一种改进的机器人视觉点云手眼标定方法

An Improved Hand Eye Calibration Method Based on Robot Vision Point Cloud

本文提出一种基于视觉点云测量系统的标定方法,设计了一套Eye-in-Hand手眼系统,包含手眼关系误差与机械臂结构参数的系统误差模型,在机器人工作空间中固定一些标定球作为实验对象,控制机器人改变不同的空间位姿,以不同位姿测量球心坐标。首先用机器人运动学公式得到初始的手眼关系;之后通过系统误差模型、矩阵运算、最小二乘法等不断迭代到达预期精度。在UR10工业机器人进行了系统标定实验,进行图像采集、点云处理、球心拟合等工作,完成Eye-in-Hand手眼系统标定实验,并使用激光跟踪仪完成精度验证。得出结论:标定前后机器人在局部空间运动的距离测量标准差由0.4963mm降低到0.1834mm,充分验证了改进方法的高效性、便捷性和实用性。

This paper proposes a calibration method based on the visual point cloud measurement system,and designs a set of Eye-in-Hand hand-eye system,which includes the system error model of the hand-eye relationship error and the mechanical arm structure parameters,and fixes some calibration balls in the robot workspace.The experimental object controls the robot to change different spatial poses and measure the coordinates of the center of the sphere in different poses.First,the initial hand-eye relationship is obtained with the robot kinematics formula;then,the expected accuracy is achieved through continuous iteration through the system error model,matrix operation,and least square method.A system calibration experiment was carried out on the UR10 industrial robot,image acquisition,point cloud processing,spherical center fitting,etc.,the Eye-in-Hand hand-eye system calibration experiment was completed,and the laser tracker was used to complete the accuracy verification.It is concluded that the standard deviation of the distance measurement of the robot movement in the local space before and after the calibration is reduced from 0.4963mm to 0.1834mm,which fully verifies the efficiency,convenience and practicability of the improved method.