Research On Water Surface Target Recognition And Path Planning Of Unmanned Surface Vehicle

Unmanned surface vehicles(USV)have the advantages of autonomy,intelligence and low cost,which can improve the efficiency of task execution,reduce personnel risk and save cost,while the autonomous navigation technology is the basis of autonomous task execution for USV.This paper focuses on the autonomous navigation technology of USV,which mainly includes the following three key elements: path planning in static environment,recognition and detection of surface targets required for local obstacle avoidance,and local obstacle avoidance in dynamic environment.This paper proposes an improved surface target recognition algorithm,global and local path planning algorithms for the perception and path planning problems in autonomous navigation,and verifies the effectiveness of the improved algorithms through simulation and experiment,which can provide technical support for the autonomous navigation of USV.The following are the main works of the paper.(1)A global path planning method for USV based on the improved BTO-RRT algorithm is proposed,which can plan a smooth path with short path length and minimum energy consumption for USV in a static environment.Firstly,to address the problem that the original BTO-RRT fails to sample and the third spline fitting in the case of large step length,an adaptive step length sampling strategy based on the artificial potential field is proposed in this paper,the adaptive step effectively solves the above problems,and the improved algorithm can generate optimized trajectories with shorter paths,taking into account the completeness of generated paths and the speed of path search.Then,in order to save the energy of the USV when navigating on the water,the minimum Snap trajectory optimization method with Corridor constraint is used instead of the cubic spline fitting method of BTO-RRT algorithm,and the optimized trajectory can reduce the energy consumption of USV in autonomous navigation.Finally,experiments were conducted in the river area using the built USV platform to verify the feasibility of the improved algorithm.(2)A water surface target detection method based on sea-sky line division is proposed to provide USV with water surface visual information in an unknown environment.The necessary condition for the USV to realize local collision avoidance in dynamic environment is to obtain surface target information.In this paper,firstly,to address the problems of low detection accuracy and poor real-time performance of traditional Hough transform-based sea-sky line detection method,a sea-sky line detection method based on Canny edge detection and line segment detection algorithm is proposed,which can quickly and stably identify sea-sky line and delineate navigable areas for USV.Then the YOLOv5 s target detection algorithm is used to identify the surface targets in combination with the above-mentioned sea-sky line delineated region of interest.After testing and comparing,the target detection algorithm based on sea-sky line delineation has higher detection accuracy,and finally,it is shown through experiments that the algorithm performs well in the application of practical scenarios.(3)A local collision avoidance method based on the improved dynamic window approach is proposed to make the USV reasonably avoid obstacles under the dynamic environment in accordance with the maritime rules.Firstly,collision avoidance timing is determined using collision hazard evaluation index,then collision avoidance waypoints are set to guide the USV to avoid obstacles,and finally,constraints are made on the heading of the USV during collision avoidance to ensure that the USV complies with the maritime rules.After simulation and verification,the improved algorithm is able to avoid obstacles under three situations of overtaking,head on and crossing from the right in strict accordance with the rule constraints.