Parameter identification of robotic arms: A comprehensive study on friction modeling and precision optimization

Abstract

In this paper, a global identification approach, including both the friction parameter and inertia parameter, is proposed to improve the accuracy of the modeling of robotic arms. The modified Denavit–Hartenberg parameter method is used to capture the kinematics of the robotic arm. A complete dynamic model considering both the torque from collision and perturbation is obtained using the Lagrange method. Subsequently, the dry friction model and parameter identification methods are introduced, including deriving the regression matrix of the dynamic model and designing an optimal excitation trajectory. A friction extraction method is proposed by incorporating an enhanced friction model to enhance the overall accuracy. Two novel tangent function-based models are proposed to fit the measured friction torque, effectively identifying the friction parameters and resolving the nonlinear problem associated with the friction force. Comparisons between the proposed model and the traditional model demonstrate that the improved friction model not only enhances the fitting accuracy of the friction but also improves the identification of the inertia parameters. Based on the experimental results, a switch model criterion is proposed to select the best friction model for each joint and to mitigate the coupling effect among different joints. During this approach, various models of friction are provided for the selection. By integrating the optimized friction model into the robot arm dynamics, an overall nonlinear dynamic model with higher accuracy is obtained. Finally, the semi-definite programming algorithm is employed to obtain identification parameters that satisfy the physical constraints of the robotic arm.