Integration of IMU and Velodyne LiDAR sensor in an ICP-SLAM framework

Abstract: Simultaneous localization and mapping (SLAM) of an unknown environment is a critical step for many autonomous processes. For this work, we propose a solution which does not rely on storing descriptors of the environment and performing descriptors filtering. Compared to most SLAM based methods this work with general sparse point clouds with the underlying generalized ICP (GICP) algorithm for point cloud registration. This thesis presents a modified GICP method and an investigation of how and if an IMU can assist the SLAM process by different methods of integrating the IMU measurements. All the data in this thesis have been sampled from a LiDAR scanner mounted on top of an UAV, a car or on a backpack. Suggested modification on GICP have shown to improve robustness in a forest environment. From urban measurements the result indicates that IMU contributes by reducing the overall angular drift, which in a long run is contributing most to the loop closure error.