You are on page 1of 8

25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.


Smooth transition between fuzzy regions to overcome failures in fuzzy membership
functions of decisions in collision avoidance of ocean navigation

L. P. Perera
Centre for Marine Technology and Engineering, Technical University of Lisbon,
Instituto Superior Tecnico, Lisbon, Portugal, prasad.perera@mar.ist,utl.pt

J. P. Carvalho
INESC-ID, Technical University of Lisbon,
Instituto Superior Tecnico, Lisbon, Portugal, joao.carvalho@inesc-id.pt

C. Guedes Soares
Centre for Marine Technology and Engineering, Technical University of Lisbon,
Instituto Superior Tecnico, Lisbon, Portugal, guedess@mar.ist.utl.pt

ABSTRACT. This paper proposes a method to overcome the failures on a fuzzy logic based Decision
Making (DM) process applied to collision avoidance in ocean navigation when contradictory decisions
result from the inference on boundaries of the represented fuzzy membership functions. The method
consists in the insertion of a smooth transition region. Further a decision making process of ocean
navigation is analyzed, input and output Fuzzy Membership Functions are derived, If-Then rule based
Fuzzy Inference System (FIS) are formulated and simulation results are presented regarding several
intercepted decision boundary regions.

KEYWORDS. Decision Making Process; Fuzzy logic; Decision failures; Ocean Navigation; Collision
Avoidance.

1. INTRODUCTION
Decision making process is an important part of the human life and several industrial applications have
been formulated to simulate the human decision making capabilities. Fuzzy-logic based systems, which
are formulated to approach the human type of thinking, facilitate a human friendly environment during
the decision making process. Hence several fuzzy-logic based decision making systems have been
developed in research as well as commercial applications recently (Hardy (1995)). The conjunction of
human behavior and decision making process have been illustrated by various fuzzy functions in the
study of Rommelfanger (1998) and Ozen et al. (2004).
Autonomous Guidance and Navigation (AGN) systems in ocean navigation are one of the industrial
applications of the human type of decision making process that has been considered in this study. The
functionalities and recent developments of AGN systems in ocean navigation are summarized by Fossen
(1999) and Ohtsu (1999) and ocean applications of AGN systems have been further studied theoretically
as well as experimentally by Moreira et al. (2007), Healey and Lienard (1993) and Do and Pan (2006).
However to improve autonomous navigational capabilities, AGN systems should be integrated with
intelligent decision making facilities as proposed in the recent literature. The decision making process
and strategies in interaction situations in ocean navigation, including collision avoidance situations, are
presented by Chauvin and Lardjane (2008). Benjamin and Curcio (2004) present the decision making
process of ocean navigation based on the interval programming model for multi-objective decision
making algorithms.
This paper focuses on finding a solution to a problem associated with the fuzzy-logic based Decision
Making process (Perera et al. (2009)) to be implemented on ocean navigation to improve safety of the
vessel by avoiding the collision situations. Fuzzy rules are formulated in accordance with the rules and
regulations that expressed in the Convention on the International Regulations for Preventing Collisions at
Sea (COLREGs) IMO (1972) to facilitate a regulated prevention of collision and to eliminate
navigational conflicts. However it observed that the Fuzzy-logic based decision making systems can
inherent failures in the intersection of the decision boundaries of two contradictory decisions which may
1
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.


lead to unavoidable collision situations. Therefore, to overcome decision failures, a method of smooth
transition between contradictory decisions of the fuzzy regions is proposed in this work.

2. FORMATION OF COLLISION SPACE


Figure 1 presents two vessels in a collision situation with the decisions in high collision risk situation.
The Own vessel, the vessel with the AGN system, is initially located at the point O ( xo, yo), and the
Target vessel, the vessel that must be avoided, is located at point A (xa, ya). The Own and Target vessels
speeds and course could be written as Vo, Va and o, a respectively. The speed Va, relative speed Va,o,
course a, and relative course a,o values of the Target vessel with respect to the Own vessel can be
estimated using the range and bearing values within a given time interval. All angles are measured with
respect to the positive Y-axis.

Figure 1. Relative Collision Situation with the Decisions in High Collision Risk Situation

The Own vessel ocean domain is divided into three circular sections with radius Rvd, Rb and Ra as
presented in Figure 1. The radius Ra represents the approximate range to the Target vessel and when the
Own vessel in a Give way situation, where the vessel has low priority for navigation and should take
appropriate actions to avoid collision situations. The radius Rb represents the approximate distance where
the Own vessel is in a Stand on situation, where the vessel has high priority for navigation but should
take appropriate actions to avoid collisions due to absence of the appropriate actions from the Target
vessel. The radius Rvd represents the vessel domain where the area is bounded for the dynamics of the
marine vessel. The distances of Rvd, Rb and Ra are formulated in the Range Fuzzy Membership Function
(FMF) (see Figure 2). The Own vessel collision regions were divided into ten regions, from I to X, as
presented in Figure 1. Even though eight regions (Perera et al. (2009)) are generally enough to implement
the COLREGs (IMO (1972)) rules and regulations of ocean navigation, ten regions have been proposed
to overcome the fuzzy decision failures as further discussed in the section 3.1. The Own vessel collision
regions are separated by dotted lines that are coincident with the Bearing FMF (see Figure 3). It is
assumed that the Target vessel is located within these ten regions and the collision avoidance decisions
are formulated in accordance to each region. As represented in Figure 1 the Target vessel position II has
been divided into eight divisions (from II a to II h) of vessel orientations, considering the relative

2
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.


course. These divisions are separated by dotted lines that are coincident with the Relative course
FMF(see Figure 5).

3. FORMATION OF DECISION SPACE


Three distinct situations that are involving risk of collision with respect to ocean navigation have been
recognized (Smeaton and Coenen (1990)): Overtaking, Head-on and Crossing. The decision space of
collision avoidance can be categorized into three stages for each vessel in open ocean environment.
When none of the vessels is at collision risk range, both vessels have the options to take appropriate
actions to avoid collision situation. However when both vessels are at collision risk range, the Give
way vessel should take appropriate actions to achieve safe passing distance in accordance with the
COLREGs rules and regulations, and the Stand on vessel should maintain course and speed. Further,
when both vessels are at critical collision risk range, and the Give way vessel does not take appropriate
actions to achieve safe passing distance in accordance with the COLREGs rules, then Stand on vessel
has the option to take appropriate actions to avoid the collision.
The collisions avoidance decisions of navigation with respect to the High risk collision situation with the
Range from Rvd to Ra for the respective regions are presented in the Figure 1. The decisions on region I
were formulated as course to starboard (o>0) and no speed change (Vo = 0); In region II, the proposed
decisions were course to starboard (o>0) and decrease speed (Vo<0). Therefore the transition of course
and speed changes from region I to region II is smoother. In region III, no course change (o=0) and
decrease speed (Vo<0); In region IV, course to port (o<0) and decrease speed (Vo<0); In region V ,
course to port (o<0) and no speed change (Vo=0). As noted, the decision transitions from region from I
to II, II to III, III to IV, IV to V are also smoother where no contradictory decisions are intersected.
The special regions VI and X are introduced into the decision space to make smooth transition between
fuzzy decisions. As presented in the Figure 1, the decisions proposed to region VI were no course change
(o = 0) and decrease speed (Vo<0). In region VII, the course to starboard (o>0) and decrease speed
(Vo< 0) are proposed. Hence the fuzzy decision transition between region V to VI is smoothed due to the
formulation of the new region. Further, in region VIII, is formulated with no course change (o=0) and
decrease speed (Vo<0) and in region IX is formulated course to port (o<0) and decrease speed (Vo<0),
therefore no contradictory decisions intersection has been observed . Finally in region X were formulated
the decisions of no course (o =0) and decrease speed (Vo< 0) facilitate the smooth decision transition
between region IX and I . As noted from the Figure 1, smoother decision transitions were achieved for
regions VII to VIII, VIII to IX, IX to X and X to I.
Tables 1 presents the summarized collision risk assessments, fuzzy rules and collisions avoidance
decisions with respect to the two vessel collision situation in Figure 1. The left-half of the Table 1
contains four main columns. The first column of the left-half Tables 1 represents the Collision Regions
(Re.) with respect to the Own vessel, and the second column represents the Divisions (Div.) that have
been divided into 8 regions from a to h of the Target vessel orientations and Collision Risk (Risk)
assessments with respect to each collision regions which have been divided into three sections of Low
Risk (Low), Medium Risk (Mid.) and High Risk (High). The Target vessel Relative Range from Rvd to Ra
is presented in the third column, and from Ra to Rb is presented in the forth column. The third and forth
columns are further divided into two sub-columns of speed conditions and decisions. The Relative Speed
Ratio of Va/Vo are presented in the first sub-column of the third and forth main columns. The speed
conditions of Va/Vo <, and > 1 are represented by approximately less than, equal, and greater than one
respectively. Finally the Decisions that need to be taken to avoid collision situations with respect to the
COLREGs rules and regulations are presented in the second sub-column of the third and forth main
columns. The decisions presented in the Table 1 could be summarized as: Course to starboard o>0,
course to port o<0 , no course change o=0, increase speed (Vo>0), decrease speed (Vo<0), no speed
change (Vo<0) and Not Applicable NA. Similar organization has been implemented for the right half of
the Table 1.

3
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.


Table 1: Collision risk assessments, Fuzzy rules and Collisions avoidance decisions
Re. Div./Risk Range (Rvd Ra) Range (Ra Rb) Re. Div./Risk Range (Rvd Ra) Range (Ra Rb)
Va Decision Va Decision Va Decision Va Decision
Vo Vo Vo Vo
I d / Mid. <1 NA <1 NA VI* a / Mid. <1 NA <1 NA
1 NA 1 NA 1 Vo<0 1 NA
>1 NA >1 NA >1 Vo<0 >1 NA
e / High <1 o>0 <1 o>0 b / High <1 NA <1 NA
1 o>0 1 o>0 1 Vo<0 1 NA
>1 o>0 >1 o>0 >1 Vo<0 >1 NA
f / Mid. <1 NA <1 NA c / Mid. <1 NA <1 NA
1 NA 1 NA 1 Vo>0 1 NA
>1 NA >1 NA >1 Vo>0 >1 NA
II e / Mid. <1 NA <1 NA VII a / Mid. <1 NA <1 NA
1 Vo>0 1 Vo>0 1 o>0 1 NA
>1 Vo>0 >1 Vo>0 >1 o>0 >1 NA
f / High <1 NA <1 NA b / High <1 NA <1 NA
1 o>0,Vo<0 1 o>0,Vo<0 1 o>0,Vo<0 1 NA
>1 o>0,Vo<0 >1 o>0,Vo<0 >1 o>0,Vo<0 >1 NA
g / Mid. <1 NA <1 NA c / Mid. <1 NA <1 NA
1 o>0 1 o>0 1 Vo>0 1 NA
>1 o>0 >1 o>0 >1 Vo>0 > NA
III f / Mid. <1 NA <1 NA VIII b / Mid. <1 NA <1 NA
1 Vo>0 1 Vo>0 1 Vo<0 1 NA
>1 Vo>0 >1 Vo>0 >1 Vo<0 >1 NA
g / High <1 NA <1 NA c / High <1 NA <1 NA
1 Vo<0 1 Vo<0 1 Vo<0 1 NA
>1 Vo<0 >1 Vo<0 >1 Vo<0 >1 NA
h / Mid. <1 NA <1 NA d / Mid. <1 NA <1 NA
1 Vo<0 1 Vo<0 1 Vo>0 1 NA
>1 Vo<0 >1 Vo<0 >1 Vo>0 >1 NA
IV g / Mid. <1 NA <1 NA IX c / Mid. <1 NA <1 NA
1 Vo>0 1 Vo>0 1 o<0 1 NA
>1 Vo>0 >1 Vo>0 >1 o<0 >1 NA
h / High <1 NA <1 NA d / High <1 NA <1 NA
1 o<0,Vo<0 1 o<0,Vo<0 1 o<0,Vo<0 1 NA
>1 o<0,Vo<0 >1 o<0,Vo<0 >1 o<0,Vo<0 >1 NA
a / Mid. <1 NA <1 NA e / Mid. <1 NA <1 NA
1 o<0,Vo<0 1 o<0,Vo<0 1 Vo>0 1 NA
>1 o<0,Vo<0 >1 o<0,Vo<0 >1 Vo>0 >1 NA
V h / Mid. <1 NA <1 NA X* c / Mid. <1 NA <1 NA
1 NA 1 NA 1 Vo<0 1 NA
>1 NA >1 NA >1 Vo<0 >1 NA
a / High <1 o<0 <1 NA d / High <1 NA <1 NA
1 o<0 1 NA 1 Vo<0 1 NA
>1 o<0 >1 NA >1 Vo<0 >1 NA
b / Mid. <1 NA <1 NA e / Mid. <1 NA <1 NA
1 NA 1 NA 1 Vo>0 1 NA
>1 NA >1 NA >1 Vo>0 >1 NA

4
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.

4. FUZZIFICATION AND DEFUZZIFICATION


The FMFs for inputs, Range (R), Bearing (o), Speed ratio (Va/Vo), and Relative course (a,o) are
presented in Figures 2 to 5, respectively. Figures 6 and 7 are formulated for the output FMFs of Speed
(Vo) and Course (o) change of the Own vessel. The Core and Supp variables are listed on the same
Figures of inputs and outputs FMFs. A Mamdani type IF <Antecedent> THEN <Consequent> rule (see
Table 1) based system has been developed and inference via Min-Max norm has been considered during
this study. Finally the defuzzification has been calculated by the center of gravity method.
The block diagram for Fuzzy Inference System (FIS) with integration of navigational instruments is
presented in Figure 8. The initial step of the fuzzy inference system consists of data collection of the
Target vessel position, speed and course. Then, the data is fuzzified with respect to the input FMFs of
Range (see Figure 2), Bearing (see Figure 3), Speed Ratio (see Figure 4) and Relative course (see Figure
5). The If-Then fuzzy rules (see Table 1) are developed in accordance with the COLREGs rules and
regulations and navigational knowledge. However near collision conditions in ocean navigation the
COLREGs does not provide clear rules and regulations. Therefore navigational knowledge (Expert
knowledge) on crash-stopping maneuvers has been considered for the formation of the fuzzy rules in
some regions. The outputs of the rule based system are the Collision Risk Warning and the Fuzzy
Decisions. Finally the fuzzy decisions are defuzzified by output FMFs of Course change (see Figure 6)
and Speed change (see Figure 7) to obtain the control actions that will be executed in the Own vessel
navigation.

5. COMPUTATIONAL IMPLEMENTATION
The fuzzy logic based Decision Making (DM) process has been implemented using the Mamdani based
(Sivanandam et al. (2007)) Fuzzy Inference System (FIS) in MATLAB software platform.

Figure 2. Range FMF Figure 3. Bearing FMF

Figure 4. Speed ratio FMF Figure 5. Relative course FMF

Figure 6. Course change FMF Figure 7. Speed change FMF

5
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.

Figure 8. Block diagram for Fuzzy Inference System

Considering the Range FMF (see Figure 2), the assigned distance values are Rvd 1NM, Rb 6NM and
Ra 10NM. The variables of 1 100, 2 800, 3 100, 4 800, 5 260 and 6 260 have been
considered for the Bearing FMF (see Figure 3). For the Speed Ratio FMF (see Figure 4) were assigned
the values 1 0.8, 2 1.2 and 3 5.0 . The Relative course FMF (see Figure 5) variables were assigned
as 1 50, 2 50, and 3 50. The output FMF of Course change (see Figure 6) was formulated by the
variables of 1 100, and 2 400. Finally the output FMF of Speed change (see Figure 7) was derived
with the variables 1 2 and 2 10.
Figures 9 and 10 present the simulations of relative trajectories of the Target vessel with respect to the
Own vessel in two vessels collision situations with the zoomed view around the origin. The vessel initial
speed condition is Vo/Va = 0.5 and initial course of the Own vessel is o = 00. The Own vessel is
initially located in the initial position of (0 NM,0 NM). The Target vessel startup points around the
around the collision point (0 NM, 5 NM) with the startup distance of 10 NM have been considered for
the analysis. Further constant speed and course towards the collision point from the initial point is
assumed. It is observed that the decision failures occur in-between relative bearing range of 1800 to 2700
and from 2700 to 3600 previously and when the decision failures occur the relative trajectory of the
Target vessel intercept the Own vessel initial position. The Target vessel relative trajectories with respect
to the relative bearing from 1800 to 2700 and from 2700 to 3600 are presented in Figures 9 and 10
respectively.
As presented in the figures, the relative trajectory of the Target vessel has not intercepted the initial
location of the Own vessel. It has instead shifted its trajectory from one side of the origin to the other
side smoothly as a result of the new fuzzy regions. The minimum distance between vessels are always
more then 0.05 NM even in critical collision situations where the Target vessel as a Give way vessel
didnt take appropriate actions to avoid collisions. Therefore the relative trajectories of the Target vessel
could be used to observe the behavior of the fuzzy rules in decision making process.

6
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.

Figure 9. Relative Trajectory of the Target vessel Figure 10. Relative Trajectory of the Target vessel
from course 1800 to 2700 with zoomed view. from course 2700 to 3600 with zoomed view.

6. CONCLUSION
This paper introduces further analysis of the decisions making process (Perera et al. (2009)) based on the
fuzzy logic, COLREGs rules and human expert knowledge in ocean navigation. The decision failures of
the fuzzy logic based systems with respect to the ocean navigation were further illustrated, and a proper
method to overcome decision failures was also presented in this study. As presented in Figures 9 and 10,
smooth transitions of change of course and/or speed could overcome the decision failures in collision
conditions as proposed. Further results on and the true trajectories of collisions avoidance could be found
on Perera et al. (2009). Although successful computational results were obtained by the proposed
method, it is assumed that more complex collision situations combination with multi vessel situations can
possibly occur and unexpected actions of the Target vessels could be experienced. Hence higher
capabilities should be formulated into the decision making process to overcome such situations.

7
25th Mini-EURO Conference on Uncertainty and Robustness in Planning and Decision Making (URPDM2010)

Coimbra, Portugal, 15 - 17 April, 2010.


ACKNOWLEDGEMENTS
The research work of the first author has been supported by a Doctoral Fellowship of the Portuguese
Foundation for Science and Technology under contract SFRH/BD/46270/2008. Further this work
contributes to the project of "Methodology for ships maneuverability tests with self-propelled models",
which is being funded by the Portuguese Foundation for Science and Technology under contract
PTDC/TRA/74332/2006

REFERENCES
Benjamin, M. R., Curcio, J. A., COLREGS - based navigation of autonomous marine vehicles.
IEEE/OES Autonomous Underwater Vehicles, 3239, 2004.
Chauvin, C., Lardjane, S.,. Decision making and strategies in an interaction situation: Collision
avoidance at sea, Transportation Research Part F (11), 259262, 2008.
Cockcroft, A. N., Lameijer, J. N. F., A Guide to The Collision Avoidance Rules. Elsevier Butterworth-
Heinemann, Burlington, MA. USA. 2001.
Do, K. D., Pan, J., Robust path-following of underactuated ships: Theory and experiments on a model
ship. Ocean Engineering 33 (3), 13541372, 2006.
Fossen, T. I. (Ed.), Recent developments in Ship Control Systems Design. World Superyacht Review.
Sterling Publication Limited, London, 1999.
Hardy, T. L., Multi-objective decision-making under uncertainty : Fuzzy logic method. NASA Technical
Memorandum 106796 - Computing in Aerospace 10 Meeting,1995.
Healey, A. J., Lienard, D., Multivariable sliding mode control for autonomous diving and steering of
unmanned underwater vehicle, IEEE Journal of Oceanic Engineering 18 (3), 327339, 1993.
IMO, Convention on the international regulations for preventing collisions at sea (COLREGs). URL :
http://www.imo.org/conventions/, 1972.
Moreira, L., Fossen, T. I., Guedes Soares, C., Path following control system for a tanker ship model.
Ocean Engineering, 34, 20742085, 2007.
Ohtsu, K., Recent development on analysis and control of ships motions. Proceedings of the 1999 IEEE
International Conference on Control Applications, 10961103, 1999.
Ozen, T., Garibaldi, J. M., Musikasuwan, S., Modeling the variation in human decision making. IEEE
Annual Meeting of the Fuzzy Information Processing - NAFIPS 04 2, 617622, 2004.
Perera, L. P., Carvalho, J. P., Guedes Soares, C., Decision making system for collision avoidance of
marine vessel navigation based on COLREGs rules and regulations, Proc. 13th Congress of International
Maritime Association of Mediterranean, Istanbul, Turkey, 1121-1128, 2009.
Perera L. P., Carvalho, J. P. and Guedes Soares, C., Autonomous guidance and navigation based on the
COLREGs rules and regulations of collision avoidance., Advanced Ship Design for Pollution Prevention,
C. Guedes Soares and J. Parunov (Eds.), Taylor & Francis Group, London, UK, 2010.
Rommelfanger, H. J., Multicriteria decision making using fuzzy logic. Conference of the North American
Fuzzy Information Processing Society, 360364,1998.
Sivanandam, S. N., Sumathi, S., Deepa, S. N., Introduction to Fuzzy Logic using MATLAB. Springer
Berlin Heidelberg, New York, 2007.

You might also like