Design of an optimal feedback linearizing-based controller for an experimental flexible-joint robot manipulator

This paper presents the design of an optimal non-linear position tracking controller for a two-link flexible joint robot manipulator. The controller is designed based on the concept of exact feedback linearization and LQG/LTR techniques. It is shown that the non-linear robot model is feedback linearizable and a characterization of the set, over which the linearizing transformation is diffeomorphic, is provided. The proposed control approach reduces the number of required measurement sensors and takes into account the effects of measurement noises. A new method for computing the non-linear state estimate is also presented. It takes advantage of the linear structure of the transformed system. Simulation results demonstrate the potential benefits of the proposed control approach in reaching the desired performance with minimum control effort. Copyright © 1999 John Wiley & Sons, Ltd.