Velocity- and Load-dependent Joint Friction Identification for a 6-DOF Industrial Robot

Industrial robots (IRs) are characterized by a lower stiffness compared to machine tools, leading to disadvantages when it comes to processes such as machining. At the same time, IRs possess several advantages, including a lower initial investment and higher flexibility. In recent years, several studies have deepened the possibility of improving the process accuracy through model-based robot control. One of the most important factors is friction, which can be modeled through analytical or data-driven models, and integrated into the robot control as a compensatory signal. In preceding works, different static, dynamic, and load-dependent friction models of the first and the second axes of the six degree-of-freedom (DOF) robot MABI MAX 100 have been identified. In this work, the friction models of the remaining four axes will be identified while considering the axis-dependent behavior of the system. The research presents substantial differences between the large and small robot axes, from the obtained data and the performed identification. This paper therefore enables a comparative analysis of these models' effectiveness in characterizing frictional behaviors across different axes and varying conditions. Our results offer a good reference for friction identification of industrial robots and provide an experimental basis for subsequent theoretical research.