Framework for Static and Dynamic Friction Identification for Industrial Manipulators

Abstract

Even if friction modeling and compensation is a very important issue for manipulators, quite simple models are often adopted in the industrial world to avoid too heavy solutions from the computational point of view, and because of the difficulty of finding and identifying a model applicable in any motion condition. This article proposes a general framework for friction identification for industrial manipulators with the goal of solving the previous problems through: first, a complete procedure managing all the steps from data acquisition and model identification up to the generation of the code for the implementation into the robot software architecture, second, the possibility of adopting static or dynamic models of different complexity, and third, the development of some modifications in the dynamic friction model so to achieve a reliable friction torque estimation at any velocity and acceleration regime, avoiding unfeasible peaks and overestimation. The results of experimental tests carried out for different manipulators prove the validity and generality of the proposed friction model and identification procedure.