Multi-LiDAR-Inertial Odometry Combined with Vehicle Kinematics for Autonomous Buses

The application of multi-LiDAR setup in autonomous buses is becoming increasingly widespread. However, challenges in real-world applications of multi-LiDAR SLAM remain largely unexplored. This paper introduces a multi-LiDAR-inertial odometry algorithm that utilizes vehicle kinematics to improve positioning accuracy and reliability. First, we introduce a sensor fusion method that synchronizes and integrates data from point cloud, IMU, and wheel encoder in the shortest possible time intervals, using IMU data to correct point cloud distortions. Subsequently, we developed a multi-sensor fusion odometry system combining multi-LiDAR, IMU, and vehicle wheel encoder data. Finally, leveraging the odometry's results, we generated a comprehensive global point cloud map. The evaluation of the proposed method in real-world scenarios demonstrates its capability to achieve high-precision odometry data and state-of-the-art performance.