An Improved Cloth Simulation Filtering Algorithm Based on Mining Point Cloud

Detection based on LIDAR point cloud data is an important step of unmanned driving perception system. The ground points, in the complex bumpy, are easily mistakenly detected as the ground object points mining area road by the traditional filtering algorithm. In order to ensure the normal operation of unmanned driving in the mining area, this paper proposes an improved cloth simulation filtering algorithm for unmanned driving system based on the cloth simulation filtering algorithm. We propose a cloth adaptive initialization method based on the mining area terrain, the secondary determination of the ground points and the ground object points, which combines with cloth simulation filtering Algorithm, to improve the ground detection ability of the cloth filtering algorithm in the complex mining area road. The experimental results show that through the optimized cloth simulation filtering algorithm for the ground detection, the extracted ground is more suitable for the actual scene than the original cloth simulation filtering algorithm, which can effectively improve the accuracy of the ground detection algorithm in the mining complex scene.