One-Step Identification of Robot Physical Dynamic Parameters Considering the Velocity-Load Friction Model

We propose a robot dynamic model to improve the accuracy of the identification, by introducing a friction model that takes into account the joint loads. Firstly, we analyze torque transfer in robot joints, assigning a physical meaning to motor inertia parameters. Then, we enhance the traditional friction model in identification by accounting for joint loads, presenting a new friction model with loads. Next, we employ a one-step method to directly identify both basic dynamic parameters and physical dynamic parameters of the robot. Experimental validation is conducted using a Rokea XMate3pro 7-DOF robot. Results demonstrate that our proposed dynamic model achieves higher accuracy in dynamic identification. It effectively describes the jitter phenomenon caused by motor torque when joints change the direction of motion. Furthermore, in identifying parameters for physical feasibility, our model outperforms traditional approaches by better fitting the dynamic of end joints.