Disturbance Observer-Based Patient-Cooperative Control of a Lower Extremity Rehabilitation Exoskeleton

Many patients with stroke are suffering lower limb locomotor dysfunctions all over the world. Body weight supported treadmill training has proven to be an effective post-stroke rehabilitation training method for these people’s recovery. Nowadays, lower extremity rehabilitation exoskeleton composed of a pair of mechanical legs has been introduced into body weight supported treadmill training, which can guide and assist the movements of the patient’s legs. However, active movements of the patient are hardly to be achieved when the rehabilitation exoskeleton is controlled by a commonly utilized position-based passive strategy. Considering the restriction above, a weight supported rehabilitation training exoskeleton device was designed in this paper to ensure the stroke patient can participate in rehabilitation training voluntarily. To realize this goal, a patient-cooperative rehabilitation training strategy based on adaptive impedance control is adopted for the swing phase in the training. Human–exoskeleton interaction torques are evaluated by a backpropagation neural network and a disturbance observer whose stability is proved by Lyapunov’s law. With no additional demand of interaction torque sensors, the complexity of this system is simplified and the cost is reduced. In order to promote the involvement of patient during the rehabilitation training, fuzzy algorithm is used to adjust the impedance parameters according to the human–exoskeleton interaction torques. The effectiveness of the whole rehabilitation control strategy is demonstrated by experimental results.