基于M型标准块的视觉引导机器人系统手眼标定方法研究

针对线结构光传感器引导的机器人系统的手眼标定问题，提出了一种以M型标准块为标定物的方法。该M型标定物的两条平行的脊线作为约束，基于两条平行脊线的约束建立包含手眼关系、机器人运动学以及两条直线位姿参数误差的模型。首先基于定点约束求解手眼关系初值并以此为基础解算出直线位姿参数的初值，然后通过最小二乘法解算误差参数并补偿到模型中，不断迭代直至计算的误差参数小于阈值，最终得到最终的机器人手眼关系及运动学误差参数。为了验证标定方法的有效性，以某精加工平面为被测物，利用线结构光机器人系统对平面进行测量，得到平面点云；拟合最小二乘平面，计算点到平面距离的均方根值作为评价依据。分别对所述M型标准块和标准球两种方法进行了实验对比，结果表明，相较于标准球方法，所述M型标准块方法得到的均方根误差由0.152 mm减少到0.080 mm，均方根误差的标准差由0.043 mm减少到0.005 mm，其标定结果的精度及稳定性得到显著提高。

Research on hand-eye calibration method of linear structured light vision guided robot system based on M-type standard block

Aiming at the hand-eye calibration problem of the linear structured light sensor-guided robot system, a method using M-type standard block as the calibration object is proposed. Based on the constraints of the two parallel ridges of the M-type calibrator as constraints, a model including the hand-eye relationship, robot kinematics, and the error of the two linear pose parameters is established. Firstly, the initial value of the hand-eye relationship is solved based on the fixed-point constraint, and the initial value of the linear pose parameter is solved on this basis. Then, the error parameters are solved by the least squares method and compensated into the model, and the iterations are continuously performed until the calculated error parameters are less than the threshold. Finally, the final hand-eye relationship and kinematic error parameters of the robot are obtained. In order to verify the effectiveness of the calibration method, a certain finishing plane is used as the measured object, and the linear structured light robot system is used to measure the plane and obtain a plane point cloud; The least-squares plane is fitted, and the RMS of the distance from the point to the plane is calculated as the evaluation basis. The two methods of M-type standard block and standard ball were experimentally compared. The results show that compared with the standard sphere method, the RMSE obtained by the M-type standard block method is reduced from 0.152 mm to 0.080 mm, and the standard deviation of the RMSE is reduced from 0.043 mm to 0.005 mm. The accuracy and stability of the calibration results are significantly improved. Key words:Industrial robots; hand-eye relationship; kinematic; parameter identification; accuracy assessment.