Passivity Guaranteed Dynamic Friction Model with Temperature and Load Correction: Modeling and Compensation for Collaborative Industrial Robot

In this paper, a new comprehensive dynamic friction model for a collaborative industrial robot joint that considers the velocity, temperature, and load torque is proposed. The variation of load-dependent friction among the four-quadrant operation depending on the sign of load-torque and speed is studied. The new model's passivity property is analyzed to obtain a physically meaningful and experimentally identified friction model. A sufficient condition is presented in terms of a simple algebraic inequality involving the parameters of the model. The model parameter identification procedure and validation of model effectiveness are demonstrated experimentally on a commercial collaborative robot manipulator. Moreover, the proposed friction model's benefits are demonstrated in two different robot applications: friction compensation and direct teaching (smooth lead-through programming) applications. Significant tracking performance improvement in the root-mean-square errors up to 76% was achieved with the proposed friction model compared to the uncompensated cases in the friction compensation application. In the direct teaching application, the new model, which precisely estimates joint friction, results in a significant decrease in interaction forces up to 66%. These experimental results validate the performance of the proposed friction model.