Extrinsic Calibration of Dual LiDARs Based on Plane Features and Uncertainty Analysis

Solutions equipped with two or more light detection and ranging (LiDAR) systems have been applied in several fields (navigation, mapping, target detection, and others) to obtain increasingly detailed 3D environmental information. Accurate calibration of multiple LiDAR systems is necessary for integrating and using their data. This study improves a convenient and reliable method for calibrating the relative poses between 3D LiDARs. The planar surfaces of a scene are measured from different poses; then, the planes are extracted and constraints are constructed to solve extrinsic parameters. The factors influencing calibration uncertainty are also investigated. It is found that the uncertainty is closely related to the geometric structure of the plane observations, and the rotation dilution of precision and translation dilution of precision are defined to denote this effect. The influence of different types of spatial plane geometries on calibration errors is discussed, and optimal observation conditions are derived. Experimental results for simulation and real data show that the proposed method can achieve an accurate extrinsic calibration of LiDARs, and the point cloud registration errors after calibration are less than 2 cm. The analysis of the calibration uncertainty provides a valuable reference and guide for experimental design and error evaluation.