Lightweight LiDAR Odometry Based on Intensity Optimization

SLAM attracts attention in the field of autonomous driving in recent years. However, facing with the outdoor complex environment such as uneven ground and uneven feature zone, most of the existing open source algorithms produce a large drift. In this paper, a lightweight LiDAR odometry is proposed based on the intensity optimization. Firstly, based on point cloud intensity and geometric information, a ground segmentation method is proposed for reducing odometry error. Then, an optimization strategy based on adaptive voxel filtering is used to improve the efficiency of the general feature extraction method in complex environment. Compared with the current excellent algorithm framework, experiments show that the proposed LiDAR odometry has higher positioning accuracy and faster running speed in the outdoor complex environment.