A Point Cloud Mapping Algorithm for Outdoor Scenes Based on LiDAR

In order to solve the problems of poor mapping effect, insufficient robustness and large accumulated error of laser odometer in outdoor scenes, this paper proposes a point cloud mapping algorithm for outdoor scenes based on LiDAR. For outdoor scenes with sparse point clouds, the method of splicing point clouds is used to obtain sufficient point clouds; the surface feature point clouds are extracted by voxelization and curvature and RANSAC-based methods to improve the computational efficiency and robustness of the algorithm; the point cloud matching is completed by using the point to plane-iterative closet point (ICP) algorithm, which effectively improves the accuracy and efficiency of feature matching; a global map is constructed, and the accumulated error of odometer is reduced by factor graph optimization and adding GNSS constraints. The algorithm is verified using the KITTI dataset and campus environment data. The experimental results show that the proposed algorithm has better mapping accuracy than the current mainstream algorithms A-LOAM and LeGO-LOAM while meeting the real-time requirements, and has good robustness and environmental adaptability.