustness
link was more than 50 degrees, the control goal was not
achieved and the control system as a whole was not robust
(Figure 17).
dV n

dt
(
= ∑qi ⋅ ϕ q, t , k ( H O − ( H Тc + H Cc ) ) , u )
 
i =1
 
Stability Controllability

(2)
( )
+ ( H O − ( H Тc + H Cc ) ) ⋅ H O − ( H Тc + H Cc ) ≤ 0

Robustness

Figure 18. Changing the position of the second link under

the control of the intelligent control system on SCO on
soft computing with divided control
The results presented on Figures 17 and 18 demon-
strated quantum soft computing supremacy in intelligent
robust control and can guarantee goal control achievement
with required accuracy (in case on Figure 17 the accura-
cy of control goal achievement achieved more than 120
times).
This result is especially noteworthy in the situation
when organizing coordination control due to a single
knowledge base (correspondingly using one FC in the in-
telligent control systems for the SCO on soft computing)
the number of input variables. For this case parameters
(that determine the functioning of the system) limited by
the computing resources of the processor on which the
where and means the total thermodynamic and infor-
mation entropies of the conventional intelligent (Tc) and
Cognitive (Cc) controllers, respectively.
It follows from Eq. (1) that the robustness of an intelli-
gent control system can increased by the entropy produc-
tion of the cognitive controller, which reduces the loss of
useful work, and Eq. (2) shows that the negentropy of the
cognitive controller reduces the minimum requirements
for initial information amount to achieve robustness.
3.2 Information Control Extraction and Action
Work
Moreover, information based on knowledge in the knowl-
edge base of the cognitive regulator allows get an addi-
tional resource for useful work, which is equivalent to the
appearance of a targeted action on the control object to

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567 75

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
ensure the guaranteed achievement of the control goal.
One of the key tasks of modern robotics is the devel-
opment of technologies for the cognitive interaction of ro-
botic systems, which allow solving the tasks of intelligent
hierarchical management by redistributing knowledge and
control functions, for example, traditionally between a
leader and a subordinate (“master - slave” system). Mod-
ern approaches to solving this problem based on the theo-
ry of multi - agent systems, the theory of swarm artificial
intelligence, and many others [33 - 35].
Due to the synergetic effect, an additional information
resource created and the multi-agent system is able to
solve complex dynamic tasks for performing team mutu-
al work. The given task may not fulfill by each element
(agent) of the system separately in various environments
without external management, control or coordination,
however, exchange of knowledge and information allows
performing useful teamwork to achieve the management
goal under the conditions of uncertainty of the initial in-
formation and limited consumption of useful resources.
In particular, it is known that for closed-loop control
systems, the amount of useful work W that is extracted
t
=
satisfies the inequality Wmax ( t ) k Tmin Ic dt ′ ≤ kTI c ,
∫ easy to see the speed of the recoverable work Wext limited
0
where k is the Boltzmann constant, Tmin ( t ) is interpret-
ed as the lowest achievable temperature by the system in
time t for feedback control, assuming Tmin ( 0 ) = T and Ic
determines the amount of Shannon information (entropy
transfer) , extracted by the system from the measurement
process [36].
Physically, the synergetic effect means the design of
knowledge self-organization and the extraction of addi-
tional information that allows the multi-agent system to
perform the most useful work with a minimum loss of
useful resource and with a minimum of the required initial
information, and without destroying the lower executive
level of the control system.
Together with the information-thermodynamic law of
intelligent control (optimal distribution of the manage-
ment qualities as “stability - controllability - robustness”),
an intelligent control system (ICS) is designed with multi-
agent systems, ensuring the achievement of the manage-
ment goal under the conditions of uncertain initial infor-
mation and limited useful resource [37].
A network of loosely coupled groups of robots working
together to solve tasks that go beyond individual capa-
bilities. Different nodes of such a system, as a rule, have
a different intelligent level (knowledge, algorithms, and
computational bases) and various information resources
for design process. Each node should be able to modify
76 Distributed under creative commons license 4.0
its behavior depending on the circumstances, as well as
to plan its communication and cooperation strategies with
other nodes. Here the indicators of the level of coopera-
tion are the nature of the distribution of tasks, the unifica-
tion of various information resources and, of course, the
possibility of solving a common problem in a bounded
time interval.
Example. As noted above, if microscopic degrees of
freedom are available to the observer in the form of the
Maxwell demon, then the second law of thermodynam-
ics can violated. Szilard showed from an analysis of the
Maxwell demon model that work extracted from the ther-
modynamic cycle in the form as kT ln 2 . Moreover, it
was shown [38] that recoverable work WextS from the system
is determined by the amount of information (or quan-
tum-classical mutual information) I, which estimate the
extracted knowledge of the system from measurements.
At the same time, a similar ratio in the form of a lower
M
boundary exists for the full cost Wcost measuring and eras-
M
ing information WextS ≤ −∆F S + kTI and Wcost ≥ kTI , where
∆F S determines the free energy of the system. Then it is
by Wext ≤ kTI , i.e. limited by the speed of information re-
trieved [39].
Model of self-organized intelligent control system pro-
posed based on the principles of minimum informational
entropy (in the “intelligent” state of control signals) and
the generalized thermodynamic measure of entropy pro-
duction (in the system “control object + regulator”). The
main result of the application of the self-organization pro-
cess is the acquisition of the necessary level of robustness
and the flexibility (adaptability) of the reproducible struc-
ture.
It has noted that the property of robustness (by its phys-
ical nature) acts as an integral part of self-organization,
and the required level of robustness in intelligent control
system achieved by fulfilling the principle of minimum
production of generalized entropy noted above. The prin-
ciple of minimum entropy production in object control
and system control serves as the physical principle of op-
timal functioning with a minimum consumption of useful
work and underlies the development of robust intelligent
control system.
This statement is based on the fact that for the general
case of controlling dynamic objects, the optimal solu-
tion to the finite variational problem of determining the
maximum useful work W according to, it is equivalent
to solving the final variational problem of finding the
minimum of entropy production S. Thus, the study of the
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

mands is not used in the learning process (the operator

maximum functional condition max (W ) (where qi , u are
qi ,u is fully concentrated on his own cognitive processes). In
the generalized coordinates object control and control sig- particular, the possibility of controlling the prosthetic arm
nal respectively) is equivalent, the study of the associated in on line using an electroencephalograph and the corre-
sponding neurointerface software considered.
problem of minimum entropy production, i.e. min ( S ) .
qi ,u

Therefore, in the developed model, the principle of 4. Development of Prosthetic Models

minimum informational entropy guarantees the necessary A lot of research has done on prosthetics and the imple-
condition for self-organization - the minimum of the re- mentation of projects to create new types of limbs.
quired initial information in the teaching signals; the ther-
modynamic criterion of the minimum of a new measure of 4.1 Related Prototypes of Prosthetic Arms and
generalized entropy production provides a sufficient con- Differences from Known Approaches - advantag-
dition for self-organization - the robustness of control pro- es and Disadvantages
cesses with a minimum consumption of useful resource.
More significant is the fact that the average amount of For illustration purposes, Figure 19a shows modern hand
prostheses. Figure 19b shows promising interfaces -
Wdiss
work performed by dissipation force = S KL ( PF PB ) , hands, virtual reality helmet, cognitive helmets.
kT
Remark. Such (and similar) projects have their advan-
i.e., the work of the dissipation forces is determined by the
tages and disadvantages. Many prostheses don’t have
Kullback - Leibler divergence for probability distributions
feedback from the user (they do not allow him to convey
PF , PB . Note that the left side of this relation represents
emotional or physiological sensations), cannot be a com-
physically thermal energy, and the right side defines pure-
plete replacement for a lost limb (for example, a long -
ly extracted information about the system.
term phantom of having a healthy limb). Most prostheses
Information entropy is a measure of the amount of
are expensive, and their operation and implementation
information about the system and the Kullback - Leibler
are very costly. They include invasive interfaces and have
divergence, as well as the determination of the amount of
more features, which, accordingly, affect their cost.
Fisher information.
Of the abundance of prostheses presented, three classes
A similar relationship exists between the work pro-
can be distinguished: 1) Cosmetic (do not carry any func-
duced by the forces of dissipation and the difference be-
tionality in themselves, only increase the aesthetic lev-
tween Renyi [37].
el); 2) Mechanical (use the power of levers and rods for
Thus, substituting the relations between the informa-
movement, which creates some inconvenience in the way
tion and the extracted free energy and work in (1) and (2),
they are used); and 3) Bioelectric (have great functional-
we obtain the conclusion noted above. Therefore, the ro-
ity, but are very expensive and require a constant charge
bustness of the intelligent control system can increase by
of batteries).
the entropy production of the cognitive controller, which
The ability to use 3D printing greatly facilitates the cre-
reduces the loss of useful resource of the control object;
ation of a prosthesis. There are many open resources that
negentropy of the cognitive regulator reduces the require-
provide drawings, sketches, and assembly technologies for
ments for minimum initial information to achieve robust-
various parts and mechanisms, including anthropomorphic
ness.
parts.
Moreover, the extracted information, based on knowl-
However, the interface development process accom-
edge in the knowledge base of the cognitive controller, al-
panied by complexity of execution and high cost of such
lows getting an additional resource for useful work, which
products. An open source project from the Thingiverse
is equivalent to the appearance of a targeted action on the [42, 43]
website adopted as the basis for a robotic prosthesis
control object to guarantee the achievement of the control
with a cognitive control system.
goal.
Purpose and goal of Part 1: creation of a model of
In this Part I of article the solution of intelligent robotic
an anthropomorphic intelligent robotic arm. The work
prosthetic arm design is considered.
uses the limb of a robot to demonstrate the application of
Remark. This work is a continuation of [40] and based
developed technologies. It decided to use the cognitive
on the concept of an intelligent simulator [41], which in-
helmet company EmotivEpoc+ [44] as an interface for re-
cludes advanced information technology for the design of
cording EEG signals. The combination of these devices
intelligent control systems. In contrast to the methodology
[14] creates a software and hardware basis for the laboratory
, visual reinforcement of the generation of mental com-

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567 77

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

bench of an intelligent robotic simulator. In order to implement a system of mechanical “tendons”

that transmit force from “muscle” motors to the limbs
(Figure 20b), a nylon thread with a diameter of 2 mm
used. Servomotors used as drives, which made it possible
to vary the angle of rotation of the output shaft, and then
transfer the force with the help of “tendons” to the limbs.
Note that it is possible to set the compression ratio of the
fingers. The results of the first stage of work (creating a
finished prototype) presented in Figure 20c.
To control the servomotors and control the prosthesis
itself, a unified Arduino Uno controller based on the AT-
mega 328 microcontroller used, to control many servomo-
tors, Sensor Shield v 4.0 company Arduino [45] connected
to the controller. A schematic diagram of the connection
of servomotors shown in Figure 21.

Figure 19. Modern prostheses of hands and hands (a).

Computer interfaces in the form of brushes with installed
sensors, augmented and virtual reality, cognitive interfac-
es (b). Examples of application in control tasks (c)
Let us consider in more detail the lower (executive)
level of the developed device.

4.2 Prosthesis Manufacturing

The first stage of work is the manufacture of prosthesis
parts on a 3D printer. For this purpose, ABS plastic was
used (a polymer consisting of acrylonitrile, butadiene, sty-
rene, which, due to its technical characteristics, was wide- Figure 21. Connection diagram of devices
ly used as an engineering and structural material in layout In this paper, for presenting simplicity of results, the
engineering) and a 3D printer. The project [42, 43] dedicated training and operation process performed on a personal
to the creation of an anthropomorphic robot was taken as computer, and the microcontroller used only to receive
the basis. and send specific control commands. The implementation
Figure 20 illustrates an anthropomorphic robot, sketch- scheme presented on Figure 22.
es of its individual parts and the manufactured part of the
project.

Figure 22. Microcontroller to PC connection diagram

Comment. From the point of view of information tech-
Figure 20. Project to create an anthropomorphic robot (a). nology, it is necessary to take into account the consump-
Sketches of the robot (b). Manufactured prototype pros- tion of computing resources in the process of training, up-
thesis (c) dating and operating the device. The rational distribution

78 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

of information-thermodynamic load [46], allows designing

the optimal structure of neural networks and integrating
quantum cognitive computing on-board processor of the
final device. In this case. The processor load in the learn-
ing process distributed between the central computer and
the controller of the actuator.

4.3 EEG Signal Removal and Data Information

Processing
The general concept of using a cognitive simulator is de-
scribed in [19, 20, 41].
Figure 23a, shows the filtered (refined) with
QSCOptKBTM toolkit EEG channel placement on the hu-
man scalp.
Figure 23. EEG channel placements on the human scalp
Each scalp electrode is located at the brain centers. In
2001 Pfurtscheller identified that many of the neural ac-
tivity related to fist movements are found in channels C3,
C4 and Cz as shown in Figure 23b. F7 is for rational ac-
tivities, Fz is for intentional and motivational data, P3, P4
and Pz contain perception and differentiation, T3, T4 is for
emotional processes, T5, T6 has memory functions and
O1 and O2 contain visualization data. In order to remove
the noise from the obtained signal, any of the suitable fil-
tering techniques may adopted. Further, the extracted data
may move for classification phase.
A well - known marker of cognitive processes is the
restructuring of brain rhythms, manifested in a surface-re-
corded electroencephalogram (EEG) of a person. To signal
the brain activity, we used the cognitive helmet company
Emotiv EPOC+ (see Figure 25a), and the functional dia-
gram of the software implementation is shown in Figure
24.
Figure 24. Functional structure of software in on line
In on line, the EEG data received in the block “Analysis
of the extracted data”. Then, after filtering and frequency
decomposition, the signals enter the recognition unit (see
Figure 23). The recognition result is the degree of similar-
ity with previously recorded commands during training.
As neural networks, deep machine learning patterns of
pattern recognition are used.
Further, when the activation level is exceeded, such
signals enter the fuzzy neural network of decision-mak-
ing, designed using the soft computing optimizer [30, 31].
The output of such a neural network is the target values of
the indicators of a managed device. At the same time, the
training and operation process is supported by an emo-
tional (positive or negative) reaction of the operator [21],
thereby evaluating the quality of training and adaptation
of the control system. The design of the “Cognitive Reg-
ulator” block based on quantum soft computing optimizer
considered in the Part II of this paper.
EPOC has 14 electrodes, which are passive sensors
that allow you to register electromagnetic signals. Sensors
are mounted on the surface of the skin (non-submersible,
non-invasive interface).
Figure 25 b, shows the structure of Emotiv EPOC+
consisting of channels AF3, F7, F3, FC5, T7, P7, O1, O2,
P8, T8, FC6, F4, F8, AF4 (plus CMS/DRL и P3/P4).
Figure 25. EmotivEPOC+ cognitive helmet(a).The layout
of the electrodes of the cognitive helmet Emotiv EP-
OC+(b)

Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1567 79

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
The supplied software allows receiving, recognizing
and recording the EEG signal from the helmet. To ana-
lyze the received signal, the so - called EEG frequency
rhythms are distinguished.
Remark. The term “frequency rhythm” means a certain
type of electrical activity corresponding to a certain state
of the brain, for which the boundaries of the frequency
range are defined. In the process of cognitive activity,
characteristic rhythms of the beta, alpha, theta and delta
ranges appear [47] (Figure 26).
The set of simultaneously present rhythms forms a
specific spatial-frequency EEG pattern. Patterns are char-
acteristic of different types of cognitive activity and are
highly individually specific [15, 19].The ability of an individ-
ual to establish rhythmic EEG patterns when performing
certain cognitive tasks makes up an “encephalographic”
portrait of his personality.
Comment. One of the main components of cognitive
neurointerface technology is gaming simulators. It is
important to note that during training, the control model
acts as a system model on the monitor screen. This, in
turn, allows not only to train the brain to generate men-
tal commands, but also to tune the control system of the
control object itself, adjusting it to the operator, thereby
ensuring increase of overall performance in the system
“brain - computer”. This kind of feature is because when a
person works with a cognitive helmet, the program adapts
to the characteristics of the operator, adjusting the control
system. Developers of such equipment with the level of
operator training to generate various mental commands
usually associate the quality of team recognition.
Figure 26. EEG frequency rhythms
The software bundle provides specialized games - sim-
ulators, in which the learning and training process takes
place. Typically, computer games used for this, where the
operator must perform an action that can matched with
some command used to control the object. When playing,
80 Distributed under creative commons license 4.0
the operator develops his skill in working with a helmet,
which subsequently makes it possible to control real tech-
nical devices (manipulators, wheelchairs, and other de-
vices). Training can be either active, with external stimuli
(for example, light of a certain frequency, video and audio
material, pictures, etc.), or passive if the operator presents
it without external influence during the generation of a
mental command.
Also, additional hardware, such as a virtual reality hel-
met, which will visualize the user’s actions in the virtual
world, thereby increasing the effectiveness of training,
can be used to form a user with a more complete picture
of learning. The duration of training and the intensity
of training affects the quality and number of recognized
mental teams. If the first time a person can learn how to
generate signals in the brain for one team, then for good
recognition of two or more teams, several exercises are
necessary.
Accordingly, both the psychophysiological character-
istics of a person’s state (including his position in space)
and his level of training will influence the management
system and the achievement of the management goal.
Usually (according to information from the manufacturer)
for the system to work well using four teams. It is neces-
sary to conduct regular training for 2-3 weeks, and after
training the operator experiences fatigue. And accordingly,
time is required to restore strength. The learning process
described in more detail in [37, 45]. A program based on the
Epoc cognitive helmet SDK written to control the prosthe-
sis servomotors.
The learning process consists of two main blocks: (1)
analysis of electroencephalographic indications and iden-
tification of frequency patterns specific to a particular op-
erator; (2) generating and recording mental commands.
At the training stage, using the wet non-immersed
brain-computer interface with passive electrodes of Emo-
tivEpoc+, the activity of certain parts of the operator’s
brain recorded. Initially, it is necessary to obtain a fre-
quency slice of the operator’s neutral state (the operator
is at rest), this frequency slice is necessary to improve
detection of the nervous excitation of the operator during
further analysis.
In the future, to generate mental commands, the op-
erator for a certain period of time (8 seconds) mentally
models the necessary action, which will serve as the basis
of the mental command (contraction, relaxation of various
muscle groups, etc.). And also at this moment performs
this action with «Live» hand. An individual signal record-
ed for each mental command.
From the viewpoint of the control system, the operator
must be able to re-generate the recorded signals, which
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
interpreted by the system to control the movement of the
device. The frequency rhythm shown in Figure 27.
Figure 27. Frequency rhythm of the neutral state (a) and
the state of excitation (b)
Typically, statistical methods used to evaluate and rec-
ognize signals, including the calculation of variance and
mathematical expectation.
Figure 28 demonstrate highlights the most different sig-
nals from each other, which the operator associates with a
mental command and a neutral state.
Figure 28. Dispersion of mental command signals
Dispersion of signals generated by the operator is a
fairly simple and effective method for comparing signals
of mental commands. Additional various methods de-
scribed in [47].
Figure 29 shows the interface of software tools that al-
lows you to receive and process data from the brain-com-
puter interface of EmotivEpoc+ (the program code created
in the C# programming language using the Windows
Forms graphical interface).
Figure 29. Program interface
Data exchange between the cognitive helmet and the
computing center done using the Bluetooth protocol stack,
as well as the UART protocol. To initialize a cognitive
Distributed under creative commons license 4.0
helmet, it is necessary to mark the “Connect” field, then
select a mental command from the drop-down list or cre-
ate a new one. Now of pressing the “Train” button, the
operator begins active cognitive activity associated with
the selected team. After completing the training process,
it is necessary to save the obtained data, connect the pros-
thesis to the computer, set the boundaries for activating
the mental command. And then put the application into
the prosthesis control mode (by marking the “Rсactive”
field). When click on the “Train” button, information sent
to the helmet about the currently selected mental team
for training, after which some time activity indicators of
the human brain recorded. When the helmet fixes similar
parameters of mental activity, the key of the current team
will change from neutral to the team associated with these
parameters.
In addition to connecting directly to the helmet and
prosthesis, it is possible to test the program using the
third-party program Emotiv Xavier Composer, which
simulates the work of a cognitive helmet and allows send-
ing virtual mental commands to the application. In the
program code, this implemented as a call to the Remote
Connector Connect methods.
If the user sends a mental command to the program,
then the spatial-frequency pattern evaluated, after the
scaling of the received signal relative to the set minimum
and maximum, the corresponding signal sent to the COM
port. The following event occurs when the timer is started,
which starts to control the prosthesis during the connec-
tion of the cognitive helmet, which is shown on the main
screen.
Comment. The variable “power” is the power of mental
effort. Since the prosthesis has physical limitations during
operation (it is not possible to send servomotors an an-
gle of rotation that extends beyond the interval [0.180]), the
variable “a” trimmed in possible values to the boundaries
of this interval. In the program code, the second possi-
ble value is another such construct. Thus, a binary tree
implemented. After the operations performed, a line of
the form “a A” is sent to the COM port, where “A” is the
angle from 0 to 180, by which all five servomotors will be
rotated, taking into account the physical restrictions previ-
ously established in the Arduino program for each one in
particular. Such methods prevent possible damage to the
prosthesis if the commands coming from the helmet are
misinterpreted.
Figure 30 shows a graph showing the frequency rhythm
of the operator. The trained cognitive control system, ana-
lyzing this frequency pattern, recognizes the mental com-
mands present in the frequency rhythm.
DOI: https://doi.org/10.30564/aia.v2i2.1567 81
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Figure 30. Execution of mental commands
Figure 31 shows a graph showing the registration of
mental commands at discrete time intervals.
Figure 31. Regulatory signaling for the actuator
Figure 32 shows the process of the experiment.
Figure 32. Conducting an experiment: execution of the
“unload” command (a), the execution of the command
“compression” (b)
The operator, using mental commands that read by the
EPOC+ cognitive interface, controls the compression-un-
loading process of the prosthesis brush.
In accordance with the scheme shown in Figure 10,
EmotivEpoc in on line captures the excitations of the
cerebral cortex at the time of a certain cognitive activity
(generation of mental command). The received data goes
to the computer center, where they are recognized. Based
on the analysis of the obtained frequency pattern and
estimation of the signal level at a discrete time instant, a
control action generated for the actuator.
82 Distributed under creative commons license 4.0
This experiment demonstrates the possibilities of sharing
electromechanical, anthropomorphic devices of a modern
class of computer neurointerfaces and software, which in turn
is one of the strongest incentives for the development of end-
to-end digital information technologies of “strong” artificial
intelligence and intelligent assistants of service use.
5. Conclusion
(1) This work (the first part) presents the development
of several high-tech areas of robotics, which have practi-
cal scientific and technical interest, both in separate and in
joint developments.
(2) It has been shown that the prospect of developing
cognitive intellectual control using soft and quantum
computing technologies is one of the important tasks in
creating a robotic prosthetic arm, such as a simple case of
a robot avatar, and is integral to the development of infor-
mation technology in the framework of the concept of an
intelligent simulator [41].
(3) The use of expert recommendation systems with a
deep representation of knowledge [6,7] and quantum end-to-
end technologies of deep machine learning with quantum
EEG processing [17,49] allows the appointment, selection
of control of robotic prostheses of the hand, taking into
account the individual psychophysiological characteristics
of the patient and the operating environment.
(4) On the one hand, these products if properly devel-
oped can presented on the market of commercially attrac-
tive products, on the other hand, technologies for using
new types of intelligent information technologies and
human-machine cognitive interfaces.
(5) The next stage of development is the creation of a
cognitive intellectual control system for a robotic arm-pros-
thesis for maintenance based on IT quantum soft comput-
ing, quantum EEG processing filters [17] and Kansei / Affec-
tive Engineering intelligent computing technologies with an
assessment of the user’s emotional state [49 - 51].
(6) The work, in its essence, reflects the completeness
of the formation of a new educational approach in intel-
ligent robotics [32, 40, 41, 52] - a hybrid cognitive intelligent
robotics based on neural interfaces with new types of IT
data processing.
Appendix 1. Contemporary models of affect, motiva-
tion, emotion and cognitive control for Kansei /Kawaii /
Affective Engineering
Let us briefly consider the models of negative affect,
pain and emotion that play important role in Kansei engi-
neering and cognitive control.
Example. In humans and other primates, the cingulate
- a thick belt of cortex encircling the corpus callosum - is
one of the most prominent features on the mesial surface
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
of the brain (Figure 33a). Early research suggested that
the rostral cingulate cortex (Brodmann’s ‘precingulate’;
architectonic areas) plays a key part in affect and motiva-
tion (Figure 33b). More recent research has enlarged the
breadth of functions ascribed to this region; in addition to
emotion, the rostral cingulate cortex has a central role in
contemporary models of pain and cognitive control. The
most basic question is whether emotion, pain and cogni-
tive control are segregated into distinct subdivisions of
the rostral cingulate or are instead integrated in a common
region. There is a growing recognition that aMCC might
implement a domain-general process that is integral to
negative affect, pain and cognitive control [22].
Cognitive control is a range of elementary processes (such
as attention, inhibition and learning) that are engaged when
automatic or habitual responses are insufficient to sustain
goal-directed behavior. Control can be engaged proactively
or reactively. Activation foci maps between negative affect,
pain and cognitive control is shown on Figure 33d.
Figure 33. (a) - Divisions of the human rostral cingulate
Distributed under creative commons license 4.0
cortex; (b) - Negative affect, pain and cognitive control
activate a common region within the anterior subdivision
of the midcingulate cortex - aMCC (The map depicts the
results of a coordinate-based meta-analysis (CBMA) of
380 activation foci derived from 192 experiments and
involving more than 3,000 participants); (c) - Activation
likelihood estimate (AlE) maps of the three behavioral do-
mains (left) and pairwise ALE minimum conjuction maps;
(d) - Activation foci maps [22]
Example: Emotional learning occurs mainly in the
amygdala. The system operation consists of two levels:
the amygdale learns to predict and react to a given rein-
forcement signal. This subsystem cannot unlearn a con-
nection. The incompatibility between predictions and the
actual reinforcement signals causes inappropriate respons-
es from the amygdala. As depicted in Figure 34, the sys-
tem on Figure 34a, consists of four main parts [54]. Sensory
input signals first enter the thalamus. Since the thalamus
must provide a fast response to stimuli, in this model the
maximum over all stimuli S is sent directly to the amyg-
dala as another input. The amygdala receives inputs from
the thalamus and sensory cortex, while the orbitofrontal
cortex (OFC) part receives inputs from the sensory cortex
and the amygdala. The system also receives a reinforcing
signal (REW - emotional signal).
For each Anode in the amygdala, there is a plastic con-
nection weight Vi. Any input is multiplied by this weight
to provide the output of the node. The O nodes show sim-
ilar behavior, with a connection weight Wi applied to the
input signal to create an output. There is one output node
in common for all outputs of the model, called E (see Fig-
ure 34a). The E node sums the outputs from A except for
the Ath and then subtracts the inhibitory outputs from the O
nodes. The result is the output from the model.
Figure 34. Graphical depiction of the brain emotional
learning process (a) and the process of generating emo-
tional reactions in the limbic part of human brain (b)
The OFC learns to prevent the system output if such
mismatches occur. The learning in the amygdala and the
OFC is performed by updating the plastic connection
weights, based on the received reinforcing and stimulus
signals.
Example. Let us consider briefly Brain Emotional
DOI: https://doi.org/10.30564/aia.v2i2.1567 83
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Learning Based Intelligent Controller (BELBIC) structure
[53]
. In a biological system, emotional reactions are utilized
for fast decision‐making in complex environments or
emergency situations. It is thought that the amygdala and
the orbitofrontal cortex are the most important parts of the
brain involved in emotional reactions and learning. The
amygdala is a small structure in the medial temporal lobe
of the brain that is thought to be responsible for the emo-
tional evaluation of stimuli. This evaluation is in turn used
as a basis for emotional states and reactions and is used for
attention signals and laying down long‐term memories.
The amygdala and the orbitofrontal cortex compute their
outputs based on the emotional signal (the reinforcing
signal) received from the environment. The final output
(the emotional reaction) is calculated by subtracting the
amygdala’s output from the orbitofrontal cortex’s (OFC)
output (see Figure 34b). A control system strategy, based
on brain emotional learning, was proposed by Caro Lucas
in the early 2000s (Lucas et al., 2004). The limbic model
used was based on the neural link, between the amygdale
and the orbitofrontal cortex, proposed by Balkenius and
Moren. This control paradigm is commonly designated by
BELBIC which stands for brain emotional learning based
intelligent control. The reasoning behind the integration of
the limbic model into a closed loop control system can be
tracked down to the seemingly robust way that the brain
performs decision making. Actually, control has all to
do with decision making: the controller goal is to devise
the best input actions based on the incoming information
according to the system states. These actions can be tak-
en considering the past, the present or even forecasts on
the future system states. Hence the controller produces a
mapping between its input signals and the output control
signals by means of an arbitrary decision function which
can be described by means of differential equations, as
in PID-controllers, or by an inference mechanism such
as in Fuzzy or Neuro-Fuzzy control. Alternatively, it can
be based on the result of the optimization of a cost func-
tion such as linear quadratic regulators (LQR) or model
predictive control (MPC). In the BELBIC control system
architecture this input-to-output transformation is imposed
by means of the limbic system model. In this case, both
the external stimuli and reward signals are generated in
such a way as to produce a closed loop system response
according to some target characteristics. In addition, due
to the recursive nature of the weights update law, this con-
troller is able to gradually learn how to handle changes in
the system dynamics. A key point in BELBIC is the exter-
nal stimulus and reward signals definition.
Notice that there are not universal rules to carry out this
task. This choice is flexible and must be custom defined
84 Distributed under creative commons license 4.0
according to the end application. For example in Lucas
et al. the reward signal r(t) is obtained as a weighted sum
of the error signal and the control effort and the external
stimulus signal i(t) is defined as a linear combination of
the system output and its first derivative.
It should be observed that it essentially converts two
sets of inputs (sensory inputs and emotional cues or rein-
forcing signals) into the decision signal (the emotional re-
action) as its output. Closed loop configurations using this
block (BELBIC) in the feedforward control loop of the to-
tal system in an appropriate manner have been implement-
ed so that the input signals have the proper interpretations.
The block implicitly implemented the critic, the learning
algorithm and the action selection mechanism used in the
functional implementations of emotionally based (or, gen-
erally, reinforcement learning based) controllers, all at the
same time.
Reza Keramat et al. [54] consider that in practical sys-
tems important information originates from two sourc-
es. One of the sources is the experts, who define their
knowledge of the system using the natural language. The
other consists of measurements and mathematical models
derived from physical laws. Hence, what matters is how
to incorporate these types of information into the design
of systems. The question is that how it is possible to for-
mulate human knowledge within a framework similar to
mathematical models.
Basically, the main function of a fuzzy system is to
make such a conversion possible. Fuzzy systems are
based on knowledge or rules. The core of a fuzzy system
is a knowledge base following the IF-THEN fuzzy rules.
If a fuzzy system is used by a controller, the controller is
called “fuzzy controller”.
The fuzzy controller designed for the full bridge DC-
DC converter takes two inputs: error (e) and error varia-
tions (Δe). The membership input functions for this con-
troller are in the range of (-5, 5). Membership functions
for each of the two input components of error and error
variations are shown in Figure 35.
Figure 35. The membership functions for e and Δe
Seven membership functions are used here for the in-
put: negative big (NB), negative medium (NM), negative
small (NS), zero (Z), positive big (PB), positive medium
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
(PM), and positive small (PS). The controller output is as-
sumed to be equal to the duty cycle, which varies between
0 and 0.5. Figure 36 displays the controller output func-
tions.
Figure 36. The output membership functions
Reza Keramat et al. show the effectiveness of BELBIC
for the full bridge DC-DC converter. Figure 37 shows the
system output voltage after the application of the fuzzy
and BELBIC controllers.
Figure 37. The comparison of the fuzzy controller with
the BELBIC
As seen in this figure, the BELBIC has a slight over-
shoot while the fuzzy controller has a considerable over-
shoot (the initial overshoot is 35 V). The time required for
stabilization is almost the same in both cases. It is there-
fore concluded that the BELBIC outperforms the fuzzy
controller.
In Reza Keramat’s experiment the BELBIC controller
outperforms the fuzzy controller. Considering the uncer-
tainty of system parameters (including inductance, ca-
pacitance, and input voltage and acceptable variations of
load), the BELBIC presents better performance than the
fuzzy controller. Although fuzzy control is a robust and
effective method for a large number of engineering sys-
tems, but its design (and consequently its performance) is
almost depend on the experience and tact of the designer.
Furthermore, after design and installation, its performance
is not improved, and in other word, it is not a learn-
ing-based or intelligent controller.
It can be stated that a Learning Based Intelligent Con-
trol which “may” in the first step, act not as satisfying as
any another modern controller, any straight-forward-de-
signed controller after few iterations thanks to its learning
Distributed under creative commons license 4.0
automata feature. Based on this deduction, we felt no need
to emphasize the comparison of these controllers after set-
ting optimization.
References
[1] Robbin A. Miranda, Casebeer William D., Heinc
Amy M., etc.. DARPA-funded efforts in the devel-
opment of novel brain-computer interface technolo-
gies[J]. Journal of Neuroscience Methods, 2015, 244:
52-67.
[2] An analytical review of the global robotics market
2019[M]. Sberbank Robotic Lab., Sberbank, 2019,
272.
[3] Guangzhong Yang, Jim Bellingham, Pierre E. Du-
pont, etc. The grand challenges of Science Robot-
ics[J]. Sci. Robot. 3, 2018, eaar7650: 1-14.
[4] An analytical review of the global robotics mar-
ket[R]. Laboratory of Robotics, Sberbank. 2018: 79.
[5] Artificial Intelligence: Approaches to Formation. AI
development strategies in the Russian Federation[R].
Sberbank, 2019.
[6] Amirova E.K., Efimov V.A., Ulyanov S.V., etc. Ex-
pert system for selecting lower-extremity (thigh)
prostheses and diagnosis of the quality of artificial
replacement. Pt 1[J]. ISSN 0006-3398, Consultants
bureau, New York, J. of Biomedical Engineering,
May-June, 1991, 3: 26-31.
[7] Amirova E.K., Efimov V.A., Kuzhekin A.P., Ulyanov
S.V., etc. An expert system for selecting lower-ex-
tremity (thigh) prosthesis and evaluation of prosthet-
ic quality. Part 2[J]. ISSN 0006-3398, Consultants
bureau, New York, J. of Biomedical Engineering,
November-December, 1991, 6: 5-12.
[8] Lupina I.V., Slepchenko A.N., Ulyanov S.V., etc. Hy-
brid expert system with an in-depth representation of
knowledge for the design and diagnosis of biotechno-
logical products[J]. Izv. USSR Academy of Sciences.
Ser. Technical cybernetics, 1991, 5(152-175).
[9] Kaplan A., Shishkin S., Ganin I., etc. The prospects
of the P300-based brain computer interface in game
control[M]. IEEE Transactions on Computational In-
telligence and AI in Games, 2013
[10] Shishkin S.L., Ganin I.P., Kaplan A.Y. Event-related
potentials in a moving matrix modification of the
P300 brain-computer interface paradigm[J]. Neuro-
science Letters, 2011, 496(2): 95-99.
[11] Leigh R. Hochberg, Mijail D. Serruya, Gerhard M.
Friehs, etc. Neuronal ensemble control of prosthetic
devices by a human with tetraplegia[J]. Nature, 2006,
442: 164-171.
[12] Höller Y., Bergmann J., Kronbichler M., etc. E. Real
movement vs. motor imagery in healthy subjects[J].
DOI: https://doi.org/10.30564/aia.v2i2.1567 85
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
Int. J. Psychophysiol, 2012: S0167-8760.
[13] Salvaris M., Cinel C., Citi L., etc. Novel protocols
for P300-based brain-computer interfaces[J]. IEEE
Trans. Neural. Syst. Rehabil. Eng., 2012, 20: 8-17.
[14] Musallam S., Corneil B.D., Greger B., etc. Cognitive
Control Signals for Neural Prosthetics[J]. Science,
2004, 305: 258-262.
[15] Ivanitsky G. A., Determining the nature of mental ac-
tivity according to the rhythmic pattern of the EEG]
[A]. in Russia. Available URL:
https://www.rfbr.ru/rffi/portal/project_search/
o_53385
[16] Sunny T.D., Aparna T., Neethu P., etc. Robotic Arm
with Brain Computer Interfacing[M]. International
Conference on Emerging Trends in Engineering, Sci-
ence and Technology, ICETEST, 2015.
[17] Gandhi V. Brain-Computer Interfacing for Assistive
Robotics 1st Edition: Electroencephalograms, Re-
current Quantum Neural Networks, and User-Centric
Graphical Interfaces[J]. N.Y. Academic Press, 2015.
[18] Meng Jing. Brain Co CEO says his “mind-reading”
tech is here to improve concentration, not surveil-
lance[M]. South China Morning Post May, 2019.
Available URL:
https://www.scmp.com/tech/innovation/arti -
cle/3008439/brainco-ceo-says-his-mind-reading-
tech-here-improve-concentration
[19] Kazuki Yanagisawa, Hitoshi Tsunashima, Kaoru
Sakatani. Brain-Computer Interface Using Near
Infrared Spectroscopy for Rehabilitation, Infrared
Spectroscopy[J]. Life and Biomedical Sciences, Prof.
Theophile (Ed.), ISBN: 978-953-51-0538-1, InTech.
Available URL:
https://www.intechopen.com/books/infrared-spec-
troscopy-life-and-biomedical-sciences/brain-comput-
er-interface-using-near-infrared-spectroscopy-for-re-
habilitation
[20] Vaughan TM, McFarland DJ, Schalk G, etc.. The
Wadsworth BCI Research and Development Pro-
gram: at home with BCI[J]. IEEE Trans Neural Syst.
Eng., 2006, 14(2): 229-233.
[21] Ulyanov S. V., Shevchenko A. V., Mamaeva A. A.,
etc. Soft Computing Optimizer as Deep Machine
Learning in Intelligent Cognitive Estimation of Hu-
man Being Emotion for Robotic Control[J]. Robots
and Engineering Cybernetics, 2019.
[22] D. Purves, G. J. Augustine, D. Fitzpatrick, etc.. Neu-
roscience[M]. Sinauer Associates, 2001.
[23] J. Moren. Emotion and Learning - A Computational
Model of the Amygdala[D]. Lund University, Lund,
Sweden, 2002.
[24] J. Moren, C. Balkenius, A Computational Model of
86 Distributed under creative commons license 4.0
Emotional Learning in the Amygdala [J], Cybernetics
and Systems, 2000, 32(6): 611-636.
[25] C. Lucas, D. Shahmirzadi, N. Sheikholeslami. Intro-
ducing BELBIC: Brain Emotional Learning Based
Intelligent Controller[J]. International Journal of
Intelligent Automation and Soft Computing, 2004,
10(1): 11-22.
[26] H. Rouhani, M. Jalili, B. N. Araabi, W. Eppler, C.
Lucas. Brain emotional learning based intelligent
controller applied to neuro-fuzzy model of mi-
cro-heat exchanger[J]. Expert Systems with Applica-
tions, 2007, 32: 911-924.
[27] Rajesh P. N. Rao, Paul G. Allen. Towards Neural
Co-Processors for the Brain: Combining Decoding
and Encoding in Brain-Computer Interfaces[J]. In-
vited submission to the journal Current Opinion in
Neurobiology, 2018.
[28] José del R. Millán, Pierre W. Ferrez, Anna Buttfield.
Non Invasive Brain-Machine Interfaces [A]. IDIAP
Research Institute. Available URL:
https://www.esa.int/gsp/ACT/doc/ARI/ARI%20
Study%20Report/ACT-RPT-BIO-ARI-056402-Non_
invasive_brain-machine_interfaces_-_Martigny_IDI-
AP.pdf
[29] Maksimenko V. A., Kurkin S. A., Pitsik E. N.. Artifi-
cial Neural Network Classification of Motor-Related
EEG: An Increase in Classification Accuracy by Re-
ducing Signal Complexity[A]. Hindawi Complexity,
2018, Article ID 9385947. Available URL:
https://doi.org/10.1155/2018/9385947
[30] US Patent No 7,219,087B2, Soft computing optimiz-
er of intelligent control system structures[P]. (Inven-
tor: S. V. Ulyanov), Date of patent: May 15, 2007
[WO 2005/013019 A3, 2005].
[31] US Patent No 2006, 0218 A1, System for soft com-
puting simulation[P]. (Inventor: S. V. Ulyanov), Date
of patent: Sept. 2006, 2006.
[32] Nikolaeva A. V., Ulyanov S.V. Intelligent robust con-
trol of an autonomous robot-manipulator[A]. Soft-
ware systems and computational methods. 2014(1):
34-62.
DOI: 10.7256/2305-6061.2014.1.11466
Available URL: https://en.nbpublish.com/library_
get_pdf.php?id=28019
[33] Tarasov V. B. From multi-agent systems to intelli-
gent organizations: philosophy, psychology, com-
puter science[J]. M.: URSS editorial, 2002: 352. (in
Russ).
[34] Zaitsev A.A., Kureichik V.V., Polupanov A.A. Over-
view of evolutionary optimization techniques based
on swarm intelligence[J]. News SFU. Technical sci-
ence, 2010(12). (in Russian).
DOI: https://doi.org/10.30564/aia.v2i2.1567
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
[35] Kureichik V.M., Kazharov A.A.. The use of swarm
intelligence in solving difficult problems[J]. News
SFU. Technicalscience, 2011(7). (in Russ).
[36] Sandberg H. et al. Maximum work extraction and
implementation costs for nonequilibrium Maxwell’s
demon[J]. Physical Review E. 2014, 4: 042119.
[37] Chatzis S. P., Korkinof D., Demiris Y.. A quan-
tum-statistical approach toward robot learning by
demonstration[J]. IEEE Transactions on Robotics,
2012, 28(6): 1371-1381.
[38] Sagawa T, Ueda M. Minimal Energy Cost for Ther-
modynamic Information Processing: Measurement
and Information Erasure[J]. Phys. Rev. Lett., 2009,
102(25): 250602. Erratum. Phys. Rev. Lett. 106,
189901, 2011.
[39] Horowitz J. M., Sandberg H. Second-law-like
inequalities with information and their interpreta-
tions[J]. New Journal of Physics, 2014, 16: 125007.
[40] Ulyanov S. V., Reshetnikov A. G., Mamaeva A. A.,
etc.. Hybrid cognitive control systems on the exam-
ple of driving[J]. Journal of Systems Analysis in Sci-
ence and Education, 2010, 3. (in Russia)
Available URL: http://sanse.ru/download/261
[41] Ulyanov S. V., Reshetnikov A. G. Basis of cognitive
computer training in robotics. Intellectual simulator
for the formation of active knowledge[J]. Journal of
Systems Analysis in Science and Education, 2016(4).
(in Russia)
Available URL: http://sanse.ru/download/277
[42] Gael Langevin, InMoov is the first Open Source 3D
printed life-size robot[R]. Available URL: http://in-
moov.fr
[43] Hand robot InMoov, opensource - project[R]. Avail-
able URL:
https://www.thingiverse.com/thing:17773
[44] EmotivEpoc, Neurocomputer interface[R]. Available
URL: https://www.emotiv.com
[45] Arduino, electronic device design tool[R]. Available
URL: http://arduino.ru/
[46] Petrov B.N., Ulanov G.M., Ulyanov S.V. etc. The
theory of models in control processes: information
and thermodynamic aspects[J]. M.: Nauka, 1978:
Distributed under creative commons license 4.0
224. (in Russia).
[47] Sotnokov P., I., Overview of EEG Signal Processing
Techniques at Brain Computer Interfaces[J]. Engi-
neering Bulletin, Electronic Journal, 2014(10). (in
Russia).
[48] Ulyanov S.V., etc. Intelligent self-organizing cogni-
tive regulators. Part 2: models of cognitive interfaces
“brain - device”[J]. Journal of Systems Analysis in
Science and Education, 2015, 1. (in Russia).
Available URL: http://sanse.ru/download/236
[49] Ulyanov S.V.. Intelligent self-organized robust con-
trol design based on quantum / soft computing tech-
nologies and Kansei engineering[J]. Computer Sci-
ence Journal of Moldova. 2013, 21, 2(62): 242-279.
[50] Ulyanov S.V., Yamafuji K.. Intelligent self-orga-
nized cognitive controllers. Pt. 1: Kansei / affective
engineering and quantum / soft computing technol-
ogies[J]. System Analysis in Science and Education,
2014, 4. Available URL: http:/www.sanse.ru/ar-
chive/48
[51] Ulyanov S.V.. Quantum fast algorithm computational
intelligence Pt I: SW / HW smart toolkit. Artificial
Intelligence Advances, 2019, 1(1): 18-43. Available
URL:
https://doi.org/10.30564/aia.v1i1.619
[52] Ulyanov S.V., Reshetnikov A.G., Tyatyushkina O.Yu.
Intelligent robotics, part 2: socio-economic-techni-
cal platform of the cognitive educational process[J].
Journal of Systems Analysis in Science and Educa-
tion, 2019. No 4. (in Russia) Available URL:
http://sanse.ru/download/277
[53] D. Shahmirzadi. Computational Modeling of the
Brain Limbic System and its Application In Control
Engineering[D]. Texas A&M University, U.S.A.,
2005.
[54] Reza Keramat, Mohammad Hosein Ershadi,
Shahrokh Shojaeian. A Comparison of Fuzzy and
Brain Emotional Learning-Based Intelligent Control
Approaches for a Full Bridge DC-DC Converter[J].
International Journal of Industrial Electronics, Con-
trol and Optimization, 2019, 2(3): 197-206.
DOI: https://doi.org/10.30564/aia.v2i2.1567 87
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020

Artificial Intelligence Advances

https://ojs.bilpublishing.com/index.php/aia

REVIEW
Research and Analysis of Machine Learning Algorithm in Artificial In-
telligence
Yang Li1　Xueyuan Du1*　Yiheng Xu2
1. University of Liverpool, Liverpool, L69 3BX, England
2. Xi’an Jiaotong-Liverpool University, Suzhou, Jiangsu, 215213, China

ARTICLE INFO ABSTRACT

Article history This article firstly explains the concepts of artificial intelligence and
Received: 14 April 2020 algorithm separately, then determines the research status of artificial in-
telligence and machine learning in the background of the increasing pop-
Accepted: 20 April 2020 ularity of artificial intelligence, and finally briefly describes the machine
Published Online:15 July 2020 learning algorithm in the field of artificial intelligence, as well as puts for-
ward appropriate development prospects, in order to provide theoretical
Keywords: reference for industry insider.
artificial intelligence
machine learning algorithm
application of science and technology

　
1. Introduction 2. An Overview of Artificial Intelligence and

W
ith the continuous development of science and
technology, artificial intelligence technology
has been widely used in various fields with
great effects. As the core technology of artificial intelli-
gence, the initial research goal of machine learning is to
enable computer systems to have the ability to “learn” just
like humans, and to lay the foundation for the realization
of artificial intelligence. With the unremitting efforts of
growing experts and scholars, the development of artifi-
cial intelligence technology can be described as progress
with each passing day. In some complicated but ordinary
fields, machines can gradually replace manual work and
exert extremely efficient work.
Machine Learning algorithm
2.1 Concept of Artificial Intelligence
In the year of 1950, Allen Matheson Turing, the father
of modern computer science, named a test with his own
name. The test will randomly ask questions to the testees
using some devices when the tester is separated from the
testees. If more than 30% of all testees cannot determine
whether the tester is a person or a machine, then it shows
that the machine is intelligent. This is exactly the famous
Turing test, which also has its shortcomings: it takes hu-
mans--which have much uncertainty because of influence
of subjective consciousness--as the actual test standard of
the entire test, so the test results of the Turing test will be

*Corresponding Author:
Sergey V. Ulyanov,
INESYS LLC (EFKO GROUP), Russia;
Email: ulyanovsv@mail.ru

88 Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/aia.v2i2.1801

 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
lack of some imprecision [1]
With the continuous development of information tech-
nology, the definition of artificial intelligence also devel-
ops continuously. Professor Nielsen of the United States
believes: “Artificial intelligence is a discipline about
knowledge--how to express, obtain and use knowledge.”;
Professor Winston, who is also from the United States,
believes: “ Artificial intelligence is to study how to enable
computers to do intelligent work that could only be done
by humans in the past.” These two definitions of artificial
intelligence can show basic ideas and contents of it to a
certain extent: to study the characteristics of human in-
telligence activities and construct artificial systems with
certain intelligence, in other words, a basic theory, method
and technology to study how to use computer technology
to simulate human intelligent behavior[2].
2.2 Concept of Algorithm
The academics have not yet reached a unification on the
concept of the algorithm, among which philosophers
mostly study algorithms just in terms of the learning algo-
rithm. These philosophers believe that the algorithm has
a strong correlation with human cognition, but they rarely
study algorithm [3]. To summarize the research results of
today’s academics, there are following four views on the
definition of the algorithm:
① Algorithm is the method to determine a certain pro-
cess route.
② Take the algorithm as a program. After determining
a problem, the program will tell us to perform steps con-
tinuously, and analyze the problem, finally get the corre-
sponding result.
③ The third view is similar to the second one, both of
which take the algorithm as a program. There are some
differences in details: for the third view, the program can
always tell us how to deal with a series of rules systemati-
cally.
④ Algorithms are limited rules that present a series of
operations to solve specific types of problems. And algo-
rithm has the characteristics of finiteness, certainty, input,
output and effectiveness [3].
2.3 Machine learning algorithm
As the core technology of artificial intelligence, the initial
research goal of a machine learning algorithm is to enable
computer systems to have the ability to “learn” just like
humans and to complete tasks that humans do now. To
achieve this goal, learning will be an indispensable part.
Without learning, artificial intelligence cannot complete
tasks that humans do now, neither can it keep up with the
Distributed under creative commons license 4.0
pace of human development. In other words, learning is
the basis for the continuous development of artificial in-
telligence [4]
Compared with traditional algorithm, learning algo-
rithm emphasizes the attribute of “learning”, which means
the algorithm can develop itself, find new laws, and even
improve the logic and behavior of algorithm program
through “learning”. In today’s academics, a large num-
ber of scholars believe that a computer system cannot be
called an intelligent system if it is not able to learn.
Nowadays, the machine learning algorithm is usually
divided into three types: supervised learning, unsuper-
vised learning, and reinforcement learning. Supervised
learning means that learning algorithm works out predic-
tion models from the labeled training data, so as to effec-
tively predict unknown situations in the future; Unsuper-
vised learning means that learning algorithm performs
the exploration of data structures and extracts meaningful
information or results [5] without the guidance of known
variables or reward functions; Reinforcement learning is
mainly used as a trial and error method, which is able to
help learning algorithm improve prediction effect of pre-
diction models through interaction with the environment.
Compared with traditional algorithm, computer systems
can accomplish different things by using the same learn-
ing algorithm, and the operation process is simpler and
more convenient, even a single learning algorithm can
derive countless different programs by itself. To achieve
the same effect, the traditional algorithm requires more
process.
3. Research status of artificial intelligence and
machine learning algorithm
3.1 Research status of artificial intelligence
Since its birth in the 1950s, the annual investment of sci-
entific research in the field of artificial intelligence has
also increased year by year with strong support from glob-
al governments. As a first-class power in the century and
a major R & D country, many public funds in the United
States have entered into the field of artificial intelligence
under the guidance of government departments, and arti-
ficial intelligence has gained rapid development. In order
to further speed up the R & D of artificial intelligence, the
United States released the report of “2019 R & D Strategic
Planning of National Artificial Intelligence “in June 2019.
The report not only effectively updates the first edition of
“ R & D Strategic Planning of National Artificial Intelli-
gence” issued by the US government in 2016, but also
proposes research goals for problems like human-machine
collaboration methods, artificial intelligence safety, ethics,
DOI: https://doi.org/10.30564/aia.v2i2.1801 89
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
legal and social impacts, which lays a planning foundation
for the development of country artificial intelligence.
At present, with the support of advanced technologies
such as information technology and computer technology,
two key technical breakthroughs have emerged in the field
of artificial intelligence, namely cloud robot technology and
human brain bionic computing technology. As cloud robot
technology related aspects have gradually matured, there
are much more relevant existing researches of cloud robot
technology than that of human brain bionic computing
technology. With the rapid development and popularization
of high-tech technologies such as science and technology,
computer networks, cloud computing, and big data, the
costs of R & D and production of robot technology have
decreased year by year, also it is possible for robots to do
cloud control and cloud computing as well as obtain and
screen network data through big data technology. At this
stage, the main research aspects of cloud robot technology
are building a networked robot system platform, an open
system robot architecture, robot network platform algo-
rithm, image processing system development, and cloud
robot related network infrastructure construction.[6]
In China, Baidu is the company with the deepest re-
search in the field of artificial intelligence. As an Internet
company, Baidu’s most significant research achievement
in recent years is the Deep Speech 2 speech recognition
system, which has not only been recognized by the world
but also successfully been chosen by MIT Technology
Review as one of 2016 Top Ten Progressive Technolo-
gies. As an Internet company, Alibaba has developed an
intelligent customer service system based on the actual
needs of its own e-commerce platform, which effectively
deals with current restriction problem on the sharp rise in
customer service demand for e-commerce platforms, and
promotes basic customer service tasks such as automatic
voice transfer, emotion recognition and keyword recog-
nition to be replaced by intelligent customer service. This
intelligent customer service system not only effectively
reduces the workload of manual customers, but also re-
duces the company’s labor demand costs, generates more
economic benefits, and lays the foundation for further
development. Tencent has also achieved certain achieve-
ments in the field of artificial intelligence, among which
the most advanced artificial intelligence technology is
the face recognition technology developed by “Tencent
Youtu”. This technology is at the leading level even in the
whole world, it also provides strong image and pattern
recognition technology support for Tencent subordinate
industries such as QQ, Tenpay, Webank, etc. In addition to
the above three major companies, other companies such
as iFLYTEK and AISpeech have also achieved great R &
90 Distributed under creative commons license 4.0
D achievements in the field of artificial intelligence, and
applied these achievements to industrial development. [7]
4. Classic learning algorithm in artificial in-
telligence
4.1 regression algorithm
The so-called regression algorithm is an algorithm that
attempts to explore the relationship among variables by
measuring the error. In the field of artificial intelligence,
it means a regression algorithm when people say “regres-
sion”. The common regression algorithm includes the
least square method, logistic regression, stepwise regres-
sion, multivariate adaptive regression splines, local scatter
smoothing estimation, etc.
4.2 Instance-based Algorithm
An instance-based algorithm is often used to build game
problem models. In the actual application process, after
the model is established, it is often necessary to select a
batch of sample data, then the computer system will ana-
lyze and compare new data with sample data according to
certain similarity characteristics, and find the best match-
ing data from new data. Therefore, the instance-based
algorithm is also called memory-based learning. Common
instance-based algorithms include KNN, learning vector
quantization, self-organizing mapping algorithms, etc.
4.3 Regularization algorithm
Generally speaking, the regularization algorithm can be
regarded as an extension of the regression algorithm. This
algorithm will adjust algorithm reasonably according to
the complexity of it. In the adjustment process, the regu-
larization algorithm usually rewards simple models, and
punishes complex algorithm correspondingly. Common
regularization algorithm includes Ridge Regression-LAS-
SO-and resilient network.
4.4 Decision-Tree Learning Algorithm
In the actual application process, the decision-tree al-
gorithm will reward the tree-structure decision model
according to data attributes. This decision model is often
used to solve classification and regression problems. The
common decision-tree algorithm includes regression trees,
ID3, C4.5, CHAID, Decision Stump, random forest, mul-
tivariate adaptive regression splines and gradient boosting
machines.
4.5 Bayesian method
Bayesian method algorithm is a kind of algorithm de-
DOI: https://doi.org/10.30564/aia.v2i2.1801
 Artificial Intelligence Advances | Volume 02 | Issue 02 | October 2020
veloped on the basis of Bayesian positioning, which is
often used to solve classification and regression problems.
Common Bayesian method algorithm includes the Bayes-
ian naive algorithm, averaged one-dependence estimator
and BBN.
4.6 Kernel-based algorithm
Among all kernel-based algorithms, the support vector
machine algorithm is the most famous. In practical appli-
cations, the kernel-based algorithm will map input data to
a higher-order vector space, in which data is processed. In
high-order vector space, some classification or regression
problems can be better solved and data resolution can be
improved. The common kernel-based algorithm includes
support vector machines, radial basis functions, and linear
discriminant analysis.
4.7 Clustering algorithm
In application, the clustering algorithm will classify and
summarize all input data in a layered manner according to
the central point process, then find out internal structures
of data and determine the biggest common point, based
on which input data will be classified. The common clus-
tering algorithm includes the k-Means algorithm and the
expectation maximization algorithm.
4.8 Association rules algorithm
Computer systems can choose rules that can explain the
relationship among variables through association rules
learning, then find out relevant ones from a large amount
of data. Common association rules algorithm includes the
Apriori algorithm and the Eclat algorithm.
4.8 Artificial Neural Network
In application, an artificial neural network algorithm will
simulate biological neural networks and solve classifica-
tion and regression problems. Nowadays, as the focus of
academic research, an artificial neural network algorithm
has covered hundreds of different types of algorithms,
among which deep learning algorithm is the most com-
monly used one. Common used artificial neural network
algorithm includes perceptron neural network, reverse
transfer, Hopfield network, self-organizing mapping and
learning vector quantization. As an important algorithm
category derived from the development process of artifi-
cial neural network algorithms, deep learning algorithm
has now received extensive attention from home and
abroad and experienced in-depth research and develop-
ment. However, most of today’s deep learning algorithms
Distributed under creative commons license 4.0
are semi-supervised learning algorithms, which are mostly
used to process large data sets with a small amount of un-
identified data. [8]
5. Conclusion
With the rapid development of computer technology and
information technology, artificial intelligence technol-
ogy has also been deeply researched and popularized.
Although existing artificial intelligence technology has
learning ability to some extent, it is difficult to meet the
ever-changing social demand due to limitations of theory,
algorithm and technology. Therefore, it is necessary for
academics from home and abroad to further study and
develop artificial intelligence, to promote the further im-
provement of machine learning algorithms in the field of
artificial intelligence, to ensure the effective growth of
the learning ability of artificial intelligence technology, to
meet social demand, and to contribute more to social de-
velopment.
References
[1] Xie Juanying, Liu Ran. Research progress of target
detection algorithm based on deep learning[J]. Jour-
nal of Shaanxi Normal University (Natural Science
Edition), 2019(5):1-9.
[2] Sun Shaojing, Chen Changfeng, Li Shigang. Devel-
opment and Challenge of ‘Algorithm Recommen-
dation and Artificial Intelligence’[J]. Journalism
Bimonthly, 2019(6):40-47.
[3] Tian Qichuan, Wang Manli. Research progress of
deep learning algorithms[J]. Computer Engineering
and Applications, 2019(22):25-33.
[4] Chang Yingchun. Anomaly detection of game artifi-
cial intelligence based on layered time series memory
algorithm[D]. Journal of Hebei University of tech-
nology, 2015.
[5] Gan Jingzhong, Yang Xiulan, Lv Jie. Overview of
Unsupervised Learning Algorithms in Artificial In-
telligence[J]. Technology and Industry Across the
Straits, 2019(1):134-135.
[6] Klaus Mainzell. Artificial Intelligence and Machine
Learning: Algorithm Basics and Philosophical Per-
spectives[J]. Journal of Shanghai normal university
(Philosophy and Social Sciences), 2018(3):13-24.
[7] Zhou Jin. Deep learning algorithms on artificial intel-
ligence[J]. Computer Programming Skills & Mainte-
nance, 2019, (7):115-117.
[8] Huang Fayang. Face Recognition Technology Based
on Neural Network Deep Learning Algorithm[J]. In-
telligent building, 2018(10):65-67.
DOI: https://doi.org/10.30564/aia.v2i2.1801 91
 Author Guidelines
This document provides some guidelines to authors for submission in order to work towards a seamless submission
process. While complete adherence to the following guidelines is not enforced, authors should note that following
through with the guidelines will be helpful in expediting the copyediting and proofreading processes, and allow for
improved readability during the review process.

Ⅰ. Format

● Program: Microsoft Word (preferred)

● Font: Times New Roman
● Size: 12
● Style: Normal
● Paragraph: Justified
● Required Documents

Ⅱ. Cover Letter

All articles should include a cover letter as a separate document.

The cover letter should include:
● Names and affiliation of author(s)
The corresponding author should be identified.
Eg. Department, University, Province/City/State, Postal Code, Country
● A brief description of the novelty and importance of the findings detailed in the paper
Declaration
v Conflict of Interest
Examples of conflicts of interest include (but are not limited to):
● Research grants
● Honoria
● Employment or consultation
● Project sponsors
● Author’s position on advisory boards or board of directors/management relationships
● Multiple affiliation
● Other financial relationships/support
● Informed Consent
This section confirms that written consent was obtained from all participants prior to the study.
● Ethical Approval
Eg. The paper received the ethical approval of XXX Ethics Committee.
● Trial Registration
Eg. Name of Trial Registry: Trial Registration Number
● Contributorship
The role(s) that each author undertook should be reflected in this section. This section affirms that each credited author
has had a significant contribution to the article.
1. Main Manuscript
2. Reference List
3. Supplementary Data/Information
Supplementary figures, small tables, text etc.
As supplementary data/information is not copyedited/proofread, kindly ensure that the section is free from errors, and is
presented clearly.

Ⅲ. Abstract

A general introduction to the research topic of the paper should be provided, along with a brief summary of its main
results and implications. Kindly ensure the abstract is self-contained and remains readable to a wider audience. The
abstract should also be kept to a maximum of 200 words.
Authors should also include 5-8 keywords after the abstract, separated by a semi-colon, avoiding the words already used
in the title of the article.
Abstract and keywords should be reflected as font size 14.

Ⅳ. Title

The title should not exceed 50 words. Authors are encouraged to keep their titles succinct and relevant.
Titles should be reflected as font size 26, and in bold type.

Ⅳ. Section Headings

Section headings, sub-headings, and sub-subheadings should be differentiated by font size.

Section Headings: Font size 22, bold type
Sub-Headings: Font size 16, bold type
Sub-Subheadings: Font size 14, bold type
Main Manuscript Outline

Ⅴ. Introduction

The introduction should highlight the significance of the research conducted, in particular, in relation to current state of
research in the field. A clear research objective should be conveyed within a single sentence.

Ⅵ. Methodology/Methods

In this section, the methods used to obtain the results in the paper should be clearly elucidated. This allows readers to be
able to replicate the study in the future. Authors should ensure that any references made to other research or experiments
should be clearly cited.

Ⅶ. Results

In this section, the results of experiments conducted should be detailed. The results should not be discussed at length in
this section. Alternatively, Results and Discussion can also be combined to a single section.

Ⅷ. Discussion

In this section, the results of the experiments conducted can be discussed in detail. Authors should discuss the direct and
indirect implications of their findings, and also discuss if the results obtain reflect the current state of research in the field.
Applications for the research should be discussed in this section. Suggestions for future research can also be discussed in
this section.

Ⅸ. Conclusion

This section offers closure for the paper. An effective conclusion will need to sum up the principal findings of the papers,
and its implications for further research.

Ⅹ. References

References should be included as a separate page from the main manuscript. For parts of the manuscript that have
referenced a particular source, a superscript (ie. [x]) should be included next to the referenced text.
[x] refers to the allocated number of the source under the Reference List (eg. [1], [2], [3])
In the References section, the corresponding source should be referenced as:
[x] Author(s). Article Title [Publication Type]. Journal Name, Vol. No., Issue No.: Page numbers. (DOI number)

Ⅺ. Glossary of Publication Type

J = Journal/Magazine
M = Monograph/Book
C = (Article) Collection
D = Dissertation/Thesis
P = Patent
S = Standards
N = Newspapers
R = Reports
Kindly note that the order of appearance of the referenced source should follow its order of appearance in the main manu-
script.
Graphs, Figures, Tables, and Equations
Graphs, figures and tables should be labelled closely below it and aligned to the center. Each data presentation type
should be labelled as Graph, Figure, or Table, and its sequence should be in running order, separate from each other.
Equations should be aligned to the left, and numbered with in running order with its number in parenthesis (aligned
right).

Ⅻ. Others

Conflicts of interest, acknowledgements, and publication ethics should also be declared in the final version of the manu-
script. Instructions have been provided as its counterpart under Cover Letter.
About the Publisher
Bilingual Publishing Co. (BPC) is an international publisher of online, open access and scholarly peer-reviewed
journals covering a wide range of academic disciplines including science, technology, medicine, engineering, education
and social science. Reflecting the latest research from a broad sweep of subjects, our content is accessible world-
wide——both in print and online.

BPC aims to provide an analytics as well as platform for information exchange and discussion that help organizations
and professionals in advancing society for the betterment of mankind. BPC hopes to be indexed by well-known
databases in order to expand its reach to the science community, and eventually grow to be a reputable publisher recognized
by scholars and researchers around the world.

BPC adopts the Open Journal Systems, see on ojs.bilpublishing.com

Database Inclusion

Asia & Pacific Science China National Knowledge

Creative Commons
Citation Index Infrastructure

Google Scholar Crossref MyScienceWork