A COLREGs-compliant decision-making method for collision avoidance of intelligent ships based on Monte Carlo Tree Search

Realizing reliable collision avoidance is a key capability for ships to ensure safe navigation. To address the decision-making issue for collision avoidance of intelligent ships, an algorithm based on Monte Carlo Tree Search (MCTS) is introduced to optimize sequential ship maneuvering actions throughout the entire collision avoidance process. By integrating relevant rules of the International Regulations for Preventing Collisions at Sea (COLREGs) and the ship maneuvering motion model into the MCTS decision-making process, a novel decision-making method for collision avoidance along with re-navigation function is proposed. The developed method generates a set of appropriate collision avoidance behaviors and corresponding trajectories of the ship in line with COLREGs rules within a short time frame. Three scenarios involving different encounter situations between an own ship and a target ship are designed and simulated to evaluate the performance of the proposed method. In particular, each scenario is conducted with 20 independent runs to ensure statistical reliability of the method. Finally, the simulation results verify the effectiveness and robustness of the proposed decision-making method in optimizing ship collision-avoidance maneuvers and achieving good re-navigation performance, highlighting the remarkable potential of MCTS-based decision-making method in tackling collision avoidance problems considering ship maneuvering dynamics.