Dual-Weight Optimized Anisotropic Kriging Compensation for Industrial Robots in Heterogeneous Error Fields With Drilling Verification

Abstract

Compared to kinematic calibration, spatial compensation methods can address both kinematic and non-kinematic errors. However, existing spatial compensation approaches often overlook the anisotropic characteristics and random error distribution inherent in industrial robots, resulting in limited accuracy. This paper proposes a dual-weight optimized anisotropic Kriging compensation method to enhance robotic positioning accuracy. First, the heterogeneous error field is analyzed through differential error equations, establishing direction-specific variograms for the X, Y, and Z axes. To mitigate residuals from variogram fitting, a dual-weight function integrating distance-based and covariance-based indices is formulated, refining error weight allocation. The positioning errors in each axis are then predicted using this optimized function and mapped to the joint space via inverse kinematics for error compensation. Experimental validation on an ABB IRB4600-60/2.05 robot demonstrates that the proposed method achieves maximum and mean prediction errors of 0.144 mm and 0.103 mm, respectively, significantly outperforming four benchmark methods in accuracy. A drilling application on aircraft parts further validates its practicality, reducing maximum and mean positioning errors from 3.222 mm and 2.537 mm to 0.543 mm and 0.254 mm, respectively, confirming the method's practical effectiveness.