A Robust LiDAR-based SLAM for Autonomous Vehicles aided by GPS/INS Integrated Navigation System

We present a LiDAR-based SLAM aided by GPS/INS integrated navigation system which improves the robustness of real-time mapping and localization for autonomous vehicles. We add the pose of GPS/INS as a penalty term in the scan-to-map registration to obtain high precision localization and clear map simultaneously. When the scan-to-map registration failed, we use the pose of GPS/INS to create a new submap and apply the map-to-map registration to register between the submaps, which improves robustness. Because the map-to-map registration cannot run in real-time, we generate the global map and local map at different frequency levels, which match the requirements of global and local path planning respectively. The experimental results show that our method has more robust performance and better map quality than the existing methods.