Research On Unmanned Surface Vehicle Collision Avoidance Path Planning Algorithm

Under the development of the new era,the strategic development direction of countries and research institutions around the world has gradually spread on the ocean.Unmanned surface vehicle technology is a hot spot for development.In the civil and military fields,compared with traditional ships,unmanned surface vehicles have high potential for improvement in both safety and economy.The unmanned surface vehicle's collision avoidance path planning capability is one of its key technologies and an important manifestation of intelligence.Based on " unmanned surface vehicle system project ",this thesis carried out the following aspects of research.(1)Through the overview of the path planning algorithm for unmanned surface vehicles,combined with the research at home and abroad,the analysis of unmanned surface vehicle collision avoidance path planning can be divided into global path planning and local path planning,and the characteristics of each algorithm are studied and analyzed.Finally,PRM algorithm and speed obstacle avoidance algorithm are selected as the research basis of this thesis.(2)Research on the global path planning of unmanned ships.Through the research of the original PRM algorithm,the steps of the learning phase and the query phase are analyzed,and it is found that there are three aspects to be improved: The complexity of the connection between the paths;The distribution of the sampling points in the obstacle is not sufficient;The path Not smooth.In view of the above deficiencies,the following improvements are made: Path optimization based on node reconnection;Enhancement of sampling points around obstacles;Optimization of secondary B-spline curve for paths.According to the above improvements,the algorithm is more suitable for the path planning of unmanned ships,and then the simulation calculation is carried out in three simple,complicated and narrow channels.The results show that the improved algorithm can plan a short distance,safe and effective.path.(3)According to the environmental construction problem of the unmanned ship,the environment of the unmanned ship is automatically constructed according to the aerial image of the drone or satellite.Since the electronic chart and the AIS system sometimes cannot cover the environmental information when the unmanned ship performs the task,in order to ensure the intelligentization of the unmanned ship,the U-net neural network structure is used to automatically identify the navigable area and the non-sailing area.Redundant recognition is used.In the case of ensuring accuracy,the program can identify all obstacle areas,and then extract the contour coordinate information of the obstacles for the environment construction.(4)Combined with the rules in the Collision Avoidance Convention,the local collision avoidance of unmanned ships is studied.Firstly,the risk of collision avoidance of unmanned ships is judged,and then the speed of unmanned ship and obstacle ship is taken as the basic parameter.According to the relationship between speed and position and ship safety radius,the speed field where collision risk will occur is determined.The minimum change in speed u is found in the field,and the speed of the unmanned ship is changed.Finally,combined with the Collision Avoidance Convention,the simulation calculations are carried out under the three conditions of encounter,cross encounter and chase.The simulation results show that the unmanned ship and the obstacle ship successfully carried out obstacle avoidance under various conditions.