Improved Cartographer Algorithm Based on Map-to-Map Loopback Detection

As the traditional Cartographer SLAM (Simultaneous Localization and Mapping) algorithm mainly uses Scan-to-Map method to finish loopback detection, which has limitation of single data matching and leads to loopback errors in scenes with few environmental features such as tunnels, corridors and garages. To solve this problem, an improved Cartographer algorithm based on map-to-map loopback detection is proposed in this paper. The algorithm matches the submap of the current frame with the previously generated submap, and takes advantage that the submap contains more information than the single frame to solves the matching limitation problem. In the process of detection, the algorithm improves the matching speed by preliminary matching and proposes the method of lazy decision to improve the real-time performance of the algorithm and improve the robustness of mapping. To verify the improvement effect of the algorithm improvement, the simulation results before and after the algorithm improvement are compared in Gazebo simulator. A trolley equipped with 16-wire LiDAR was established as the experimental platform, and a Raspberry Pi 4B was used as the algorithm running platform to test the indoor environment. The experimental results show that the mapping and localization effect is improved in the scenes with fewer environmental features, and meets the demand of higher accuracy (distance error is less than 0.05m, horizontal angle error is less than 0.25°).