A kinematic parameter calibration method for industrial robot based on a novel designed 3D ballbar device

Abstract

The limited positioning accuracy of industrial robots restricts their application in high-precision manufacturing. Existing calibration methods based on coordinate measuring machines or laser trackers provide high accuracy but require expensive equipment, complex setup, and are inconvenient for on-site applications. Conventional ballbars are simple and convenient but mainly measure axial length variation between two balls. To address these issues, this study develops a three-dimensional (3D) ballbar device and proposes a robot kinematic parameter calibration method based on the center-to-sphere distance residual. The developed device consists of a 3D displacement-sensing base and an axial-measurement ballbar body, enabling small-displacement measurement in the X, Y, and Z directions as well as axial variation measurement along the line connecting the centers of the two balls. Chebyshev-polynomial fitting is used to establish the mapping between sensor voltage and displacement. Based on the Denavit–Hartenberg (D–H) parameter error model, the relationship between the center-to-sphere distance residual and the robot kinematic parameter errors is established. The LM algorithm is then used to identify the kinematic parameter errors and update the nominal robot model for error compensation. Experiments are conducted on a TB6-R10 six-axis collaborative robot. The results show that the maximum distance error is reduced from 2.9208 mm to 0.5195 mm, and the mean distance error is reduced from 1.0761 mm to 0.0967 mm after calibration. In addition, a discrete spatial path geometry validation using twenty validation points that were not used for parameter identification shows that the RMS adjacent path-segment length error decreases from 1.4420 mm to 0.2245 mm after compensation. The main contribution of this work is the integration of the developed 3D ballbar measurement device with a distance-residual-based kinematic parameter identification model, providing a practical and effective calibration method for improving the distance-based positioning performance and spatial geometric consistency of industrial robots.