autonomous guidance and navigation based on the colregs rules and regulations of collision avoidance

8/9/2019 Autonomous guidance and navigation based on the COLREGs rules and regulations of collision avoidance.

1/12

Advanced Ship Design for Pollution Prevention, C. Guedes Soares and J. Parunov (Eds.), Taylor & Francis Group, London, UK, 2010, ISBN 978-0-415-58477-7,

pp 205-216. (In Proc. Of the International Workshop Advanced Ship Design for Pollution Prevention, Split, Croatia, Nov. 23-24, 2009)

1 INTRODUCTION

1.1 Autonomous guidance and navigation

Automatic steering is a most valuable invention ifproperly used. It can lead to disaster when it is left tolook after itself while vigilance is relaxed. It is onmen that safety at sea depends and they cannot makea greater mistake than to suppose ...

The statement was made by the Justice Cairns withrespect to a collision that had occurred due to thefault in the automatic pilot system when the Britishcargo ship Trentbank was overtaking the Portu-guese tanker Fogo in the Mediterranean in Sep-tember 1964 (Cockcroft and Lameijer (2001)). Theverdict not just shows the importance of the AGNsystems in ocean navigations but also of the humanvigilance of its capabilities and requirements for fur-ther developments.

The automatic pilot systems are primary level de-velopments units of the Autonomous Guidance and

Navigation (AGN) Systems and their applicationshave been in the dreams of ship designers in severaldecades. The development of computer technology,satellite communication systems, and electronic de-vices, including high-tech sensors and actuators,have turned these dreams into a possible realitywhen designing the next generation ocean AGN sys-tems.

The initial step of the AGN system, which madethe foundation for applied control engineering, wasformulated around 1860 to 1930 with the inventionof the first automatic ship steering mechanism by

Sperry (1922). Sperrys work was based on replica-tion of the actions of an experienced helmsman thatwas formulated as a single input single output sys-tem (SISO).

Similarly, the research work done by Minorski(1922) is also regarded as the key contribution to thedevelopment of AGN Systems. His initial work wasbased on the theoretical analysis of automated shipsteering system with respect to a second order shipdynamic model. The experiments were carried outby Minorski on New Mexico in cooperation with theUS Navy in 1923 and reported in 1930 (Bennett,1984).

Hence both works done by Sperry and by Minor-ski are considered as the replication of non-linearbehavior of an experienced helmsman (Roberts et al.(2003)) in ocean navigation. From Sperrys time topresent, much research has been done and consider-able amount of literature has been published onAGN systems with respect to the areas of marinevessel dynamics, navigation path generations, and

controller applications, environmental disturbancerejections and collision avoidance conditions.

The functionalities of the Multipurpose Guidance,Navigation and Control (GNC) systems are summa-rized by Fossen (1999) on a paper that focuses notonly on course-keeping and course-changing maneu-vers (Conventional auto pilot system) but also inte-gration of digital data (Digital charts and weatherdata), dynamic position and automated docking sys-tems.

Autonomous guidance and navigation based on the COLREGs rules andregulations of collision avoidance.

L. P. PereraCentre for Marine Technology and Engineering (CENTEC), Technical University of Lisbon,

Instituto Superior Tecnico, Lisbon, Portugal

J. P. CarvalhoINESC-ID, Technical University of Lisbon, Instituto Superior Tecnico, Lisbon, Portugal

C. Guedes SoaresCentre for Marine Technology and Engineering (CENTEC), Technical University of Lisbon,

Instituto Superior Tecnico, Lisbon, Portugal

ABSTRACT: Autonomous Guidance and Navigation (AGN) is meant to be an important part of the futureocean navigation due to the associated navigational cost reduction and maritime safety. Furthermore intelli-gent decision making capabilities should be an integrated part of the future AGN system in order to improveautonomous ocean navigational facilities. This paper is focused on an overview of the AGN systems with re-spect to the collision avoidance in ocean navigation. In addition, a case study of a fuzzy logic based decisionmaking process accordance with the International Maritime Organization (IMO) Convention on the Interna-

tional Regulations for Preventing Collisions at Sea (COLREGs) has been illustrated.


2/12

Figure 1: Autonomous Guidance and Navigation System.

Recent developments of design, analysis and con-

trol of AGN systems are also summarized by Ohtsu(1999), and several ocean applications of AGN sys-tems have been further studied theoretically as wellas experimentally by (Moreira et al., (2007); Healeyand Lienard, (1993); Do and Pan, (2006)). This areais bound to become increasingly important in the fu-ture of ocean navigation due to the associated costreduction and maritime safety.

The block diagram of main functionalities of aAGN system integrated with collision avoidance fa-cilities is presented in Fig. 1. The AGN system con-tains four units of Collisions Avoidance Conditions(CAC), Control System(CS), Filters and Sensors.The sensor unit consists of sensors that could meas-ure course and speed characteristics of the vesselnavigation. The filter unit is formulated for filter-outnoisy signals. The output of the Filter unit will befeedback into the CAC unit as well as the CS unit.The sensor signals will be used by the CS unit tocontrol overall vessel navigation of the vessel andused by the CAC unit to make decisions on colli-sions avoidance process.

1.2 Collision avoidance

Having an intelligent decision making process is animportant part of the future AGN systems in oceannavigation. However, conventional ocean naviga-tional systems consist of human guidance, and as aresult, 75-96 % of marine accidents and causalitiesare caused by some types of human errors(Rothblum et al. (2002); Anto and Guedes Soares2008). Since most of the wrong judgments andwrong operations of humans at the ocean ended inhuman casualties and environmental disasters, limit-ing human subjective factors in ocean navigation andreplacing them by an intelligent Decision Making(DM) system for navigation and collision avoidance

could reduce maritime accidents and its respectivecausalities. However development of collisionavoidance capabilities into the next generation AGNsystems in ocean navigation is still in the hands offuture researchers and this part of the intelligentAGN systems has been characterized as e-Navigation (eNAV (2008)).

The terminology used in recent literature regard-ing the collision avoidance conditions designates thevessel with the AGN system as the Own vessel,and the vessel that needs to be avoided as the Tar-get vessel. With respect to the Convention on the

International Regulations for Preventing Collisionsat Sea (COLREGs) rules and regulations, the vesselcoming from the starboard side has higher priorityfor the navigation and is called the Stand on vesseland the vessel coming from the port side has lowerpriority and is called the Give way vessel. Thesedefinitions have been considered during the formula-tion of collision situations in this work.

The decision making process and strategies in in-

teraction situations in ocean navigation, includingcollision avoidance situations, are presented byChauvin and Lardjane (2008). The analysis of quan-titative data describing the manoeuvres undertakenby ferries and cargo-ships and behaviour of theGive way and the Stand on vessels with respectto verbal reports recorded on board a car-ferry in theDover Strait are also presented in the same work.

The detection of the Target vessel position and ve-locity are two important factors assessing the colli-sion risk in ocean navigation as illustrated in recentliterature. Sato and Ishii (1998) proposed combiningradar and infrared imaging to detect the Target ves-

sel conditions as part of a collision avoidance sys-tem. The collision risk has been presented with re-spect to the course of the Target vessel and imageprocessing based course measurement method pro-posed in the same work.

The vessel domain could be defined as the areabounded for dynamics of the marine vessel and thesize and the shape of the vessel domain are otherimportant factors assessing the collision risk inocean navigation. Lisowski et al. (2000) used neural-classifiers to support the navigator in the process ofdetermining the vessels domain, defining that the

area around the vessel should be free from stationaryor moving obstacles. On a similar approach,Pietrzykowski and Uriasz (2009) proposed the no-tion of vessel domain in a collision situation as de-pending on parameters like vessel size, course andheading angle of the encountered vessels. Fuzzylogic based domain determination system has beenfurther considered in the same work.

Kwik (1989) presented the calculations of two-ship collision encounter based on the kinematics anddynamics of the marine vessels. The analysis of col-lision avoidance situation is illustrated regarding thevessel velocity, turning rate and direction, and de-

sired passing distance in the same work. Yavin et al.(1995) considered the collision avoidance conditionsof a ship moving from one point to another in a nar-row zig-zag channel and a computational open loopcommand strategy for the rudder control system as-sociated with the numerical differential equationsolver is proposed. However, the dynamic solutionsbased on differential equations could face implemen-tation difficulties in real-time environment.

The design of a safe ship trajectory is an importantpart of the collision avoidance process and has usu-ally been simulated by mathematical models based


3/12



on manoeuvring theory (Sutulo et al. 2002). An al-ternative approach based on neural networks has alsobeen proposed by Moreira and Guedes Soares(2003). Optimization of a safe ship trajectory in col-lision situations by an evolutionary algorithm is pre-sented by Smierzchalski and Michalewicz (2000),where comparison of computational time for trajec-tory generation with respect to other manoeuvringalgorithms, and static and dynamic constrains for the

optimization process of the safe trajectories are alsoillustrated. However, the optimization algorithmsalways find the solution for the safe trajectory basedon assumptions; hence the optimum solutions maynot be realistic and may not have intelligent features.As an example, it is observed that some of the opti-mization algorithms always find the safest path be-hind the Target vessel and that may lead to a conflictsituation with the COLREGs rules and regulations.

1.3 The COLREGs

It is a fact that the COLREGs rules and regulations

regarding collision situations in ocean navigationhave been ignored in most of the optimization algo-rithms. The negligence of the IMO rules may lead toconflicts during ocean navigation. As for the re-ported data of the maritime accidents, 56% of themajor maritime collisions include violation of theCOLREGS rule and regulations (Statheros et al.(2008)). Therefore the methods proposed by the lit-erature ignoring the COLREGs rules and regulationsshould not be implemented in ocean navigation. Onthe other hand, there are some practical issues onimplementation of the COLREGs rules and regula-

tions during ocean navigation. Consider the crossingsituations where the Own vessel is in Give waysituations in Figures 4, 5, 6, and 7 and in Stand onsituations in Figures 9, 10, 11 and 12, there are ve-locity constrains in implementing COLREGs rulesand regulations of the Give way and the Standon vessels collision situations when the Target ves-sel has very low or very high speed compared to theOwn vessel.

In the collision avoidance approach of repulsiveforce based optimization algorithms proposed byXue et al. (2009), the Own vessel is kept away fromthe obstacles by a repulsive force field. However this

approach may lead to conflict situations when themoving obstacles present a very low speed or veryhigh speed when compared to the Own vessel speed.In addition, complex orientations of obstacles maylead to unavoidable collision situations. On the otherhand, repulsive force based optimization algorithmsare enforced to find the global safe trajectory forOwn vessel navigation, and this might not be a goodsolution for the localized trajectory search. In addi-tion the concepts of the Give way and the Standon vessels that are derived on COLREGs rules andregulations during the repulsive force based optimi-

zation process are not taken into consideration andtherefore may not be honoured.

The intelligent control strategies implemented oncollision avoidance systems could be categorized asAutomata, Hybrid systems, Petri nets, Neural net-works, Evolutionary algorithms and Fuzzy logic.These techniques are popular among the machinelearning researchers due to their intelligent learningcapabilities. The soft-computing based Artificial In-

telligence (AI) techniques, evolutionary algorithms,fuzzy logic, expert systems and neural networks andcombination of them (hybrid expert system), for col-lision avoidance in ocean navigation are summarizedby Statheros et al. (2008).

Ito et al. (1999) used genetic algorithms to searchfor safe trajectories on collision situations in oceannavigation. The approach is implemented in thetraining vessel of Shioji-maru integrating Auto-matic Radar Plotting Aids (ARPA) and DifferentialGlobal Position System (DGPS). ARPA system data,which could be formulated as a stochastic predictor,is designed such that the probability density map of

the existence of obstacles is derived from theMarkov process model before collision situations aspresented by Zeng et al. (2001) in the same experi-mental setup. Further, Hong et al. (1999) have pre-sented the collision free trajectory navigation basedon a recursive algorithm that is formulated by ana-lytical geometry and convex set theory. Similarly,Cheng et al. (2006) have presented trajectory optimi-zation for ship collision avoidance based on geneticalgorithms.

Liu and Liu (2006) used Case Based Reasoning(CBR) to illustrate the learning of collision avoid-

ance in ocean navigation by previous recorded dataof collision situations. In addition, a collision riskevaluation system based on a data fusion method isconsidered and fuzzy membership functions forevaluating the degree of risk are also proposed. Fur-ther intelligent anti-collision algorithm for differentcollision conditions has been designed and tested onthe computer based simulation platform by Yang etal. (2007) Zhuo and Hearn (2008) presented a studyof collision avoidance situation using a self learningneuro-fuzzy network based on an off-line trainingscheme and the study is based on two vessel colli-sion situation. Sugeno type Fuzzy Inference System

(FIS) was proposed for the decision making processof the collision avoidance.

1.4 Fuzzy-logic based systems

Fuzzy-logic based systems, which are formulated forhuman type thinking, facilitate a human friendly en-vironment during the decision making process.Hence several decision making systems in researchas well as commercial applications have been pre-sented before (Hardy (1995)). The conjunction ofhuman behavior and decision making process was


4/12

formulated by various fuzzy functions in Rommel-fanger, (1998) and Ozen et al., (2004). A fuzzy logicapproach for the collision avoidance conditions withintegration of the virtual force field has been pro-posed by Lee et al. (2004). However the simulationsresults are limited to the two vessel collision avoid-ance situations. The behaviour based controls formu-lated with interval programming for collision avoid-ance of ocean navigation are proposed by Benjamin

et al. (2006). Further, the collision avoidance behav-iour is illustrated accordance with the Coast GuardCollision Regulations (COLREGS-USA).

Benjamin and Curcio (2004) present the decisionmaking process of ocean navigation based on the in-terval programming model for multi-objective deci-sion making algorithms. The computational algo-rithm based on If-Then logic is defined and testedunder simulator conditions by Smeaton and Coenen(1990) regarding different collision situations. Fur-ther, this study has been focused on the rule-basedmanoeuvring advice system for collision avoidanceof ocean navigation.

Even though many techniques have been proposedfor avoidance of collision situations, those tech-niques usually ignore the law of the sea as formu-lated by the International Maritime Organization(IMO) in 1972. These rules and regulations are ex-pressed on the Convention on the InternationalRegulations for Preventing Collisions at Sea (COL-REGs)(IMO (1972)). The present convention wasdesigned to update and replace the Collision Regula-tions of 1960 which were adopted at the same timeas the International Convention for Safety of Life atSea (SOLAS) Convention (Cockcroft and Lameijer

(2001)).The detailed descriptions of collision avoidancerules of the COLREGs, how the regulations shouldbe interpreted and how to avoid collision, have beenpresented by Cockcroft and Lameijer (2001). Fur-ther, the complexity of autonomous navigation notonly in the sea but also in the ground has been dis-cussed by Benjamin and Curcio (2004). The legalframework, rules and regulations, is discussed andthe importance of collision avoidance within a set ofgiven rules and regulations is highlighted in thesame work.

2 COLREGS RULES AND REGULATIONS

The COLREGs (IMO (1972)) includes 38 rules thathave been divided into Part A (General), Part B(Steering and Sailing), Part C (Lights and Shapes),Part D (Sound and Light signals), and Part E (Ex-emptions). There are also four Annexes containingtechnical requirements concerning lights and shapesand their positioning, sound signalling appliances,additional signals for fishing vessels when operatingin close proximity, and international distress signals.

Three distinct situations involving risk of collision inocean navigation have been recognized in recent lit-erature (Smeaton and Coenen (1990)), Overtaking(see Figures 3 and 8); Head-on (see Figures 2) andCrossing (see Figures 4 to 7 and 9 to 12) and therules and regulations with respect to these collisionconditions have been highlighted by the COLREGs.

Figure 2: Head-on Figure 3: Overtake

Figure 4: Crossing Figure 5: Crossing

Figure 6: Parallel-crossing Figure 7: Crossing

Figure 8: Overtake Figure 9: Crossing


5/12



Figure 10: Parallel-crossing Figure 11: Crossing

Figure 12: Crossing

The maritime collision could be defined as abrief dynamic event consisting of the close prox-

imity of two or more stationary or moving ocean ob-stacles or vessels. Hence, maintenance of safe dis-tance among vessels and other obstacles areimportant factors of the maritime safety as illustratedby the COLREGs.

The safe distance maintenance by both vessel in aovertake situation (see Figures 3 and 8) has been il-lustrated by the COLREGs rule 13(a)(IMO (1972)).

..., any vessel overtaking any other shall keep out ofthe way of the vessel being overtaken.

Hence both vessels have the responsibility to

maintain safe distances during the Overtake encoun-ter. Further, the course change of the head-on situa-tion (see Figure 2) has been specified by the COL-REGs rule 13(a)( IMO (1972)).

When two power-driven vessels are meeting on re-ciprocal or nearly reciprocal course so as to involverisk of collision each shall alter her course to star-board so that each shall pass on the port side of theother.

As required by the COLREGs, the vessels should

take early action to avoid situations of crossingahead with the risk of collision in starboard to star-board and must be passing by port to port. The cross-ing situations further formulated by the COLREGsand according to rule 15

When a two power-driven vessel are crossing so asto involve risk of collision, the vessel which has theother her own starboard side shall keep out of thewan and shall, if the circumstances of the case admit,avoid crossing ahead of the other vessel

As specified by the COLREGs, the vessel comingfrom the starboard side has higher priority for thenavigation and the vessel coming from the port sidehas lower priority as mentioned before. This conceptis previously defined as the Give way and the Standon vessels. However, considering the collision con-ditions, where the Give way vessel did not takeany appropriate actions to avoid collisions, as illus-trated by the COLREGs rule 17(b) (IMO (1972)), the

Stand on vessel has been forced to take appropri-ate actions to avoid collision situation,

When, from any cause, the vessel required to keepher course and speed finds herself so close that colli-sion cannot be avoided by the action of the Giveway vessel alone, she shall take such action as willbest aid to avoid collision

However, the decision making process of the Ownvessel in this critical collision situation should becarefully formulated, because the collision avoidancein this situation alternatively depends on the Stand

on vessel manoeuvrability characteristics. Further,this situation might lead to a Crash stopping ma-neuver of the Stand on vessel due the lack of dis-tance for speed reductions. As recommended by theCOLREGs rule 6( IMO (1972)) with respect to thereduction of vessel speed for the safe distance,

Every vessel shall at all times proceed at a safespeed so that she can take proper and effective actionto avoid collision and be stopped within a distanceappropriate to the prevailing circumstances and con-ditions. In determining a safe speed the following

factors shall be among those taken into account ...

Hence special factors should be considered for in-tegration of course and speed controls due to the factthat the Own vessel may not response to the requiredchanges of course or speed. Further information withrespect to the Stopping distance and Turning cir-cles should be considered for formations of the de-cision making process. Vessel course changes and/orspeed changes in ocean navigation must be formu-lated in order to avoid collision situations. The spe-cific controllability of either course or speed changehas been highlighted in the COLREGs rule 8(b)

(IMO (1972)):

Any alteration of course and/or speed to avoid col-lision shall, if the circumstances of the case admit,be large enough to be readily apparent to anothervessel observing visually or by radar; a succession ofsmall alterations of course and/or speed should beavoided

Hence integrated controls of course as well as speedchanges should be implemented during ocean navi-gation to avoid collision situations. Similarly special


6/12

measures should be considered for integration ofcourse and speed controls due to the fact that theOwn vessel may not response to the requiredchanges of course or speed.

3 COLLISION AVOIDANCE METHODOLOGY

3.1 Identification of Obstacles

The stationary and moving obstacles in ocean navi-gation can be identified by several instruments andsystems like Eye / camera, radar / Automatic RadarPlotting Aid (ARPA), and Automatic IdentificationSystem (AIS). ARPA provides accurate informationof range and bearing of nearby obstacles and AIS iscapable of giving all the information on vessel struc-tural data, position, course, and speed. The AISsimulator and marine traffic simulator have beenimplemented on several experimental platforms fordesign of safe ship trajectories (Hasegawa (2009)).

3.2

Collection of Navigational InformationThe navigational information of other vessels can becategorized into static, dynamic and voyage relatedinformation (Imazu (2006)). Static information iscomposed of Maritime Mobile Service Identity(MMSI), Call Sign and Name, IMO number, lengthand beam, type of ship and location and position ofcommunication antenna. Dynamic information canbe divided into vessel position, position time stamp,course over ground, speed over ground, heading,navigational status and rate of turn. Finally, voyagerelated information can be expanded into vessel

draft, cargo type, destination and route plan. Collec-tion of navigational information is an important partof the decision making process of the collisionavoidance in ocean navigation and can be achievedby collaboration with the Automatic IdentificationSystem (AIS).

3.3 Analysis of Navigational Information

The collected obstacles and other vessels informa-tion should be considered for further analysis ofnavigational information. However continuous care-ful observation, collection and analysis of naviga-

tional information should be done in small regularintervals to obtain early warning for collision situa-tions. Further in ocean navigation, complex collisionsituation with the combination of above situationcould be occurred and identification of each situationwith respect to the each collision conditions willhelpfully for the overall decisions on ocean naviga-tion.

3.4 Assessments of the Collision Risk

The analysis of navigational information will help toassess the collision risk. The assessment of collisionrisk should be continuous and done in real-time bythe navigational system in order to guarantee theOwn vessel safety. As illustrated in the literature, themathematical analysis of collision risk detection canbe divided into two categories (Imazu (2006)): Clos-est Point Approach Method (CPA - 2D method) and

Predicted Area of Danger Method (PAD - 3Dmethod). CPA method consists of calculation of theshortest distance from the Own vessel to the Targetvessel and the assessment of the collision risk, whichcould be predicted with respect to the Own vesseldomain. However, this method is not sufficient toevaluate the collision risk since it doesnt take intoconsideration the target vessel size, course andspeed. The extensive study of CPA method with re-spect to a two vessel collision situation has been pre-sented by Kwik (1989).

The PAD method consists of modelling the Own

vessel possible trajectories as an inverted cone andthe Target vessel trajectory as an inverted cylinder,being the region of both object intersection catego-rized into the Predicted Area of Danger. The Targetvessel size, course and speed could be integrated intothe geometry of the objects of navigational trajecto-ries.

3.5 Decisions on Navigation

The decisions of collision avoidance in ocean navi-gation are based on the speed and course of eachvessel, distance between two vessels, distance of the

Closest Point of Approach (RDCPA), time toDCPA, neighbouring vessels and other environ-mental conditions. The decision space of collisionavoidance can be categorized into three stages foreach vessel in open ocean environment:

When both vessels are at non collision risk range,both vessels have the options to take appropriateactions to avoid a collision situation;

When both vessels are at collision risk range, theGive way vessel should take appropriate ac-tions to achieve safe passing distance in accor-dance with the COLREGs rules and regulationsand the Stand on vessel should keep the courseand speed;

When both vessels are at critical collision riskrange, and the Give way vessel does not takeappropriate actions to achieve safe passing dis-tance in accordance with the COLREGs rules,then Stand on vessel should take appropriateactions to avoid the collision situation.


7/12



3.6 Implementation of Decisions on Navigation

As the final step, the decisions on vessel navigationshould be formulated with respect to the collisionrisk assessments. The actions that are taken by theOwn vessel are proportional to the Target vessel be-haviour as well as the COLREGs rules and regula-tions. The expected Own and Target vessel actionsof collision avoidance could be formulated into twocategories: Course change and speed change. How-

ever the initiation for actions should be formulatedwith respect to the Target vessel range and the rateof change of range. Further if the actions taken bythe Target vessel are not clear or there is doubt aboutactions, sound signals should be used as recom-mended by the COLREGs.

In a collision situation, when the Target vessel isahead or fine on the bow of the Own vessel and theTarget vessel overtaking the Own vessel from asternor fine on the quarter without the safe distance, al-teration of course is more effective than speed altera-tion. However in a collision situation, when the Tar-

get vessel is approaching from abeam or near theabeam of the Own vessel, alteration of speed is moreeffective than course alteration but course alterationcould be achieved the same.(Cockcroft and Lameijer(2001)). On conventional navigational systems,power driven vessels usually prefer course changesover speed changes due to the difficulties and delaysin controllability of engines from the bridge unlessthe engines are on stand-by mode. However theseproblems could be overcome with the AGN systemswith the integration of speed and course control sys-tems.

4 CASE STUDY: FUZZY LOGIC BASEDDECISION MAKING PROCESS

This section focuses on a fuzzy logic based DecisionMaking (DM) system to be implemented on vesselnavigation to improve safety of the vessel by avoid-ing the collision situations and is an illustration ofthe study of Perera et al. (2009). The experiencedhelmsman actions in ocean navigation can be simu-lated by a fuzzy logic based Decision Making (DM)process, being this one of the main advantages in

this proposal.4.1 Formation of collision conditions

Figure 13 presents two vessels in a collision situa-tion in ocean navigation that is similar to a Radarplot in the Own vessel. The Own vessel is initiallylocated at the point O (xo, yo), and the Target vesselis located at the point A (xa, ya). The Own and Tar-get vessels velocities and course, are represented byVo, Va and o, a. The relative speed (Va,o) andcourse (a,o) of the Target vessel with respect to theOwn vessel can be estimated using the range and

bearing values in a given time interval. The relativetrajectory of the target vessel has been estimatedwith the derivation of relative speed, Va,o and rela-tive course, a,o. All angles have been measured re-garding the positive Y-axis. Further, it is assumedthat both vessels are power driven vessels regardingthe IMO categorization.

Figure 13: Relative Collision Situation in Ocean Navigation

Further, in Figure 13, the Own vessel ocean do-main is divided into three circular sections with ra-dius Rvd, Rb and Ra. The radius Ra represents the ap-proximate distance to the Target vessel identificationand this distance could be defined as the distance

where the Own vessel is in a Give way situationand should take appropriate actions to avoid colli-sion. The distance Rb represents the approximate dis-tance where the Own vessel is in Stand on situa-tion, but should take actions to avoid collisions dueto absence of the appropriate actions from the Targetvessel. Further the distance Rvd represents the vesseldomain. The approximate distances considered forthis study are Rvd 1NM, Rb 6NM and Ra 10NM.

The Own vessel collision regions are divided intoeight regions from I, to VIII. It is assumed that theTarget vessel should be located within these eight

regions and the collision avoidance decisions areformulated in accordance to each region. As repre-sented in target vessel position II in Figure 13, theTarget vessel positions have been divided into eightdivisions of vessel orientations regarding the relativecourse (from IIa to IIh).

4.2 Fuzzification and Defuzzification

The overall design process of the fuzzy logic basedDM system could be categorized into the followingsix steps.


8/12

Identification of the input Fuzzy MembershipFunctions (FMFs).

Identification of the output FMFs. Creation of FMF for each inputs and outputs. Construction of If-Then fuzzy rules to operate

overall system. Strength assignment of the fuzzy rules to execute

the actions. Combination of the rules and defuzzification of

the output.

4.2.1 Fuzzy Sets and Membership Functions

Fuzzy sets are defined by Fuzzy Membership Func-tions (FMF), which can be described as mappingsfrom one given universe of discourse to a unit inter-val. However it is conceptually and formally differ-ent from the fundamental concept of the probability(Pedrycz and Gomide (2007)). A fuzzy variable isusually defined by special FMF called LinguisticTerms that are used in fuzzy rule based inference.The Linguistic terms for the inputs (Collision Dis-

tance, Collision Region, Relative Collision Angleand Relative Speed Ratio) and outputs (Speed andCourse change of the Own vessel) were defined forthis analysis.

Mamdani type "IF is and/or is and/or .... THEN is " rule based fuzzy system and inference viaMin-Max norm was used during this study. OnMamdani fuzzy inference systems, the fuzzy sets re-sulting from the consequent part of each designedfuzzy rule are combined through an aggregation op-

erator (Ibrahim 2003) according to the activationlevel of the antecedents. The Min-Max norm is theaggregation operation considered in this study. Inthis norm the Min (minimum) operator is consideredfor intersection and Max (maximum) operator isconsidered for the union of two fuzzy sets.

Finally the defuzzification was made using thecenter of gravity method. In this method, one calcu-lates the centroid of the resulting FMF and uses itsabscissa as the final result of the inference.

4.2.2 Fuzzy Inference System (FIS)

The block diagram for Fuzzy Inference System (FIS)with integration of navigational instruments is pre-sented in Figure 14. The initial step of the fuzzy in-ference system consists of data collection of the tar-get vessel position, speed and course. As the nextstep, the relative trajectory, relative speed and rela-tive course of the Target vessel are estimated. Then,the data is fuzzified with respect to the input FMF ofCollision Distance, Collision Region, Relative SpeedRatio and Relative Collision Angle.

The If-Then fuzzy rules are developed in accor-dance with the COLREGs rules and regulations and

using navigational knowledge. The outputs of therule based system are the Collision Risk Warningand the Fuzzy Decisions. Finally the fuzzy decisionsare defuzzified by output FMF of Course Changeand Speed Change to obtain the control actions thatwill be executed in the Own vessel navigation. The

Figure 14: Block diagram for Fuzzy Inference System.

control actions are further expanded as Course con-trol commands and Speed control commands that

can be implemented on Rudder and Propulsion con-trol systems as presented in the Figure.

4.2.3 Computational Implementation

The fuzzy logic based DM system has been im-plemented on the software platform of MATLAB.MATLAB has support for the fuzzy logic schemesMamdani and Sugeno Types (Sivanandam et al.(2007)). However this work has been implementedon the Mamdani based Fuzzy Inference System(FIS). The Mamdani type fuzzy logic scheme con-sists of utilizing membership functions for both in-

puts and outputs. As previously mentioned, If-Thenrules are formed by applying fuzzy operations intothe Mamdani type membership functions for giveninputs and outputs.

The MATLAB simulations for two vessels colli-sion situations with respect to the different speedsand course conditions in the Cartesian coordinatespace have been presented in Figures from 15 to 22.These figures contain the start and end positions ofthe Own and Target vessels with respect to naviga-tional trajectories. The initial vessel speed conditionis Vo/Va = 0.5 and initial course of the own vessel is


9/12



o = 00. The star position of the Own vessel (0, 0)

and the collision point for both vessels (0, 5) hasbeen considered for all simulations. As noted fromthe simulations fuzzy logic based DM systems hasmade proper trajectory for all the collision condi-tions.

Figure 15: Heading Situation

Figure 16: Crossing Situation




10/12

Figure 19: Overtake Situation





11/12



5 CONCLUSION

An overall discussion about inclusion of collisionavoidance in AGN systems has been presented. Inthe case study, the decision making process of AGNsystem that was based on the fuzzy logic and humanexpert knowledge in ocean navigation is introduced.As observed, the DM system was able to overcomecollision conditions by fuzzy logic based decisionmaking process. Furthermore the DM system hastaken proper maneuvers to avoid close-quarter situa-tion during the ocean navigation where the collisionrisk is high. Therefore the DM system would nothave to take quick decisions based on inadequate in-

formation and time as human decisions.

ACKNOWLEDGMENT

The research work of the first author has been sup-ported by the Doctoral Fellowship of the PortugueseFoundation for Science and Technology (Fundaopara a Cincia e a Tecnologia) under the contractSFRH/BD/46270/2008. Furthermore, this work con-tributes to the project Methodology for ships ma-

neuverability tests with self-propelled models,which is being funded by the Portuguese Foundationfor Science and Technology ( Fundao para a Cin-cia e Tecnologia ) under the contractPTDC/TRA/74332/2006

REFERENCES

Anto, P., Guedes Soares, C., 2008. Causal factors in accidentsof high speed craft and conventional ocean going vessels.Reliability Engineering and System Safety 93, 12921304.

Benjamin, M. R., Curcio, J. A., 2004. COLREGS - based navi-gation of autonomous marine vehicles. IEEE/OES Autono-mous Underwater Vehicles, 3239.

Benjamin, M. R., Curcio, J. A., Newman, P. M., 2006. Naviga-tion of unmanned marine vehicles in accordance with therules of the road. Proceedings of the 2006 IEEE Interna-tional Conference on Robotics and Automation, 35813587.

Bennett, S.,1984. Nicholas minorsky and the automatic steeringof ships. Control Systems Magazine, IEEE 4 (4), 1015.

Chauvin, C., Lardjane, S., 2008. Decision making and strate-gies in an interaction situation: Collision avoidance at sea.Transportation Research Part F (11), 259262.

Cheng, X. D., Liu, Z. Y., Zhang, X. T., 2006. Trajectory opti-mization for ship collision avoidance system using genetic

algorithm. Proceedings OCEANS- Asia Pacific, Singapore,385-392.

Cockcroft, A. N., Lameijer, J. N. F., 2001. A Guide to The Col-lision Avoidance Rules. Elsevier Butterworth-Heinemann,Burlington, MA. USA.

Do, K. D., Pan, J., 2006. Robust path-following of underactu-ated ships: Theory and experiments on a model ship. OceanEngineering 33 (3), 13541372.

eNAV, eNavigation 2008. URL http://www.enavigation.org/.Fossen, T. I. (Ed.), 1999. Recent developments in Ship Control

Systems Design. World Superyacht Review. Sterling Publi-

cation Limited, London.Hardy, T. L., 1995. Multi-objective decision-making under un-

certainty: Fuzzy logic method. NASA Technical Memoran-dum - Computing in Aerospace 10 Meeting (106796), 16.

Hasegawa, K., 2009. Advanced marine traffic automation andmanagement system for congested waterways and coastalareas. Proceedings of International Conference in OceanEngineering (ICOE), Chenai, 110.

Healey, A. J., Lienard, D., 1993. Multivariable sliding modecontrol for autonomous diving and steering of unmannedunderwater vehicle. IEEE Journal of Oceanic Engineering18 (3), 327339.

Hong, X., Harris, C. J.,Wilson, P. A., 1999. Autonomous shipcollision free trajectory navigation and control algorithms.Proceedings of 7th IEEE International Conference onEmerging Technologies and Factory Automation, Barce-lona, Spain 923929.

Ibrahim A. M., 2003, Fuzzy Logic for Embedded Systems Ap-plications, Elsevier Science (USA).

Imazu, H., 2006. Advanced topics for marine technology andlogistics. Lecture Notes on Ship collision and integrated in-formation system, Tokyo University of Marine Science andTechnology, Tokyo, Japan.

IMO, 1972. Convention on the international regulations forpreventing collisions at sea (COLREGs). URLhttp://www.imo.org

Ito,M., Zhang, F., Yoshida, N., 1999. Collision avoidance con-trol of ship with genetic algorithm. Proceedings of the 1999IEEE International Conference on Control Applications,

17911796.Kwik, K. H., 1989. Calculations of ship collision avoidancemanoeuvres: A simplified approach. Ocean Engineering 16(5/6), 475491.

Lee, S., Kwon, K., Joh, J., 2004. A fuzzy logic for autonomousnavigation of marine vehicles satisfying colreg guidelines.International Journal of Control Automation and Systems 2(2), 171181.

Lisowski, J., Rak, A., Czechowicz, W., 2000. Neural networkclassifier for ship domain assessment. Mathematics andComputers in Simulations 51, 399406.

Liu, Y., Liu, H., 2006. Case learning base on evaluation systemfor vessel collision avoidance. Proceedings of the Fifth In-ternational Conference on Machine Learning and Cybernet-ics, Dalian, China, 20642069.

Minorski, N., 1922. Directional stability of automatic steeredbodies. Journal of American Society of Naval Engineers 34(2), 280309.

Moreira, L., Fossen, T. I., Guedes Soares, C., 2007. Path fol-lowing control system for a tanker ship model. Ocean Engi-neering 34, 20742085.

Moreira, L., Guedes Soares, C., 2003. Dynamic model of ma-neuverability using recursive neural networks. Ocean Engi-neering 30 (13), 16691697.

Ohtsu, K., 1999. Recent development on analysis and controlof ships motions. Proceedings of the 1999 IEEE Interna-tional Conference on Control Applications, 10961103.

Ozen, T., Garibaldi, J. M., Musikasuwan, S., 2004. Modelingthe variation in human decision making. Proceedings IEEE


12/12

Annual Meeting of the Fuzzy Information Processing -NAFIPS 04 2, 617622.

Pedrycz, W., Gomide, E., 2007. Fuzzy Systems Engineering -Toward Human Centric Computing. A John Wiley & Sons,INC, Hoboken, New Jersey.

Perera, L. P., Carvalho, J. P., Guedes Soares, C., 2009. Deci-sion making system for the collision avoidance of marinevessel navigation based on COLREGs rules and regulations.Proceedings 13th Congress of International Maritime Asso-ciation of Mediterranean, Istambul, 1121-1128.

Pietrzykowski, Z., Uriasz, J., 2009. The ship domain - a crite-

rion of navigational safety assessment in an open sea area.The Journal of Navigation 63, 93108.

Roberts, G., Sutton, R., Zirilli, A., Tiano, A., 2003. Intelligentship autopilots - a historical perspective. Mechatronics 13(1), 10911103.

Rommelfanger, H. J., 1998. Multicriteria decision making us-ing fuzzy logic. Proceedings Conference of the NorthAmerican Fuzzy Information Processing Society - NAFIPS1998, 360364.

Rothblum, A.M.,Wheal, D.,Withington, S., Shappell, S.A.,Wiegmann, D. A., Boehm,W., Chaderjian, M., 2002.Key to successful incident inquiry. Proceedings 2nd Interna-tional Workshop on Human Factors in Offshore Operations(HFW), Houston, Texas, 16.

Sato, Y., Ishii, H., 1998. Study of a collision-avoidance system

for ships. Control Engineering Practice 6, 11411149.Sivanandam, S. N., Sumathi, S., Deepa, S. N., 2007. Introduc-

tion to Fuzzy Logic using MATLAB. Springer Berlin Hei-delberg, New York.

Smeaton, G. P., Coenen, F. P., 1990. Developing an intelligentmarine navigation system. Computing & Control Engineer-ing Journal, 95103.

Smierzchalski, R., Michalewicz, Z., 2000. Modeling of shiptrajectory in collision situations by an evolutionary algo-rithm. IEEE Transactions on Evolutionary Computation 4(3), 227241.

Sperry, E., 1922. Automatic steering. Transactions: Society ofNaval Architects and Marine Engineers, 5361.

Statheros, T., Howells, G., McDonald-Maier, K., 2008.Autonomous ship collision avoidance navigation concepts,technologies and techniques. The Journal of Navigation, 61,129142.

Sutulo, S., Moreira, L., Guedes Soares, C., 2002. Mathematicalmodels of ship path prediction in manoeuvring simulationsystems. Ocean Engineering 29, 119.

Xue, Y., Lee, B.S., Han D., 2009, Automatic collision avoid-ance of ships, Proceedings of the Institution of MechanicalEngineers, Part M: Journal of Engineering for the MaritimeEnvironment, 223 (1), 3346.

Yang, S., Li, L., Suo, Y., Chen, G., 2007. Study on construc-tion of simulation platform for vessel automatic anti-collision and its test methods. Proceedings IEEE Interna-tional Conference on Automation and Logistics, 24142419.

Yavin, Y., Frangos, C., Zilman, G., Miloh, T., 1995. Computa-tion of feasible command strategies for the navigation of aship in a narrow zigzag channel. Computers Math. Applic.30 (10), 79101.

Zeng, X., Ito, M., Shimizu, E., 2001. Building an automaticcontrol system of maneuvering ship in collision situationwith genetic algorithms. Proceedings of the American Con-trol Conference, Arlington, VA, USA, 28522853.

Zhuo, Y., Hearn, G. E., 2008. A ship based intelligent anti-collision decision-making support system utilizing trial ma-noeuvers. Proceedings Chinese Control and Decision Con-ference - CCDC, Yantai, China, 39823987.

autonomous guidance and navigation based on the colregs rules and regulations of collision avoidance

Documents