车载三维激光雷达外参数的分步自动标定算法

Step-By-Step Automatic Calibration Algorithm for Exterior Parameters of 3D Lidar Mounted on Vehicle

三维激光雷达外参数标定是智能车通过激光雷达感知环境的基础,针对常见标定方法实施繁琐、精度低,以及依赖其他传感器的问题,提出了一种分步自动标定算法。第1步对地面点云进行拟合得到地面方程,构造水平度函数,通过粒子群优化(PSO)算法优化水平度函数完成对激光雷达俯仰角、横滚角和纵向位移的标定;第2步标定以第1步标定的完成为基础,在车辆沿直线行驶过程中采集多帧含有同一标定杆的激光点云,通过聚类得到标定杆聚类中心,然后在二维平面内对多帧同一标定杆的聚类中心进行直线拟合,根据直线斜率计算航向角。结果表明,所提算法的精度可达10^(-5)数量级,耗时0.5s,极大地提高了标定精度和效率,能满足实际工程的使用需求。上述两步自动标定算法由程序自动完成,并且不依赖于其他传感器即可得到高精度的标定结果。

The calibration of exterior parameters of a 3 Dlidar is the basis of an intelligent vehicle perceiving environment through the lidar.Aiming at the major problems of common calibration methods including cumbersome implementation,low precision,and relying on other sensors,we propose a step-by-step automatic calibration algorithm.The first step is to get the ground equation by fitting the ground point cloud,construct the levelness function,and complete the calibration of the pitch angle,roll angle,and longitudinal displacement of the lidar by optimizing the levelness function with the particle swarm optimization（PSO）algorithm.Based on the completion of the first step calibration,the second step calibration needs to collect multi-frame point clouds containing the same calibration pole during the vehicle moving along a straight line.The calibration pole is clustered to get the clustering centers.The linear equation with multi-frame clustering centers is fitted in the two-dimensional plane,and the yaw angle is calculated according to the slope of the fitted straight line.The results show that the precision of the proposed algorithm can reach the magnitude order of 10^-5,and the algorithm takes 0.5 s.The calibration precision and efficiency are greatly improved,and actual engineering demands can be satisfied.The two steps are done automatically by the program,and the calibration results with high-precision can be acquired without other sensors.