DyLESC: A Dynamic LiDAR Extrinsic Self-Calibration Method for Intelligent Driving Vehicles

Abstract

In this paper, we propose a novel LiDAR extrinsic parameter self-calibration method, DyLESC. With the driving trajectories obtained from GPS/IMU, a set of point cloud maps can be established by accumulating the LiDAR frames. The method introduces a function to evaluate the blurriness of the point cloud maps, which involves both geometrical features and point-number factors. By minimizing the blurriness of the point cloud maps, LiDAR extrinsic parameters are automatically calibrated. The method limits the frame number in a point cloud map to avoid the accumulated bumpy errors of GPS/IMU. For improving the robustness, the moving objects are removed from the LiDAR frames. We test the method both in a simulation environment and in the real world. The experiment result shows that DyLESC can accurately calibrate the LiDAR extrinsic parameters.