Abstract:

Aiming at the calibration of LiDAR and camera of driverless technology, this paper proposes an automatic calibration method based on 3D feature point sets of two sensors data. First, ArUco tags provide transform between the camera and the cardboard marker, the dimensions and location of the cardboard is known, the location of the corners can be calculated. RANSAC algorithme is used to fit edges on the points from the LIDAR, their intersection is calculated as corners, thus the two sets of point correspondences are found. Then, Kabsch algorithme is used to obtain the optimal calibration parameters between the LiDAR and the camera. Finally, we compare the optimal calibration parameters obtained from calculated against measured values using tape by a human to test the feasibility of our method. For further determine the accuracy of our method, we use our method to calibration of two cameras, the accuracy of the calibration parameters obtained is evaluated by the quality of the fusion point cloud data. The experimental results show that the automatic calibration method based on 3D feature point sets can realize the accurate solution of calibration parameters between LiDAR and camera, this method could be used to automatic calibration of two or more cameras.

Key words: calibration, ArUco tag, RANSAC algorithme, Kabsch algorithme, fusion point cloud