摘要
传感器之间的外参标定是实现高效多传感器融合的前提,针对相机和激光雷达在无靶标和传感器时间不同步情况下的外参在线标定问题,文中提出了一种基于位姿插值和自然平面目标的相机与激光雷达外参在线标定方法。通过车辆在线获得的自然场景图像和激光点云信息中语义目标的对应信息,计算出相机和激光雷达的外参变换。实验表明,文中方法在通常的自动驾驶场景中可以高效提取到特定平面目标的2D掩模以及其对应的3D点云,并且即使传感器时间未同步,依然可以在线计算出较精确的外参变换结果。
The extrinsic calibration between different sensors is a prerequisite for efficient multi⁃sensor fusion.In order to solve the problem of online calibration of extrinsic parameters of the LiDAR and camera when there is no target and sensor time is not synchronized,this paper proposes an online calibration method of camera and LiDAR extrinsic based on pose interpolation and natural plane target.The extrinsic transformations of the camera and LiDAR are calculated only through the images of the natural scene and the LiDAR 3D point cloud information obtained by the vehicle online.Experiments show that the method in this paper can efficiently extract the 2D mask of a specific plane target and its corresponding 3D point cloud in the usual automatic driving scene.Even when the sensor time is not synchronized,the precise extrinsic parameter transformation result can still be calculated online.
作者
周媛媛
易鹏
马力
ZHOU Yuanyuan;YI Peng;MA Li(Guanzhou Urban Planning&Design Survey Research Institue,Guangzhou 510000,China)
出处
《电子设计工程》
2023年第16期43-46,51,共5页
Electronic Design Engineering
关键词
激光雷达
外参
在线标定
自然平面目标
位姿插值
LiDAR
extrinsic
online calibration
natural plane targets
pose interpolation
