You are on page 1of 6

12th Conference on Learning Factories, CLF2022

A Teaching Factory on context-aware design of automation for


sustainable manufacturing processes
Panagiotis Stavropoulosa,*, Alexios Papacharalampopoulosa, Konstantinos
Kampourisb, Christos Papaioannoua, Theodoros Bekiarisa, Nikos Porevopoulosa1
a
Laboratory for Manufacturing Systems and Automation (LMS), Department of Mechanical Engineering and Aeronautics,
University of Patras, Rio Patras 26504, Greece

b
Gizelis Robotics, Schimatari Viotias 32009, Kormatzini Area, Greece

Abstract

Automation has been the one of the key concepts in the factories of the future. However, its design ought to include criteria
beyond the standard ones, related to digital manufacturing, addressing also social, personal and business sustainability aspects.
The current work targets at creating a Teaching Factory that will be able to fuse all these aspects. To this end, and under the
concept of designing robotized manufacturing processes, a Teaching Factory is considered that targeted the design of a robotic
gripper for robo-bend cells. The technical requirements included universality across potential part geometries and the social
goal has been to reduce cells operators’ physical and cognitive load. Additionally, the commercialization of the gripper was
regarded. The procedure of designing and operating the TF is described herein and a roadmap for evaluation is extracted
towards designing TF evaluation procedures.

© 2022 The Authors. This is an open access article.


Peer Review statement: Peer-review under responsibility of the scientific committee of the 12th Conference on
Learning Factories 2022.
Keywords: Teaching Factory; design; sustainability; automation of manufacturing process; context-aware design

1. Introduction

The Teaching Factory (TF) [1,2] is a mostly suitable training method for the era of Industry 4.0 and on [3], as
experiential learning is deployed on the basis of problem solving to achieve upskilling and technology transfer
simultaneously. Digitalization seems to play a two-fold role in this procedure; firstly, the communication and
learning procedure throughout the TF are facilitated and secondly, the corresponding key enabling technologies
are utilized in the procedure of innovation absorption. To facilitate this procedure, the concept of TF Network [4,5]
has also been suggested, since academia and/or industries from different or even similar ecosystems can be utilized
to boost the performance of the TF procedure. A similar concept, that of learning factories (LFs), is also based on
problem learning factories [6], and it could potentially be used as a first step towards achieving the familiarity of
the students with technologies through the use of testbed. Additionally, since design is closely related to
manufacturing, TF’s have been deployed to address this concept [7], as well as LFs [8], rendering the training
procedure far more interesting. After all, the future outlook of the overall procedure is expected to be positive
towards upskilling (students and employees) and reskilling (employees) [9].

The whole procedure of training, however, has been challenged during the global COVID-19 pandemic [10];
extra procedures are required in order to modify TF workflows to this end. In any case, experiential learning and
going through Kolb’s learning cycle have to be targeted, rendering upskilling even more challenging since d-
learning (or e-learning, more specifically) is expected to be applied [11]. An additional challenge is to keep up
with the social demands; since “Industry 5.0 centers around three interconnected core values: human-centricity,

* Corresponding author. Tel.: +30-2610-910160


E-mail address: pstavr@lms.mech.upatras.gr

Electronic copy available at: https://ssrn.com/abstract=4072720


Stavropoulos Panagiotis et al, CLF 2022

sustainability and resilience” [12]. This calls for an additional effort towards balancing automation and human
centered systems (i.e. through specific technologies, such as collaborative robotics [13]).

Furthermore, it is worth mentioning that there are several special strategies that have to be taken into
consideration, as per the literature [14]. Indicatively, the following are mentioned:

 lifelong learning and transdisciplinary education,


 sustainability, resilience, and human-centric design modules,
 hands-on data fluency and management courses and
 human-agent/machine/robot/computer interaction experiences

All these could be supported by the so-called context awareness in manufacturing, which aims at the increase
of the “ubiquitous system capabilities”, as the ultimate goal is to “adapt the service they provide to the situation of
the user” [15]. As such, the evaluation procedure of the TF’s [5] is expected to be updated, in addition to the TF
procedure, towards considering all the aforementioned (technical, social, business) aspects. The current work can
be considered as a first step towards fusing context-aware design, business and social aspects, as well as sustainable
automation in manufacturing training through the concept of TF.

2. The TF procedure

The TF procedure is well-established in literature, as aforementioned, however, herein the workflow (Fig. 1) is
slightly modified. More specifically, three distinctive extra aspects are taken into consideration: the fuzziness in
the requirements (session 1), the integration of business feasibility in filtering the solutions (during session 2) and
the consideration of opinions in evaluation (week 3). These modifications target at problem solving &
communication, business & social aspects integration, and checking upskilling, respectively.

Fig. 1. The TF procedure.

3. Implementation

The case study that has been adopted is that of automating to a further extent a robo-bend cell (Fig. 2). This cell
consists of a robotic arm used to feed the machine with the part and an air-bending machine tool. The goal has
been to design a gripper that can manipulate metal sheets. The part families have been defined as being of various
geometries, potentially with holes of random positioning and size.

Electronic copy available at: https://ssrn.com/abstract=4072720


Stavropoulos Panagiotis et al, CLF 2022

Fig. 2. The ROBO-BEND cell.

The trainees’ (students) selection has been based quite random (Table 1), based on their availability and during
the first session, a first description of the problem was made by the company, and since the trainees had no previous
experience with robotized processes, some material was given to them, related to potential solutions that have been
used in various similar cases, sensors capabilities and connectivity challenges. Two representatives from industry,
as well as two representatives from academia were engaged.

Table 1. Characteristics of the participants.


Person Background Role
Student 1 MSc candidate Trainee
Student 2 Undergraduate Trainee
Student 3 Undergraduate Trainee
Student 4 Undergraduate Trainee
Student 5 Life-long trainee (Graduate) Trainee
Business representative 1 Engineering Director Evaluator
Business representative 2 Robotics Engineer Evaluator
Academia representative 1 Professor Academic quality guarantee
Academia representative 2 Researcher Day-to-day guidance

4. Results

The students were self-organized and took roles based on their experience and liking. Even though the designs
came up after internal discussions (without the company participating), this led to specializations and discrete
workflows for both week 1 and 2. Table 2 summarizes the outcome of this procedure.

Table 2. The roles of the students during the sessions.


An example of a column heading Session 2 Role (presenting alternative Session 3 Role (focusing on
solutions) solution #3)
Student 1 State of the Art/Market Connectivity to the robot &
synchronization
Student 2 Design/Presentation of solution #1 Mechanical connectivity details
(parallel bars)
Student 3 Design/Presentation of solution #2 Electric / pneumatic triggering
(vertical bars)
Student 4 Design/Presentation of solution #3 Finalization of the design and
(fully configurable geometry) working envelope estimation

Student 5 Draft cost model Sensors integration

Electronic copy available at: https://ssrn.com/abstract=4072720


Stavropoulos Panagiotis et al, CLF 2022

During the design of the solutions, the manufacturability and the assemble-ability of the designs was taken into
consideration, as well as the cost. Three solutions came up, based on different concepts; a linear configuration, a
2D configuration and a fully customizable configuration. The first two solutions, that were rejected due to various
technical and business reasons during the second session, are given below, in figures 3 and 4. These technical
aspects included the limitations in the part manipulation handling, the flexibility in motion and the agility in
triggering the suction pumps. Business aspects had to do with the easiness in adapting (and the respective cost)
these designs for handling special geometries. All of these factors affect eventually the resilience of the automated
manufacturing cell.

Fig. 3. Solution #1: parallel bars with suction pumps.

Fig. 4. Solution #2: vertical bars with suction pumps.

The third solution, that was eventually selected, had to do with a fully adjustable geometry that consisted of
plug-in sub-modules. This was selected on the ground of utmost agility and tailoring easiness. Furthermore, the
activation of the sensors was easier, based on the fact that the sub-modules would be pre-fabricated and there would
be channels for electronic and pneumatic triggering. The sustainability aspects were also discussed during the final
TF session, while an attempt for their quantifying was also made, making the objectives clear to both sides.

Fig. 5. Solution #3: configurable geometry with sub-modules.

Electronic copy available at: https://ssrn.com/abstract=4072720


Stavropoulos Panagiotis et al, CLF 2022

5. Evaluation

An evaluation procedure was performed at the end of the TF, namely during week #3. It has been of two-fold
character; firstly, to evaluate the procedure from the trainee’s side (Table 3), adopting the trainees’ confidence as
an upskilling predictor [16]. Secondly, to allow the business (industry) to give their own feedback about this
particular procedure (Table 4).

Table 3. Evaluation from trainee’s side.


Question (I believe that…) Mean Value Min Value Max Value Range
1. The company has been responsive 4.8 4.000 5.000 1
2. The specifications were given analytically 3.6 2.000 5.000 3
3. The problem has been interesting 5 5.000 5.000 0
4. The problem has been difficult 3.6 2.000 5.000 3
5. I am happy with the final solution 4.4 4.000 5.000 1
6. I am interested to continuing this 4.8 4.000 5.000 1
particular project
7. The Teaching Factory has been a pleasant 4.8 4.000 5.000 1
experience
8. I feel confident in doing this again 4.6 4.000 5.000 1
9. The number of sessions was satisfactory 4.6 4.000 5.000 1
10. Additional teaching material would be 4.4 3.000 5.000 2
useful
11. Additional hands-on experience would be 4.8 4.000 5.000 1
useful

Table 4. Evaluation from trainers’ side


An example of a column heading Mean Value Range
1. The students have been responsive 4 0
2. The intermediate solutions made sense in terms of feasibility 4.5 1
3. The final solution covers 70% of business & financial 3 0
requirements
4. The final solution made sense covers 90% of technical 2.5 1
requirements
5. The final solution saves design time by 30% 3 0
6. The final solution increases line flexibility by 10% 4 0
7. The students have had progress in concepts comprehension & 5 0
manipulation
8. The students have had substantial progress in skills development 4.5 1
(creative thinking & problem solution)
9. The students were able to communicate their ideas 4 0
10. The students were able to communicate their intentions 4 0
11. It was evident that the students cooperated to a desired extent 4.5 1

It is quite interesting to use the range metric as an indicator for alarms. Particularly, for the cases of Table 3,
the following cases that appeared to have slightly larger ranges can be used to extract useful results, namely the:

 Trainees’ Question 2 on specifications


 Trainees’ Question 4 on problem difficulty
 Trainees’ Question 10 on teaching material availability

All these are indicative of the procedure causing cognitive conflict [17] and driving the trainees through Kolb’s
learning cycle. In addition, in the cases where the range has been equal to 1, the mean values are quite clear
regarding the conclusions that can be extracted. Furthermore, it would be equally interesting to focus on the
business side of the evaluation, namely in the cases of Table 4 where the range is non-zero:

Electronic copy available at: https://ssrn.com/abstract=4072720


Stavropoulos Panagiotis et al, CLF 2022

 Trainers’ Question 2 on feasibility: the mean value is high, so the conclusion is safe
 Trainers’ Question 4 on covering technical requirements: it is clear that some further elaboration should be
made; perhaps an additional session would be required
 Trainers’ Question 8 on skills development: the conclusion is safe
 Trainers’ Question 11 on students cooperating: the conclusion is safe

6. Conclusions and future outlook

It can be safely concluded that context-aware design for manufacturing is a very interesting area for applying
teaching factories towards two-way training. Furthermore, the current thematic area was able to address
sustainability in the threefold way; personal aspects were introduced through automation reducing physical and
cognitive load, profit and social issues ware regarded in terms of automation cost and workflow changes. In
addition, the integration of business sustainability aspects (customization easiness in this case) seems to be highly
useful for innovation absorption and digitalization of the workflow. Furthermore, the evaluation procedure can be
a very useful tool towards checking the feasibility of the solution as well as upskilling itself. However, further
research is required to standardize all of the above aspects. In addition, the direct link of opinions with upskilling,
in particular, needs to be studied in detail, through more quizzes about knowledge retention and knowledge
application, at high levels of Bloom’s taxonomy, potentially through interpreting quizzes results at statistically
significant numbers of trainees.

Acknowledgements

This research was partially supported by EIT Manufacturing Project M-NEST-RIS and EU Project
AVANGARD.

References

[1] G. Chryssolouris, D. Mavrikios, A novel initiative for the Manufacturing Enterprises and Educational Institutions of the future: The
Teaching Factory, Working Paper (2005)
[2] G. Chryssolouris, D. Mavrikios, N. Papakostas, D. Mourtzis, Education in Manufacturing Technology & Science: A view on Future
Challenges & Goals, Inaugural Keynote, in proceedings of International Conference on Manufacturing Science and Technology Melaka,
Malaysia (2006)
[3] D. Mourtzis, N. Milas, S. Fotia, A. Vlachou, Development of Skills and Competences in Manufacturing towards Education 4.0: A Teaching
Factory approach, AMP (2018) 194-210.
[4] D. Mavrikios, K. Georgoulias, G. Chryssolouris, The Teaching Factory Network: A new collaborative paradigm for manufacturing
education, Procedia Manufacturing, 31 (2019) 398-403.
[5] H. Bikas, P. E. Johansson, R. Di Falco, J. Stavridis, E. Niemi, Z. Azpilgain, L. Fumagalli, B. Thiede, P. Stavropoulos, A Teaching Factory
Knowledge Exchange Network. Available at SSRN 3859260 (2021).
[6] A. L. Andersen, T. D. Brunoe, K. Nielsen, Engineering education in changeable and reconfigurable manufacturing: Using problem-based
learning in a learning factory environment. Procedia CIRP 81 (2019) 7-12.
[7] P. Stavropoulos, H. Bikas, D. Mourtzis, Collaborative Machine Tool design: the Teaching Factory paradigm, Procedia CIRP 23 (2018)
123-128.
[8] B. Schallock, C. Rybski, R. Jochem, H. Kohl, Learning Factory for Industry 4.0 to provide future skills beyond technical training. Procedia
manufacturing, 23 (2018) 27-32.
[9] D. Mavrikios, K. Georgoulias, G. Chryssolouris, The teaching factory paradigm: Developments and outlook. Procedia Manufacturing 23
(2018) 1-6.
[10] D. Mourtzis, N. Panopoulos, J. Angelopoulos, S. Zygomalas, G. Dimitrakopoulos, P. Stavropoulos, A Hybrid Teaching Factory Model
for Supporting the Educational Process in COVID-19 era. Procedia CIRP 104 (2021) 1626-1631.
[11] T. H. Morris, Experiential learning–a systematic review and revision of Kolb’s model. Interactive Learning Environments 28(8) (2020)
1064-1077.
[12] X. Xu, Y. Lu, B. Vogel-Heuser, L. Wang, Industry 4.0 and Industry 5.0—Inception, conception and perception. Journal of Manufacturing
Systems 61 (2021) 530-535.
[13] M. Doyle-Kent, Collaborative Robotics in Industry 5.0 (Doctoral dissertation), TUWIEN (2021).
[14] J. Mitchell, D. Guile, Fusion Skills and Industry 5.0: Conceptions and Challenges, IntechOpen (2021).
[15] F. Belkadi, M. A. Dhuieb, J. V. Aguado, F. Laroche, A. Bernard, F. Chinesta, Intelligent assistant system as a context-aware decision-
making support for the workers of the future. Computers & Industrial Engineering 139 (2020) 105732.
[16] A. Gray, Upskilling through foundation skills. A literature review. A Report Prepared for the Department of Labour, Government of New
Zealand (2006).
[17] K. H. Um, J. Y. Oh, The mediating effects of cognitive conflict and affective conflict on the relationship between new product development
task uncertainty and performance. International Journal of Project Management 39(1) (2021) 85-95.

Electronic copy available at: https://ssrn.com/abstract=4072720

You might also like