Preisach model of nonlinear transmission at low velocities in robot joints

More advanced robotic applications demand novel control schemes and additional sensors to exceed the position accuracy of todays industrial robots. In that context more elaborate models of the transmission behavior of robotic joints open an avenue for advanced control system design with the objective to improve the dynamic accuracy of conventional robots without altering their mechanical structure. This paper introduces a novel advanced transmission model to capture the nonlinear behavior of elastic robot joints at low velocities. The stiffness of the robot joints is identified based upon experimental observations of the dynamic torque-torsion relationship. The prediction of the effective stiffness torque relies on the quasi-static torque equilibrium and the joint torsion observed through internal and external joint encoders. The distributed hysteresis behavior of the torsion as a function of stiffness torque is described by means of the Preisach operator which parameters are identified in a model based dynamic simulation. Finally, the accuracy of the model is verified experimentally by a comparison of the dynamic response between the actual system and the identified model. The proposed model enables the design of an advanced predictive control of robot joints, and thereby contributes to an improvement in the accuracy of standard industrial robotic systems.