Extrinsic Calibration for Low Resolution LiDAR-camera System Incorporating Printed Checkerboard and AprilTag

Extrinsic calibration of laser and vision sensors is a challenging problem. Traditional target-based methods for calibrating LiDAR and cameras have shown good results for high resolution LiDAR. But as low resolution LiDAR becomes more widely used, many algorithms have exposed issues in the extrinsic calibration of low resolution LiDAR. This paper investigates the extrinsic calibration method of low resolution LiDAR and camera using the checkerboard as calibration targets. Due to the sparsity inherent in low resolution point clouds, establishing accurate threedimensional (3D)-two dimensional (2D) or 3D-3D point correspondences throughout the calibration procedure is challenging. This paper introduces a pre-printed AprilTag as an auxiliary calibration object on the basis of using a checkerboard target. This assists in obtaining accurate point correspondences, providing additional constraints for the calibration process, and also offering sampling points at varying distances to avoid the derivation of local optima. Additionally, we use several methods to reduce the impact of noise in the target point cloud and improve the accuracy of our calibration results. Finally, we conducted quantitative and qualitative experiments in both real-world and simulated environments, demonstrating our algorithm exhibits high precision and can be applied to robotic vision applications in engineering.