Protocol-Driven A* Algorithm for Fast ASV Motion Planning in Dynamic Scenarios

This paper proposes a new local path planner for Autonomous Surface Vehicles that integrates navigation rules into the planning process. The planner operates within a three-level architecture designed for the whole motion planning process, from global path creation to real-time course adjustments. This integration aims to improve efficiency while complying with maritime collision avoidance regulations. Previous attempts to integrate the 'Convention on the International Regulations for Preventing Collisions at Sea' rules relied on reactive solutions, leading to frequent course changes and inefficiency. The planner incorporates these regulations within the A*-based algorithm of the architecture's second level. It receives obstacle data and plans a path to the next global waypoint. The correct maneuvers are determined based on the obstacle's approaching angle, and the A* algorithm's nodes are selected to ensure compliance while searching for the fastest route. Planned path safety is verified using a geometric ray-tracing algorithm that estimates future obstacle positions and checks for potential collisions. The planner was evaluated using simulations and field tests with the ULISSE autonomous catamaran. The experiments demonstrate the system's robustness in handling uncertainties and noise, while correctly avoiding dynamic obstacles.