摘要
针对电力作业过程中机器人对作业工具检测定位问题,提出一种融合单目相机图像与激光雷达数据的识别定位方法。识别作业工具上的二维码标记,得到其在相机坐标系下的位姿,以单目相机与激光雷达联合标定为基础,通过坐标变换得到该标记在激光雷达坐标系下的位姿;以该位姿计算标记平面的法向量为基准,通过过滤、聚类算法得到标记板的点云;融合两个标志板点云的特征计算出作业工具在激光雷达坐标系下的位姿。经过户外高强度光照条件下的实验验证表明,该方法可以有效地消除相机内参标定以及与雷达联合标定时产生的误差,能够准确地检测并定位电力作业工具在机器人操作空间中的位姿。
For the detecting and positioning errors of operating tools by robots during the process of power operation,this paper proposes a sensor fusion method that integrates monocular camera image and LiDAR point cloud data.ArUco markers on the operating tool are detected,and their poses under the camera coordinate system are obtained.Based on the joint callibration between camera and LiDAR,the poses of marker planes under the LiDAR coordinate system are obtained through coordinate transformation.With the normal vector of the plane as the baseline,and the point clouds of the marker plates are obtained by the filtering and clustering algorithm.And the characteristics of the two marker plate point clouds are fused to calculate the pose of the operating tool under the LiDAR coordinate system.The experiment under outdoor high-intensity light conditions verifies that the proposed method can effectively eliminate the error caused by the camera intrinsic callibration and extrinsic calibration between camera and LiDAR,accurately detect and locate the pose of the power operating tool in robot operation space.
作者
翟世雄
白玉苓
张黎明
胡益菲
张雨蔚
任书楠
ZHAI Shixiong;BAI Yuling;ZHANG Liming;HU Yifei;ZHANG Yuwei;REN Shunan(State Grid Tianjin Electric Power Company,Tianjin 300010,China;Beijing Guodian Futong Technology Development Co.,Ltd.,Beijing 100070,China)
出处
《机械制造与自动化》
2025年第3期306-310,314,共6页
Machine Building & Automation
基金
国网天津市电力公司科技项目(滨海-研发-2023-01)。
关键词
三维目标检测
多传感器融合
点云过滤
点云聚类
电力机器人
3D object detection
multiple sensor fusion
point cloud filter
point cloud cluster
electric robots
作者简介
第一作者:翟世雄(1990-),天津人,男,工程师,本科,研究方向为电力作业装备检测定位,sheksi12@163.com。
