Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance

Wu, Yizhou; Liu, Jin; Li, Xingye; Xiao, Junsheng; Zhang, Tao; Xu, Haitong; Zhang, Lei

doi:10.3390/jmse13122275

Open AccessReview

Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance

by

Yizhou Wu

¹,

Jin Liu

^2,*

,

Xingye Li

²

,

Junsheng Xiao

²,

Tao Zhang

¹,

Haitong Xu

³

and

Lei Zhang

^4,*

¹

Merchant Marine College, Shanghai Maritime University, Shanghai 201306, China

²

College of Information Engineering, Shanghai Maritime University, Shanghai 201306, China

³

Centre for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Técnico, Universidade de Lisboa, 1649-004 Lisboa, Portugal

⁴

Shanghai Science Center for Autonomous Intelligent Unmanned Systems, Tongji University, Shanghai 200092, China

^*

Authors to whom correspondence should be addressed.

J. Mar. Sci. Eng. 2025, 13(12), 2275; https://doi.org/10.3390/jmse13122275

Submission received: 4 November 2025 / Revised: 21 November 2025 / Accepted: 25 November 2025 / Published: 28 November 2025

(This article belongs to the Special Issue Emerging Topics in Intelligent Technology for Maritime Autonomous Systems)

Download

Browse Figures

Versions Notes

Abstract

This comprehensive review examines the works of reinforcement learning (RL) in ship collision avoidance (SCA) from 2014 to the present, analyzing the methods designed for both single-agent and multi-agent collaborative paradigms. While prior research has demonstrated RL’s advantages in environmental adaptability, autonomous decision-making, and online optimization over traditional control methods, this study systematically addresses the algorithmic improvements, implementation challenges, and functional roles of RL techniques in SCA, such as Deep Q-Network (DQN), Proximal Policy Optimization (PPO), and Multi-Agent Reinforcement Learning (MARL). It also highlights how these technologies address critical challenges in SCA, including dynamic obstacle avoidance, compliance with Convention on the International Regulations for Preventing Collisions at Sea (COLREGs), and coordination in dense traffic scenarios, while underscoring persistent limitations such as idealized assumptions, scalability issues, and robustness in uncertain environments. Contributions include a structured analysis of recent technological evolution, and a Large Language Model (LLM) based hierarchical architecture integrating perception, communication, decision-making, and execution layers for future SCA systems, which prioritizes the development of scalable, adaptive frameworks that ensure robust and compliant autonomous navigation in complex, real-world maritime environments.

Keywords: reinforcement learning; ship collision avoidance; multi-agent reinforcement learning; large language model; COLREGs compliance

Share and Cite

MDPI and ACS Style

Wu, Y.; Liu, J.; Li, X.; Xiao, J.; Zhang, T.; Xu, H.; Zhang, L. Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance. J. Mar. Sci. Eng. 2025, 13, 2275. https://doi.org/10.3390/jmse13122275

AMA Style

Wu Y, Liu J, Li X, Xiao J, Zhang T, Xu H, Zhang L. Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance. Journal of Marine Science and Engineering. 2025; 13(12):2275. https://doi.org/10.3390/jmse13122275

Chicago/Turabian Style

Wu, Yizhou, Jin Liu, Xingye Li, Junsheng Xiao, Tao Zhang, Haitong Xu, and Lei Zhang. 2025. "Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance" Journal of Marine Science and Engineering 13, no. 12: 2275. https://doi.org/10.3390/jmse13122275

APA Style

Wu, Y., Liu, J., Li, X., Xiao, J., Zhang, T., Xu, H., & Zhang, L. (2025). Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance. Journal of Marine Science and Engineering, 13(12), 2275. https://doi.org/10.3390/jmse13122275

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Menu

Towards LLM Enhanced Decision: A Survey on Reinforcement Learning Based Ship Collision Avoidance

Abstract

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI