Collision avoidance decision-making for stand-on ships through inference of give-way ship intention in complex encounters

Rationally determining the optimal timing and devising effective collision avoidance strategies are crucial for stand-on ships in complex encounter scenarios. Given the absence of explicit COLREGs guidelines on when and how a stand-on ship should initiate an evasive action, a collision avoidance method from the stand-on ship's perspective through inferring the give-way ship's intention is necessary to be developed. By continuously monitoring and analyzing the give-way ship's behavior, this paper employs Dempster-Shafer theory to assess its cooperative intention, enabling rational determination of the optimal timing for action. The improved nonlinear velocity obstacle method, incorporating the stand-on ship's maneuvering capabilities, is then used to generate a set of feasible anti-collision decision strategies. A NSGA-II algorithm helps optimally select a strategy that can balance safety and navigational efficiency. Simulation results demonstrate that the newly proposed approach enhances stand-on ships' situational awareness across varying levels of give-way ship cooperation, effectively mitigating collision risks while significantly reducing route recovery time and course deviation.