Abstract:Automatic collision avoidance decision making for vessels is a critical challenge in the development of autonomous ships and has become a central point of research in the maritime safety domain. Effective and systematic collision avoidance strategies significantly reduce the risk of vessel collisions, ensuring safe navigation. This study develops a multi-vessel automatic collision avoidance decision-making method based on deep reinforcement learning (DRL) and establishes a vessel behavior decision model. When designing the reward function for continuous action spaces, the criteria of the "Convention on the International Regulations for Preventing Collisions at Sea" (COLREGs) were adhered to, taking into account the vessel's collision risk under various encounter situations, real-world navigation practices, and navigational complexities. Furthermore, to enable the algorithm to precisely differentiate between collision avoidance and the navigation resumption phase in varied vessel encounter situations, this paper incorporated "collision avoidance decision making" and "course recovery decision making" as state parameters in the state set design, from which the respective objective functions were defined. To further enhance the algorithm's performance, techniques such as behavior cloning, residual networks, and CPU-GPU dual-core parallel processing modules were integrated. Through simulation experiments in the enhanced Imazu training environment, the practicality of the method, taking into account the effects of wind and ocean currents, was corroborated. The results demonstrate that the proposed algorithm can perform effective collision avoidance decision making in a range of vessel encounter situations, indicating its efficiency and robust generalization capabilities.

Grounding hazard considerations in evaluation of COLREGS collision avoidance algorithms

Exploration of COLREG-relevant parameters from historical AIS-data

COLREG Scenario classification and Compliance Evaluation with temporal and multi-vessel awareness for collision avoidance systems

Codification of Good Seamanship in Complex and Congested Waterways

Stochastic COLREGs Evaluation for Safe Navigation under Uncertainty

New frontiers in the risk assessment of ship collision

A Machine Learning Method for the Evaluation of Ship Grounding Risk in Real Operational Conditions

Autonomous Collision Avoidance at Sea: A Survey

Theoretical and Statistical Analysis of Ship Grounding Accidents

Towards A COLREGs Compliant Autonomous Surface Vessel in a Constrained Channel

Multi-Ship Collision Avoidance Decision-Making Based on Collision Risk Index

Collision Avoidance Algorithm Based on COLREGs for Unmanned Surface Vehicle

Hybrid Navigation Acceptability and Safety

COLREG and MASS: Analytical review to identify research trends and gaps in the Development of Autonomous Collision Avoidance

An Improved NSGA-II Algorithm for MASS Autonomous Collision Avoidance under COLREGs

A COLREGs-Compliant Deep Reinforcement Learning Approach

Autonomous Navigation in Confined Waters -- A COLREGs Rule 9 Compliant Framework

A Systematic Approach for Collision Risk Analysis Based on Ais Data

Developments and research directions for collision avoidance in mixed navigation environment for MASS: A systematic literature review with bibliometric analysis.

Analysis of structural crashworthiness of double-hull ships in collision and grounding

Survey on grounding incidents: Statistical analysis and risk assessment