Dynamic domain-based collision avoidance system for autonomous ships: Real experiments in coastal waters

This study proposes a dynamic domain-based collision avoidance system consisting of environment perception, situation awareness, guidance, navigation and control modules for autonomous ships. A dynamic ship domain considering the relative motion between target ships (TSs) and own ship (OS), has been established. A Nomoto model with a predefined speed during path planning is adopted to generate followable paths conforming to the ship kinematics. To handle the problem of collision avoidance uncertainties in real-world induced by environmental disturbances or inconsistent actions, a local path planning algorithm based on a virtual potential field, dynamic ship domains and a re-planning strategy, is proposed to continuously generate feasible collision avoidance paths. During experimental testing, a method based on virtual automatic identification system (AIS) information is introduced to simulate TSs, and a real ferry is utilized as the TS, the course of which is variable. The results of real experiments in coastal waters of China, demonstrate that the proposed collision avoidance method and system can achieve safe and efficient collision avoidance in the presence of environmental disturbances and randomly moving TSs.