Control of tibialis anterior FES envelop for unilateral drop foot gait correction using NARX neural network

In this paper a control methodology based on artificial neural networks (ANN) is proposed for control of ankle dorsiflexion in patients with unilateral drop foot. In the presented strategy, the electrical stimulation intensity for the disabled tibialis anterior (TA) muscle is controlled considering the existing coordination patterns between activities of the ipsilateral ankle dorsiflexor muscles and the contralateral ankle plantarflexor muscles during normal gait. Based on this coordination, in each gait cycle the TA muscle of one leg acts in close simultaneity with the calf muscle of the opposite leg. Therefore in this paper a dynamic ANN has been trained in a predictive manner, to forecast the disabled TA muscle activity based on the input from the healthy calf muscle of the opposite leg. The predicted TA activation is then used to control the TA muscle FES intensity in real time. Seven healthy volunteers participated in the experiments. Surface electromyogram was recorded from TA and calf muscle simultaneously on the opposite legs while walking in different gait frequencies. Results obtained from the controller are quite promising and show impressive generalization ability between subjects.