机器人激光检测手眼标定误差分析及优化

Analysis and Optimization of the Hand-eye Calibration Error in Robot

针对小型异构件激光检测精度不稳定、标定复杂、鲁棒性差等问题,建立基于标准球的eye to hand手眼标定模型,研究球心拟合、模型求解两个影响手眼标定精度的重要因素,提出一种基于加权的改进方案。首先,对扫描截面进行半径拟合及分析,获取标准球最优检测区域,降低标定数据采集误差;其次,用最小二乘法初步计算球心坐标,提取各个采集点误差;再次,根据误差设定阈值,通过TuKey权重函数对各采集点重新加权迭代求解,精确求解球心坐标;最后通过重加权最小二乘对标定模型求解,去除误差较大点对标定结果的影响。结果表明:优化方案相较于原有方案最大误差减少了57.2%,平均误差减少了49.7%,提高了标定的精度和抗干扰能力。

Aiming at the problems of unstable laser detection accuracy,complex calibration,and decreased robustness of small-scale heterogeneous components,an eye to hand calibration model based on a standard ball is established.Two important factors that affect the accuracy of the hand-eye calibration are studied on sphere center fitting and model solving,and a weight-based improvement scheme is proposed.Firstly,the radius of the scan section is fitted and analyzed to obtain the optimal detection area of the standard sphere and reduce the calibration data collection error.Secondly,the spherical center coordinates are preliminarily calculated by the least square method and the errors of each acquisition point are extracted.the threshold value is set according to the error,and each acquisition point is reweighted and iteratively solved through the Tukey weight function to accurately solve the spherical center coordinate.Finally,the calibration model is solved by weighted least squares to remove the influence of larger errors on the calibration results.The results show that the maximum error and average error of the optimized scheme are reduced by 57.2%and 49.7%respectively,and the calibration accuracy and anti-interference ability are improved.