基于位形相似性聚类的机器人参数标定研究

Research on Robot Parameter Calibration Based on Configuration Similarity Clustering

利用非几何参数误差引起的定位误差与机器人位形之间的相似性映射关系,通过聚类算法将机器人参数标定位形空间划分为多个子空间,并在各个子空间中完成参数标定;利用相似性关系将待补偿位形归类到对应的子空间,根据子空间的参数标定结果控制机器人运动,实现非几何参数误差引起的定位误差的有效补偿。以UR5机器人为标定对象进行试验,与普通标定补偿方法相比,采用所提方法标定补偿后的位置误差模的最大值和平均值分别减小了60.24%和66.62%。

Based on the similarity mapping between positioning errors caused by non-geometric parameter errors and robot configurations,the robot parameter calibration configuration space was divided into multiple subspaces by clustering algorithm,then the parameter calibration was completed in each subspace.The configurations to be compensated were classified into the corresponding subspaces by similarity relation,then the robot motions were controlled according to the parameter calibration results of the subspaces,and the positioning errors caused by non-geometric parameter errors were compensated effectively.An UR5 robot was used as the calibration object for the experiments.Compared with the ordinary calibration compensation method,the maximum and average of the positioning error mode of the proposed method are reduced by 60.24%and 66.62%respectively after calibration compensation.