RESEARCH ON SENSOR FUSION-BASED CALIBRATION AND REAL-TIME POINT CLOUD MAPPING METHODS FOR LASER LIDAR AND IMU

Abstract

Abstract. Simultaneous Localization and Mapping (SLAM), as one of the core technologies in autonomous driving, provides environment perception, decision-making, and path planning to ensure safe vehicle operation. This paper proposes a tightly-coupled fusion mapping method based on LiDAR and Inertial Measurement Unit (IMU). By introducing IMU, performing extrinsic calibration, and time synchronization with the LiDAR, the issue of mapping drift caused by inconsistent coordinate systems is addressed, leading to improved mapping accuracy. Furthermore, to compensate for IMU zero-bias errors, IMU preintegration is employed to calibrate the point cloud and optimize the initial values of the LiDAR odometry. A novel iVox voxel-based point cloud registration method is used to enhance registration efficiency without compromising mapping accuracy. Lastly, the Iterative Extended Kalman Filter (IEKF) is incorporated into the back-end state estimation to propagate and correct the estimated state using IMU data, enabling precise state estimation and map updates. This research is of significant importance in addressing the challenges of low mapping accuracy and real-time performance in single-sensor SLAM, providing an effective solution for environment perception and path planning in autonomous driving systems.