ENVIRONMENTAL INFORMATION PERCEPTION OF LIDAR FOR AUTONOMOUS VEHICLE NAVIGATION PLANNING

Traditional lidar perception algorithms provide limited information and lack target-level details. Therefore, this paper proposes a method that integrates prior environmental information and real-time perception data to construct a local environment model, which guides vehicle navigation planning. Using prior environmental information, the height difference algorithm is employed to reduce the dimensionality of the 3D point cloud data and construct a 2D obstacle grid map. Furthermore, prior feasible regions are used to preprocess the point cloud information, enhancing the speed of map construction. Based on the grid map, a 2D obstacle detection and tracking module is formed, utilizing the distance threshold clustering algorithm, feature extraction, and ROI (Region of Interest) detection. The information generated by this module is incorporated into the grid map, and a perception map suitable for autonomous vehicle navigation and planning tasks is finally constructed. Through simulation and real vehicle experiment, it is confirmed that the algorithm proposed in this paper is capable of constructing the 2D obstacle grid map while ensuring real-time requirements, thereby providing sufficient perception information for navigation planning.