Research On USV Dynamic Obstacle Avoidance Planning Based On LiDAR Angle Correction

The ocean is the cradle of life,with vast areas and abundant resources.It is a vital transportation route and strategic location for countries around the world.There are still many resources in the ocean that have not been explored and discovered,and intelligent unmanned surface vehicles(USVs)have become an important way to explore ocean resources.They are also a key focus of technological competition among countries in the field of ocean exploration.Currently,intelligent USVs face significant challenges,such as the urgent need for rapid and accurate environmental perception.Traditional perception solutions based on ultrasonic ranging and camera image capture not only make USVs expensive but also result in unsatisfactory environmental perception effects in harsh air environments or when detecting targets at slightly longer distances.On the other hand,during the cruising process,USVs will inevitably encounter obstacles ahead.Therefore,having a stable and effective obstacle avoidance path planning strategy is the basis for USVs to better complete exploration tasks.Although existing USV obstacle avoidance algorithms have been widely studied in theory,practical implementation in real water environments is relatively scarce.This paper proposes an automatic angle correction method for point cloud data to address the distortion problem in laser detection caused by the changing orientation of the ship-mounted 2D Li DAR during environmental perception for unmanned surface vehicles.The proposed method references the current attitude angle data of the unmanned surface vehicle and the returned obstacle point cloud data,and establishes corresponding mathematical geometry models based on the category of obstacles.Finally,the true width and position direction of obstacles in the current environment are calculated and restored.Ground experiment results demonstrate that the 2D Li DAR point cloud data after angle correction can reduce errors by approximately 5% to 15% at a certain distance compared with the original point cloud data,thereby improving the accuracy of obstacle detection data for unmanned surface vehicles.In view of the insufficiency in the traditional algorithm of obstacle avoidance,this paper puts forward an algorithm of obstacle avoidance decision-making based on Windows partition.The algorithm evenly divides the perception window of the unmanned boat into several sub-regions,comprehensively considers the motion characteristics of the surface unmanned boat,the estimated turning radius and the obstacle avoidance decision threshold,evaluates the generation value of each sub-region,and then selects the region with the best generation value from the sub-region.Finally,the center direction of the optimal region is used as the next navigation direction of the unmanned boat.The obstacle avoidance algorithm proposed in this paper can be well combined with the lidar detection model,overcome the problems existing in the traditional obstacle avoidance algorithm,and reflect the better obstacle avoidance effect.Compared with classical obstacle avoidance algorithms,the proposed obstacle avoidance algorithm has more advantages in physical memory occupancy,algorithm complexity and portability.Experimental results show that this algorithm has better performance in the aspects of path smoothness and real-time obstacle avoidance.The obstacle avoidance scheme proposed in this paper combines angle correction of 2D Li DAR data and a path obstacle avoidance decision method based on window segmentation,and corresponding obstacle avoidance tests were conducted in real water areas.The experimental results show that the angle correction method can reduce the data error generated by the 2D Li DAR detection to a certain extent.The path obstacle avoidance decision algorithm based on window segmentation can plan a safe and smooth obstacle avoidance route,and the actual obstacle avoidance trajectory of the USV during the pond test is similar to the planned obstacle avoidance route by the algorithm,demonstrating good obstacle avoidance performance.