Efficient Kinematic Calibration for Industrial Manipulators based on a Simplified POE model

The Product-of-Exponential (POE) model simplifies kinematic modeling by establishing direct relationships between kinematic and end-effector errors using base coordinate and tool coordinates. However, It contains numerous parameters and high latitude, which may lead to reduced efficiency and precision in the case of limited test data. This paper introduces the Simplified Effective Kinematic Calibration (SEC) technique, combining the Simplified Product-of-Exponential (SPOE) model, Pre-processing Sequential Quadratic Programming (PSQP), and Gaussian Process Regression (GPR) to enhance calibration accuracy and efficiency. The SEC technique reduces parameter identification dimension, identifies coordinate transformation, compensates for geometric errors, and uses GPR for non-geometric error compensation. Experiments show that SEC demonstrates better positioning accuracy and efficiency than traditional methods.