Unmanned surface vehicle (USV) is an important autonomous marine vehicle. The safe navigation of USV is directly determined by the local obstacle avoidance because it must avoid real-time local obstacle in the global path planning in a three-dimensional environment. Therefore, efficient algorithms of real-time local obstacle avoidance for USV are a critical issue. In this study, a new threelayered architecture for local obstacle avoidance algorithm was proposed to solve the local obstacle avoidance problem. First, real environment and obstacle models were established in the polar coordinate. The known static path-planning method was conducted based on particle swarm optimization (PSO). Second, the method was integrated with marine rules based on PSO. Third, an obstacle avoidance method under unknown environment was created based on rolling windows. Finally, a simulation experimental platform was developed to verify the feasibility and effectiveness of the aforementioned measure. Result shows that the proposed algorithm can effectively avoid local obstacles of USV at a computation time of less than 2 s. The USV avoids the obstacles smoothly and reaches the desired destination with complex requirements. The simulation results also demonstrate the promising application of the proposed method in studying the path planning of USV. This method can address the issues of real-time local obstacle avoidance of USV.

No abstract