论文部分内容阅读
现有的机器人参数辨识的方法所能提高的绝对定位精度有限,是由于关节和连杆柔度等非几何因素造成误差在空间中分布不均匀.为此,本文提出了一种变参数误差模型.由于各轴之间耦合,为了便于求解,提出采用空间网格来处理该变参数误差模型,同时提出了一套规则的参数辨识采样点选取方法.利用改进型的Levenberg-Marquardt迭代最小二乘法求出各网格对应的参数误差的全局收敛解.最后,利用激光跟踪仪在KUKA工业机器人上进行运动学标定验证补偿效果.验证结果表明:能将机器人的绝对定位精度平均值从0.901 mm提高到0.115 mm.
The existing methods of robot parameter identification can improve the absolute positioning accuracy is limited because of non-geometric factors such as joint and connecting rod flexibility caused by uneven distribution in space.Therefore, this paper presents a variable parameter error model Due to the coupling between the axes, in order to solve the problem, a spatial grid is proposed to deal with the variable parameter error model, and a set of rule parameter identification sampling point selection method is proposed.Using the improved Levenberg-Marquardt iterative least square method The global convergence of the parameters corresponding to each grid is obtained.Finally, the laser tracker is used to calibrate the kinematic calibration of KUKA industrial robot.Experimental results show that the average absolute positioning accuracy of the robot can be improved from 0.901 mm To 0.115 mm.