关键词: 地面无人车辆, 激光雷达, 负障碍物, 非结构化环境

Abstract: When Unmanned ground vehicles operate in unstructured environments, negative obstacles pose a significant threat to vehicle safety. However, in the existing research on negative obstacle detection, camera based detection methods are not well suited for unstructured environments with poor lighting conditions and complex backgrounds; LiDAR based detection methods are mostly designed based on the characteristics of mechanical LiDAR harnesses and have poor universality. Therefore, a highly compatible and scalable negative obstacle detection method has been proposed, which can adapt to different types of LiDAR and achieve high-precision negative obstacle detection. The method preprocesses point cloud data, extracts ground regions of interest, and performs point cloud pose correction; Adopts adaptive resolution polar coordinate rasterization technology to enhance the spatial representation capability of point cloud data; Designs a negative obstacle grid feature descriptor, combines with multiple features such as the hollow characteristics, height differences, and minimum height of point clouds, to extract potential negative obstacle regions; Introduces a multi-frame fusion strategy, and outputs high-precision negative obstacle surface occupancy range through map reprojection and Bayesian rule-based probability updates. The experimental results show that the proposed method is applicable to different scanning modes of LiDAR, and can effectively identify negative obstacle areas in complex unstructured environments, with good universality and detection accuracy.

Key words: unmanned ground vehicles, lidar, negative obstacles, unstructured environment