输入参数优化的机器人手眼标定改进方法  

Robot hand-eye calibration improvement method for input parameter optimization

作　　者：侯阳阳 苏本磊 HOU Yang-yang;SU Ben-lei(National Ethnic Affairs Commission Key Laboratory of Electronic Information Engineering,Southwest Minzu University,Chengdu 610041,China)

机构地区：[1]西南民族大学电子信息工程国家民委重点实验室,四川成都610041

出　　处：《西南民族大学学报（自然科学版）》2024年第5期554-561,共8页Journal of Southwest Minzu University(Natural Science Edition)

基　　金：四川省科技厅重点研发计划项目(2022YFG0377);西南民族大学中央高校基本科研业务费专项(2020NQN05)。

摘　　要：由于相机位姿误差以及机器人运动解算误差导致标定结果具有较大误差,为了提升机器人定位与抓取的精度,提出了一种具有误差补偿和图像校正的手眼标定方法.基于对偶四元数理论来描述螺旋运动,用奇异值分解法求解手眼变换矩阵.再通过对非线性相机成像模型进行研究后最小化相机的重投影误差,得到优化后的相机外参,同时根据机器人运动学模型对机器人运动误差做补偿.最后搭建机器人单目视觉实验平台,将该方法与对偶四元数法和Tasi两步法进行对比分析,实验结果表明,该方法在XYZ三个方向上的坐标误差均值为0.46,标准差为0.25.在标定精度和稳定性上均优于其他两种方法,具有一定的理论实践意义.Due to the camera pose error and the robot motion solution error,the calibration result has a large error.In order to improve the accuracy of robot positioning and grasping,a hand-eye calibration method with error compensation and image correction was proposed.Based on dual quaternion theory to describe spiral motion,the hand-eye transformation matrix was solved by singular value decomposition method.After studying the nonlinear camera imaging model,the reprojection error of the camera was minimized,then the optimized external reference of the camera was obtained,and the robot motion error was compensated according to the robot kinematic model.Finally,the robot monocular vision experimental platform was built,and the method was compared with the dual quaternion method and the Tasi two-step method.The experimental results showed that the coordinate error of the method in the three directions of XYZ was 0.46 and the standard deviation was 0.25.In terms of calibration accuracy and stability,it is better than the other two methods,and has certain theoretical and practical significance.

关 键 词：机器人 手眼标定 对偶四元数 图像校正 误差补偿 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]