基于视觉标记板的自动驾驶车辆激光雷达与相机在线标定研究

An On-Line Calibration Method Between Laser and Camera in Autonomous Driving Vehicle Based on Visual Marker Board

针对传统的高清相机和激光雷达外部参数手动或半自动化标定程序繁琐和耗时问题,提出一种适用于地面无人平台车载激光雷达和相机在线自动标定的方法,以安装标记板(Marker)的移动车辆为标定目标,实现车辆的快速识别和位姿估计。同时获取在两种传感器中得到的位姿估计结果,利用经典的定向求解方法计算得到相机与激光雷达之间的刚性转换关系,并在真实城区环境中利用最优评估优化减小标定误差,试验结果证明了所提方法的准确性与可靠性。

It’s tedious and time-consuming to calibrate manually or semi-automatically external parameters of traditional HD camera and laser radar.To solve this problem,a novel method is presented to calibrate the laser and camera automatically for the unmanned ground vehicles.The method uses a vehicle mounted with a marker as the calibration object to make the target to be estimated rapidly.Pose estimation result is obtained from 2 sensors,and the classical oriental solution is used to calculate the rigid conversion relation between camera and laser radar,the optimal evaluation optimization is used to reduce calibration error in real urban environment,test results prove accuracy and reliability of the proposed method.