S4OM: A Real-Time Lidar Odometry and Mapping System Based on Super4PCS Scan-Matching

Abstract

For long-distance accurate localization and mapping of mobile robots in outdoor environment, a new real-time lidar odometry and mapping system (S4OM) is proposed in this paper, it divides the complex problem of simultaneous localization and mapping into localization problem and mapping problem, and then uses two algorithms to deal with them. The localization algorithm outputs the location information at a high frequency, it achieves coarse-to-fine matching by combination the Super4PCS algorithm with ICP algorithm. The mapping algorithm corrects the location information and builds map at a low frequency, it uses NDT matching method to match the current keyframe with the local map. The S4OM system formed by the above two algorithms can achieve good balance of instantaneity and accuracy without GPS or other inertial navigation aids, and has good robustness and environmental adaptability, in the case of large displacements and low-overlap of point clouds it can still achieve good results. The method in this paper has been tested with a large number of datasets and has been submitted for evaluation on KITTI odometry benchmark. The results indicate the effectiveness and feasibility of the method.