In order to solve the safe navigation decision-making problem of autonomous ship in the complex encounter situation, an autonomous collision avoidance method based on deep reinforcement learning (DRL) is proposed. First, on basis of analyzing the difference of collision avoidance mechanisms in the different scenarios with various obstacles and target ships, we design a reward function which is constraints with the International Regulations for Preventing Collisions at Sea (COLREGs). Among them, we innovatively make collision risk assessment model, avoidance responsibility, and interactive rules integrate into the reward function. Second, the virtual potential energy of static obstacles is introduced to improve the DRL. Finally, we validate the proposed collision avoidance algorithm in the two-ship encounter scenario and multi-ship encounter scenario, separately, through the performance of the reward convergence trend and time efficiency. The results indicate that the algorithm can plan a collision-free path under the premise of obeying the COLREGs in an unknown complex environment.
COLREGs-Compliant Collision Avoidance Method for Autonomous Ships via Deep Reinforcement Learning
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2021 ; Changsha, China September 24, 2021 - September 26, 2021
Proceedings of 2021 International Conference on Autonomous Unmanned Systems (ICAUS 2021) ; Chapter : 199 ; 2016-2026
2022-03-18
11 pages
Article/Chapter (Book)
Electronic Resource
English
Collision-avoidance under COLREGS for unmanned surface vehicles via deep reinforcement learning
Taylor & Francis Verlag | 2020
|Hybrid Collision Avoidance for ASVs Compliant with COLREGs Rules 8 and 13-17
BASE | 2020
|An automatic COLREGs-compliant obstacle avoidance system for an unmanned surface vehicle
Online Contents | 2014
|