摘要：AbstractStroke patients often suffer from a partial paralysis hindering shoulder elevation movements. To support the small voluntary forces, Functional Electrical Stimulation (FES) can be applied to the shoulder-deltoid muscle for gaining additional force. In previous research we presented a positive-feedback control loop which applies a stimulation level that causes the arm abduction to be hold at a given percentage (support factor) of the currently measured angle. This approach causes an amplification of the effect of the residual volitional activity and effectively realizes an arm weight relief. The support-level strongly influences the closed-loop’s behavior especially for percentages close to 100% and, therefore, must be carefully chosen to meet the patient’s requirements. Because of parameter variations, this value cannot be precisely pre-computed, and, hence, we present an adaptive trial-to-trial based learning procedure to iteratively realize a given level of volitional amplification. During each trial, the algorithm obtains measurements of the volitional activity obtained by means of Electromyography (EMG) and the joint angle. After the arm returns to its rest position, the achieved amplification of the volitional activity is estimated. An integral controller combined with a non-linear output transformation then updates the support factor for the next trial aiming to realize the desired volitional amplification. The closed-loop adaptive controller is robust asymptotically stable. In two tests performed on a healthy subject, the desired level of volitional amplification was reached within 3 to 5 trials demonstrating the feasibility of the chosen approach.
