FES based rehabilitation of the upper limb using input/output linearization and ILC

To provide effective stroke rehabilitation, a control scheme is developed for upper arm tracking in 3D space using electrical stimulation. In accordance with clinical need, the case where stimulation is applied to two muscles in the arm and shoulder is considered, with the arm supported against gravity by an exoskeletal mechanism. An upper limb model with five degrees of freedom is first developed to represent the unconstrained upper arm, and an input/output linearization controller is applied to decouple the actuated joint angles, and combined with a state-feedback optimal tracking controller. Linear iterative learning controllers are then designed to enforce precise tracking over repeated attempts at the task, and stability conditions for the unactuated joint angles are given. Experimental results confirm practical performance.