Equation Discovery of KUKA 6-DoF Industrial Robotic Manipulator Dynamics with Backlash Using Lasso Model Selection Criteria

Abstract

The dynamic equations of robot manipulators are crucial for control system design and system analysis. Manipulator robots driven by electric motors with gearboxes often experience backlash and friction during movement, leading to inaccuracies in dynamic modeling. This study proposes functions to model backlash and friction forces to enhance the accuracy of 6-DoF robot manipulator dynamic modeling. Additionally, the nonlinear dynamic equations of the manipulator robot were derived using the LASSO model selection criteria and the Variable Segregation Algorithm (LMSCVS) method. The results demonstrate that the LMSCVS method can produce dynamic equations for robot manipulators with an improved accuracy of up to 95% when using the proposed library functions to model backlash and friction.