LiDAR-IMU Fusion for Distortionless Point Cloud Generation in Hybrid Solid-State LiDAR for 3D Mapping

LiDAR sensor generates accurate 3D environment representations by creating point cloud data, but self-motion distortion presents a significant challenge in solid-state LiDAR due to the restricted horizontal field of view. As LiDAR systems create point cloud data at frequencies ranging from 5 to 20 Hz, typically at 100 ms intervals at 10 Hz, this distortion arises due to the varying coordinate systems of each point in a frame, caused by the movement of the vehicle on which LiDAR is mounted. The paper discusses the underlying causes of this distortion, particularly in oscillating LiDAR systems, and presents a method to correct it through LiDAR and IMU (Inertial Measurement Unit) sensor fusion. By converting all points within a single frame to a consistent axis system using IMU data and timestamps, distortion can be effectively removed, leading to a more accurate, distortion-free point cloud. In this paper, the LiDAR-IMU fusion approach is employed to generate undistorted point clouds for precise perception and the creation of an accurate 3D point cloud map. The algorithm is tested in real-time at the TiHAN Autonomous Navigation Testbed in the IIT Hyderabad campus.