Research And Implementation Of Intelligent Obstacle Avoidance Strategy For Unmanned Surface Vessel

With the development of the marine economy and the intensification of disputes over maritime rights in recent years,as a new type of surface craft,unmanned surface vessel(USV)is of great significance to the development of marine resource and the maintenance of marine safety.USV avoidance technology is a prerequisite for USV to achieve safe and autonomous navigation.In consideration of the actual obstacle avoidance scenes,this thesis focuses on the obstacle avoidance algorithms of USV,and designs simulation experiments and physical experiments.Due to the obstacle avoidance scenes of USV are complex and changeable,a single obstacle avoidance algorithm is difficult to deal with all scenes,this thesis proposes an intelligent obstacle avoidance strategy based on the principle of scene division.The intelligence of this strategy mainly reflects on scene switching autonomously,which will adopt obstacle avoidance algorithms with different constraint levels for different scenes,improve the success rate of obstacle avoidance,and has a certain versatility.First of all,based on the relative distance between the USV and the obstacle,this thesis designs a scene division principle to describe different scenes as different danger levels of the obstacle.For obstacles of different danger levels,this thesis designs three obstacle avoidance algorithms: "safe steering method","virtual potential field method",and "dynamic geometric method".The "safe steering method" aims to avoid the high-risk obstacles by using geometric parameter calculations to output an obstacle avoidance angle in fixed increments which outputs to the rudder angle control module of USV in each control cycle.By controlling the rudder angle of USV,this algorithm ensures the USV responds to the high-risk obstacles quickly and sets an angle threshold constraint to prevent the USV from continuously changing its heading in a short time and causing the rotation to ensure the rationality of the algorithm.The "virtual potential field method" aims to avoid the medium-risk obstacles by adopting the idea of establishing a virtual potential field,which defines new attractive field functions and repulsive force field functions,and outputs the resultant force direction as a given heading to heading control module.This algorithm makes the USV track a given heading to avoid the medium-risk obstacles,and introduces the concept of escape angle and virtual repulsion point,which overcomes the problems of local optimization and inability to avoid the obstacles facing each other,enhances the algorithm’s robustness.The "dynamic geometry method" aims to avoid the low-risk obstacles,by adding a safety circle outside of the obstacles,and then judging the collision risk between the USV and the obstacles in the sensing range,and outputting the tangent point between the USV and the safety circle of the obstacles which may collide.This algorithm makes the USV dynamically track the obstacle avoidance point to avoid low-risk obstacles.From the results of the simulation experiments and the physical experiments,the three obstacle avoidance algorithms designed in this thesis avoid the dynamic and static obstacles with different dangerous levels successfully,which verifies the effectiveness of the above algorithms.As far as the "safe steering method" is concerned,the USV’s heading changes rapidly in a short time when faces the high-risk obstacle,which verifies the algorithm can make the USV react quickly,has a certain real-time.Judging from the experimental results of the "virtual potential field method",the algorithm overcomes the shortcomings of the traditional artificial potential field method,and effectively improves the success rate of avoiding the medium-risk obstacles.As for the "dynamic geometry method",it successfully realized the avoidance of the low-risk obstacles.