Coordinated multi-stage and multi-objective optimization approach for ship collision avoidance decision-making

Abstract

Appropriate decision-making is essential to prevent ship collisions and further ensure navigational safety at sea. Current research on collision avoidance decision-making (CADM) primarily aims to find limited feasible solutions based on a designated safe distance under the strong assumption that ships take no or little maneuvering actions. To overcome such limitations, a multi-stage and multi-objective CADM optimization approach is proposed to satisfy the requirements of safety and economy, while considering the ship path uncertainties and risk preferences of ship officers. First, a comprehensive collision risk estimation method is formulated by quantifying degree of domain violation and probability of domain violation (PDV) in dynamic and uncertain situations. Then, the CADM for each individual ship is treated as a multi-objective optimization problem, and a three-stage optimization approach outlines the process to find optimal collision avoidance path. Preliminary non-dominated solutions are obtained and filtered to enhance the robustness by an accepted threshold of the PDV. Optimal solutions are selected for ship officers with different risk preferences. Finally, the coordinated collision avoidance path for multi-ship encounter is determined based on the exchange of individual collision avoidance path through several rounds of coordination. The effectiveness of CADM method is validated through extensive case studies. The results show that it can provide feasible solutions for multi-objectives and adaptively determine the most appropriate scheme for ship officers with different risk preferences to make coordinated decisions.