KINEMATIC ERROR MODELING AND CALIBRATION OF A SIX-DEGREE-OF-FREEDOM SERIAL ROBOT INCORPORATING JOINT-DEPENDENT ERRORS

Abstract In modern industrial applications, precise kinematic error modeling is crucial for achieving high positioning accuracy in serial robots. While the traditional Circular Point Analysis (CPA) method effectively describes geometric errors, such as link length inaccuracies and assembly misalignments, it overlooks joint-dependent errors caused by factors like bearing wear and gear transmission. This limitation reduces the end-effector's positioning accuracy during high-precision tasks. To address this, this paper analyzes various kinematic error sources and proposes a comprehensive error model that integrates constant geometric errors with joint-dependent errors. The model employs Chebyshev polynomials for fitting error terms and the Levenberg-Marquardt (L-M) algorithm for identifying polynomial coefficients, enabling accurate joint-dependent error characterization in a six-degree-of-freedom serial robot. The proposed method was validated using the TB6-R10 robot, with end-effector positions measured by a laser tracker. A comparative analysis with the CPA method shows that the proposed approach reduces the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) by over 90%, compared to 70% for the CPA method. These results demonstrate the model's ability to accurately capture error characteristics and establish a more precise mapping between joint positions and end-effector positions, outperforming traditional approaches.