针对车载三维激光雷达在工作前需要对安装外参数进行标定的问题, 综合考虑激光雷达的扫描光束不可见、多线式扫描等特点, 提出了一种激光雷达外参数的标定方法。首先对普通纸箱进行扫描, 以纸箱两个侧面和地面间的相互垂直关系作为约束, 采用随机抽样一致性算法(RANSAC)在获取的点云数据中拟合出三个平面的初始模型, 并通过旋转、平移步骤优化平面模型的拟合精度, 从最优模型中提取同名向量和同名点; 然后基于空间向量的三维坐标系转换模型, 对激光雷达的旋转和平移参数进行求解, 只需采集一次数据即可完成所有外参数的标定; 最后结合仿真和在标定后对室外环境三维重建的结果, 验证了算法的有效性。

Aiming at the problem that the external parameters of the on-vehicle 3D LIDAR needs to be calibrated before installation, a calibration method for LIDAR external parameters is proposed under the consideration of LADAR's invisible scanning beam and multi-line scanning. Firstly, scanning a ordinary carton, and the random sample consensus algorithm (RANSAC) is used to fit the initial model of three planes in the obtained point cloud data, with the vertical relationship between the two sides of the carton and the ground as a constraint. Then the rotation and translation steps are used to optimize the fitting precision of the plane model, and the corresponding vector and point are extracted from the optimal model. Next, solving the rotation and translation parameters of the LIDAR using the transformation model of 3D coordinate system based on space vector, calibration of all external parameters can be done by collecting data once. Finally, the effectiveness of the algorithm is verified by the simulation and the results of 3D reconstruction of the outdoor environment after calibration.