基于激光点云特征的LiDAR-GNSS/IMU联合自动标定方法

LiDAR-GNSS/IMU Automatic Joint Calibration Method Based on Laser Point Cloud Features

车载多线激光雷达与GNSS/IMU系统间的数据融合可以提高智能驾驶定位系统的精度,两传感器间外部参数的标定是完成数据融合的前提.针对车辆传感器间外部参数手动测量困难、标定自动化程度低的问题,提出了一种不依赖标志物的自动标定方法,首先构建线、面两类特征点云地图并采用闭环约束减小累积误差,对两类点云地图分别提取面特征体素和线特征聚类,并将每一激光帧中的激光点与面特征体素和线特征聚类进行关联,结合基于运动标定的约束,构建最小二乘问题优化求解外部参数,最后通过实车实验验证了所提出算法的有效性和鲁棒性.

The data fusion between onboard Light Detection and Ranging(LiDAR)and GNSS/IMU can improve the accuracy of the localization system in intelligent driving,and the calibration of external parameters between two sensors is the premise of data fusion.It is hard to measure external parameters between onboard sensors manually and the the calibration of automation degree is lower.In this paper,an automatic calibration method that does not rely on markers is proposed.First,the line feature map and surface feature map are constructed,respectively,and closed-loop constraints are used to reduce the accumulated error.The laser points in each frame are matched with surface feature voxels and line feature clusters extracted from two feature maps.The matching errors are integrated with the constraints based on motion calibration.Then a least squares problem is built to optimize the external parameters.Finally,we evaluate the efficiency and robustness of the proposed algorithm by conducting experiments with real vehicles.