Mechanism Design And Motion Control Of Ankle-foot Exoskeleton Driven By PMAs

Ankle movement disorder is the most common complication in patients with nerve injury such as stroke/hemiplegia,which seriously affects the patient’s walking ability.Hemiplegia patients are prone to ankle dorsiflexion and foot inversion,which will reduce the contact surface between foot and ground and affect the stability of walking.Traditional walking-assisted rehabilitation lacks of targeted rehabilitation training and real-time feedback.Based on the concept of ‘repetition,concentration and task orientation’,the robot can help patients simulate the normal physiological gait patterns for walking training to improve the recovery of hemiplegic gait.However,the current exoskeletons are mostly driven by motors,which are uncomfortable and bulky to wear.Moreover,the rigid exoskeletons mostly adopt fixed configuration,which is difficult to drive and support at the sole of the foot,and cannot meet the rehabilitation needs of different patients or different rehabilitation stages.Therefore,this paper focuses on the design and control of the ankle-foot exoskeleton driven by multi-PMAs,and applies it to ankle rehabilitation training to mainly improve ankle dorsiflexion.The main work is as follows:(1)Mechanism design of flexible ankle-foot exoskeleton driven by multi-PMAs.Aiming at the problems of inconvenient wearing and lack of drive on the sole of the foot,a flexible ankle-foot exoskeleton driven by multi-PMAs is designed.The exoskeleton can realize the movement of ankle flexion/extension,eversion/inversion and adduction/abduction degrees of freedom.An effective plantar support structure is formed through the combined drive of pneumatic muscles and flat pneumatic muscle to improve ankle dorsiflexion.A kinematic model of the flexible ankle-foot exoskeleton is established,and its dynamic model based on the Lagrangian equation is established on this basis,and the stress-strain analysis of the key parts is carried out to ensure that it can meet the actual rehabilitation training intensity.Finally,the design and integration of the software and hardware control system of the flexible ankle-foot exoskeleton is completed.(2)Personalized gait prediction model based on attention-based CNN-LSTM network.As current prediction methods do not fully consider the correlation of gait data in time and space,and the importance of different dimension is different,resulting in low prediction accuracy,a personalized gait prediction method based on attention-based CNN-LSTM network is studied.The motion intention of the affected side is predicted through the motion data of the healthy side,and the upper and lower limb cooperation data is added to expand the data dimension.The attention mechanism is used to assign weights to important dimensions to effectively improve the prediction accuracy,which realize the height coincidence between the exoskeleton and the human,and reduce the reverse coupling force between the user and the exoskeleton.(3)Research on cooperative control of multi-PMAs for individualized rehabilitation.Aiming at the difficulty of improving the ankle dorsiflexion function of the current rehabilitation robot,the motion control strategy of ankle-foot exoskeleton driven by multi-PMAs for personalized rehabilitation is studied.The expected trajectory of the ankle is decomposed into the control target positions of the pneumatic muscles and flat pneumatic muscle by the peak fitting model.Multi-PMAs are timely and accurately controlled by the fuzzy adaptive PID method to track the predetermined trajectory for the patient,which provides flexible and adjustable assist force for patients in different gait stages,and realizes the natural human-robot coordinated control of the flexible ankle-foot exoskeleton during walking.