Study On Intelligent Obstacle Avoidance Method Of Unmanned Surface Vehicles

As a kind of autonomous operation platform,unmanned surface vehicles(USVs)have good maneuverability and autonomy,can replace personnel to complete various marine operations.In order to improve USVs’ reliable for marine missions such as exploration and rescue,we must focus on the intelligent obstacle avoidance technology of USVs.At present,the intelligent obstacle avoidance methods of USVs still need to be further explored and improved.In this thesis,the intelligent obstacle avoidance of USVs is divided into global intelligent obstacle avoidance and local intelligent obstacle avoidance.This thesis first introduces research background and significance,current status,and analyzes development trend of the intelligent obstacle avoidance technology.Then this thesis introduces principle and related motion parameters of the intelligent obstacle avoidance,including situations of USV and the obstacle,relative movement situations,and collision risk degree.Aiming at the intelligent obstacle avoidance of USVs,the global intelligent obstacle avoidance of USVs under known environment model and the local intelligent obstacle avoidance under unknown environment model are studied respectively.Aiming at the intelligent obstacle avoidance of USVs,the grid method,ant colony algorithm,particle swarm optimization algorithm,genetic algorithm and other algorithms are studied,and some obstacle avoidance experiments are conducted.This thesis proposes an improved algorithm based on the rapid-exploring random tree(RRT)algorithm and conducts global obstacle avoidance experiments under actual water environment.The improved algorithm uses the grid method to process the actual water surface map,selects the sampling point according to the sampling probability,and uses the line segment theorem to improve the connection method of path nodes.The improved algorithm can generate a safe path with fewer nodes and a shorter length.Finally,compared with other improved rapid-exploring random tree algorithms and the proposed improved rapid-exploring random tree algorithm in the same environment,the global obstacle avoidance planning results are shown.The results show that the proposed improved rapid-exploring random tree algorithm can plan safe paths with shorter lengths and fewer turning nodes.Aiming at the local obstacle avoidance when encountering unknown obstacles during USVs’ driving,firstly,the obstacle avoidance motion model of USV is studied.Then the real-time relative movement of obstacles and USVs is considered,according to the rules of collision avoidance.According to different division situations and the collision risk degree,a local obstacle avoidance method based on geometric dynamic method is proposed.The simulation results of obstacle avoidance in different encounters show that the method can effectively realize the local obstacle avoidance in real time.And after the obstacle avoidance ends,the USV automatically resumes sailing and continues to travel to next path node.