Experimental identification of dynamic friction parameters with the intention of precision optimal control of model based robotic systems

Abstract

In this paper, the difference in the position and the velocity of input and output angles is considered as friction and, its value is estimated using the methods of least-squares error identification with two direct and recursive approaches. The selected techniques, considering simplicity and reduction of computational volume and high execution speed, have a good performance in determining the parameters of friction model, and as a result, achieve a relatively accurate model close to the actual model of system. Then, the SDRE approach is used as a nonlinear optimal control method. The suggested approach is simulated for a two degrees of freedom mechanical arm with assuming a friction model. To verify the performance of identification algorithms, the Experimental tests have been implemented on a single mechanical arm. By satisfactorily identifying the friction parameters and compensating it by the SDRE controller, the final error of End-effector is reduced.