Tuning of Feedback Linearization Controller Based on FRIT Method

Abstract

In this study, we propose a new parameter tuning method for robotic manipulators which enables to tune linear feedback gain and dynamical parameters in exact linearization controller from only one-shot experimental data. This method is an extension of FRIT (Fictitious Reference Iterative Tuning) to nonlinear systems. Tuning process of FRIT method can be considered as model matching to the unknown plant. In this study, by providing mathematical model of the unknown plant, we can find parameters which achieve the desired response for any reference. In order to show validity of this method, we apply this method to resolved acceleration controller for a SCARA type robot, and simulation was performed. Furthermore, we analyzed the effect of error in provided model of controller and the dependence of tuned parameters on experimental data.