Article Search
닫기

Original Article

Split Viewer

International Journal of Fuzzy Logic and Intelligent Systems 2021; 21(4): 378-390

Published online December 25, 2021

https://doi.org/10.5391/IJFIS.2021.21.4.378

© The Korean Institute of Intelligent Systems

Improved Collision Avoidance Method for Autonomous Surface Vessels Based on Model Predictive Control Using Particle Swarm Optimization

Jinwan Park

Department of Maritime Transportation System, Mokpo National Maritime University, Mokpo, Korea

Correspondence to :
Jinwan Park (pjinwan2@gmail.com)
*These authors contributed equally to this work.

Received: November 10, 2021; Revised: December 11, 2021; Accepted: December 17, 2021

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Most collision accidents are caused by human errors in an encounter situation between vessels. Autonomous operations have received much interest among researchers to reduce human errors resulting from improper vessel maneuvering actions. In this paper, an intelligent collision avoidance method that considers not only the Convention on the International Regulations for Preventing Collisions at Sea (COLREGs) but also complex encounter situations with multiple obstacle vessels was proposed for autonomous surface vessels (ASVs). The results were verified by comparing the performance with that of existing methods. To take timely maneuvering actions to prevent the collision of vessels involved in an encounter, fuzzy comprehensive evaluation, and model predictive control using particle swarm were applied to path planning. As a result, ASVs can take proper action to avoid collisions in compliance with COLREGs based on the proposed method. The improved collision avoidance method exhibits better performance than existing methods.

Keywords: Autonomous surface vessels, Path planning, Collision avoidance, Model predictive control, Particle swarm optimization

No potential conflict of interest relevant to this article was reported.

Jinwan Park received his Ph.D. degree in Maritime Information Systems from Mokpo National Maritime University, Korea. His research interests include maritime safety, vessel traffic theory, ASV control, optimization method, and intelligent navigation systems.

E-mail: pjinwan2@gmail.com


Article

Original Article

International Journal of Fuzzy Logic and Intelligent Systems 2021; 21(4): 378-390

Published online December 25, 2021 https://doi.org/10.5391/IJFIS.2021.21.4.378

Copyright © The Korean Institute of Intelligent Systems.

Improved Collision Avoidance Method for Autonomous Surface Vessels Based on Model Predictive Control Using Particle Swarm Optimization

Jinwan Park

Department of Maritime Transportation System, Mokpo National Maritime University, Mokpo, Korea

Correspondence to:Jinwan Park (pjinwan2@gmail.com)
*These authors contributed equally to this work.

Received: November 10, 2021; Revised: December 11, 2021; Accepted: December 17, 2021

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Most collision accidents are caused by human errors in an encounter situation between vessels. Autonomous operations have received much interest among researchers to reduce human errors resulting from improper vessel maneuvering actions. In this paper, an intelligent collision avoidance method that considers not only the Convention on the International Regulations for Preventing Collisions at Sea (COLREGs) but also complex encounter situations with multiple obstacle vessels was proposed for autonomous surface vessels (ASVs). The results were verified by comparing the performance with that of existing methods. To take timely maneuvering actions to prevent the collision of vessels involved in an encounter, fuzzy comprehensive evaluation, and model predictive control using particle swarm were applied to path planning. As a result, ASVs can take proper action to avoid collisions in compliance with COLREGs based on the proposed method. The improved collision avoidance method exhibits better performance than existing methods.

Keywords: Autonomous surface vessels, Path planning, Collision avoidance, Model predictive control, Particle swarm optimization

Fig 1.

Figure 1.

Definition of encounter situation.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 2.

Figure 2.

Steering and sailing rules in each encounter situation defined by COLREGs: (a) overtaking, (b) head-on situation, and (c) crossing situation.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 3.

Figure 3.

Flowchart of collision avoidance for ASVs.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 4.

Figure 4.

Basic idea of the model predictive control.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 5.

Figure 5.

Structure of collision avoidance system based on MPC-PSO.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 6.

Figure 6.

Simulation results for head-on situation. (a) Trajectory plot until the distance between the ASV and the obstacle vessel are closest. (b) Full trajectory plot of the ASV and obstacle vessel. (c) The course offset angle χca and the propulsion P of the ASV for collision avoidance. (d) Actual course command χ of the ASV.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 7.

Figure 7.

Simulation results for starboard crossing situation. (a) Trajectory plot until the distance between the ASV and the obstacle vessel is closest. (b) Full trajectory plot of the ASV and obstacle vessel. (c) The course offset angle χca and the propulsion P of the ASV for collision avoidance. (d) Actual course command χ of the ASV.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 8.

Figure 8.

Simulation results for port crossing situation. (a) Trajectory plot to minimize the distance between the ASV and the obstacle vessel.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 9.

Figure 9.

Simulation results for overtaking. (a) Trajectory plot to minimize the distance between the ASV and the obstacle vessel. (b) Full trajectory plot of the ASV and obstacle vessel. (c) The course offset angle χca and the propulsion P of the ASV for collision avoidance. (d) Actual course command χ of the ASV.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 10.

Figure 10.

Simulation results for an encounter situation with multiple obstacle vessels based on MPC-PSO. (a) Full trajectory plot of the ASV and each obstacle vessel. (b) Trajectory plot up to the point where the distance between the ASV and each obstacle vessel are the closest. (c) The course offset angle χca and the propulsion P of the ASV for collision avoidance. (d) Actual course command χ of the ASV.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Fig 11.

Figure 11.

Simulation results for an encounter situation with multiple obstacle vessels based on SBMPC. (a) Full trajectory plot of the ASV and each obstacle vessel. (b) Trajectory plot up to the point where the distance between the ASV and each obstacle vessel is the closest. (c) The course offset angle χca and the propulsion P of the ASV for collision avoidance. (d) Actual course command χ of the ASV.

The International Journal of Fuzzy Logic and Intelligent Systems 2021; 21: 378-390https://doi.org/10.5391/IJFIS.2021.21.4.378

Table 1 . Parameters for simulation.

ParameterValueParameterValue
Tc5 sec5.5, 10
Tp5 secW3
Te1,700 sec, 3,500 secWP30
Hc25 secWχ0.9
Hp600 secWΔP10
Ts10 minWΔχ0.6
LOA1120 mSWsize20
LOA2100 mMaxi15

Table 2 . Initial navigation parameters for head-on situation.

ObjectPosition (x, y)Speed (knot)Course (°)
ASV(0, 0)100
Obstacle(0, 10,000)8180

Table 3 . Initial navigation parameters for crossing situation.

ObjectInitial position (x, y)Speed (knot)Initial course (°)
ASV(0, 0)100
Obstacle(4000, 4000)10270

Table 4 . Initial navigation parameters for crossing situation.

ObjectInitial position (x, y)Speed (knot)Initial course (°)
ASV(0, 0)100
Obstacle(−3000, 6000)690

Table 5 . Initial navigation parameters for overtaking.

ObjectInitial position (x, y)Speed (knot)Initial course (°)
ASV(0, 0)100
Obstacle(0, 3000)30

Table 6 . Initial navigation parameters for an encounter situation with multiple obstacles.

ObjectInitial position (x, y)Speed (knot)Initial course (°)
ASV(0, 0)100
Obstacle1(−3000, 6000)690
Obstacle2(10000,19000)6247
Obstacle3(10000,18000)5252
Obstacle4(300,20000)7180
Obstacle5(500, 20000)6182

Table 7 . The closest distance between ASV and each Obstacle in the validation simulation (measure of distance: nm).

MethodObstacle1Obstacle2Obstacle3Obstacle4Obstacle5Ave. value
MPC-PSO1.04271.45850.81381.96692.00421.4572
SBMPC0.51630.80501.33150.43710.30190.6784

Share this article on :

Related articles in IJFIS

Most KeyWord