Efficient collaborative collision avoidance algorithm considering COLREGs in complex multi-ship encounter scenarios

• Designed a collision avoidance decision-making algorithm considering the uncertainties in the motion of other ships. • Coordinated multi-ship avoidance timing by integrating COLREGs and ship maneuverability characteristics. • Generated mutually matching situation classification results in multi-ship encounter scenarios. • A novel VO method is proposed, considering both ship dynamic constraints and decision-making efficiency. In recent years, the issue of autonomous ship collision avoidance has attracted widespread attention. Although various automatic collision avoidance algorithms have been proposed, the development of decision-making systems for collision avoidance in complex multi-ship encounter scenarios and in cases where the target ship’s motion is uncertain has not received sufficient attention. To address this gap, the study considers the uncertainty in the velocity observations of other vessels and proposes a time-interactive ship domain model to assess collision risks. Combining the maneuvering characteristics of ships, a dual time-scale domain model is proposed to accurately determine the timing for evasive maneuvers. Based on collision avoidance regulations, a role-symmetric encounter situation classification algorithm is developed to clarify the coordinated actions in multi-ship encounter scenarios. Moreover, a velocity obstacle primitive method constrained by ship dynamics is proposed to generate real-time evasive actions. The experimental results show that the proposed autonomous ship collision avoidance decision-making algorithm not only ensures navigation safety but also demonstrates high decision-making efficiency, coordinating the actions in multi-ship encounters to provide safe and efficient collision avoidance strategies.