Modeling and Calibration of Posture-Dependent Denavit-Hartenberg Errors and Angular Positioning Deviations of a Six-Axis Robotic Manipulator

Prediction accuracy of a kinematic model is critical for offline compensation of volumetric errors of a six-axis robotic manipulator. Numerous scholars have employed kinematic models incorporating Denavit-Hartenberg (D-H) errors for offline compensation, and some have studied their enhancements by integrating stiffness parameters of rotary axes. However, previous works primarily focused on predicting deviations resulting from posture changes within the rotary plane. This article proposes a novel modeling scheme capable of capturing deviations both within and outside the rotary plane. A key idea is that, instead of explicitly measuring the 6 DOF stiffness parameters, the end effector trajectory is measured to quantify the position and orientation of the rotary axis average line, varied due to the posture of downstream axes, which are parameterized as posture-dependent D-H errors and posture-dependent angular positioning deviations. Measurement for the calibration of posture-dependent errors can be completed within one hour. A case study shows that the proposed posture-dependent kinematic model can significantly improve the prediction accuracy by over 36%, highlighting the importance of including posture-dependent errors.