基于刚度模型和高斯过程回归模型的重载工业机器人分步标定方法

A hierarchical calibration method for heavy-duty robot with stiffness model and Gaussian process regression model

针对串联工业机器人由于关节柔性导致的重载下绝对定位精度较低的问题,提出了一种机器人定位误差分步标定方法。采用局部指数积模型对机器人进行几何误差标定。提出了一种基于建模和机器学习的非几何误差标定方法。在该部分中,首先建立了机器人的刚度模型对非几何误差中最主要的变形误差进行标定,然后采用数据驱动的高斯过程回归(GPR)模型对残余误差进行标定。实验结果表明,该方法可以有效提高机器人带载下的绝对定位精度,并且具有位置精度不随载荷变化而产生明显波动的优点。

Aiming at the problem of poor absolute positioning accuracy under heavy loads caused by joint compliance of serial industrial robots,a hierarchical calibration method for heavy-duty industrial robots is proposed.The local product-of-exponential model is used to calibrate the geometric error of the robot.A non-geometric error calibration method based on modeling and machine learning is proposed.In this part,the stiffness model of the robot is first established to calibrate the deformation error,which is a prominent part of the non-geometric error.Then,the residual error is calibrated with the data-driven Gaussian process regression(GPR)model.Experimental results show that this method can effectively improve the absolute positioning accuracy of the robot under load,and the positional accuracy does not fluctuate obviously with the change of load.