A Novel Robot Positioning Accuracy Calibration Method Integrating Levenberg-Marquardt and Differential Evolution Algorithm

Industrial robots are widely used in processes such as milling of aircraft skins, where their absolute positioning accuracy plays a crucial role in determining the quality of the machining results. A major factor influencing this accuracy is the robot's kinematic parameters. This paper introduces a refined kinematic error model, free of redundant parameters, built upon the MD-H model and enhanced using the Hash method. It also presents an innovative calibration technique that integrates the Levenberg-Marquardt (LM) algorithm with the Differential Evolution (DE) algorithm. First, the MD-H model is constructed, followed by the development of the error model, with redundant columns eliminated through the Hash method. The LM algorithm is then employed for preliminary parameter identification, generating initial estimates that are further refined using the DE algorithm. By applying this combined approach, the KUKA KR300 R2700 robot achieved an absolute positioning accuracy of 0.3358 mm post-calibration, meeting the stringent precision demands required for milling aircraft skins.