A Study On Lower Limb Movement Recognition Based On Android Multi-sensor Acquisition System

The control of human-assisting systems such as exoskeletons,intelligent prostheses,and rehabilitation robots require the recognition of the movements of lower limbs.The surface electromyography signal contains the information of the human movement,and the control of the human-assisting system through the surface electromyography signal can provide an intuitive control experience.In order to obtain a better recognition rate,a variety of signals such as surface electromyography signal,plantar pressure signal,and attitude signal can be used in the recognition.The multi-sensor acquisition system with Android as the host computer has the advantages of being portable,easy to expand,and low cost compared with professional signal acquisition equipment.This paper proposes a portable multi-sensor acquisition system based on Android and uses different signals to recognize different lower limb movement patterns.Firstly,a multi-sensor acquisition system based on Android is designed and built,which can collect the surface electromyography signals,plantar pressure signals,and attitude sensor signals of the lower limbs and send them to mobile phone applications through Bluetooth.Mobile phone applications can display,calculate,and save signals.Secondly,the five motionless movement patterns including neutral position,dorsiflexion,plantar flexion,inversion,and eversion are recognized by using the surface electromyography signals.Linear discriminant analysis(LDA)algorithm is used to train and recognize the feature vectors of surface electromyography signals.The optimal channel set,optimal feature set,and averaged number are determined.Thirdly,seven movement patterns including walking,running,marching in place,stairs ascent,stairs descent,standing,and sitting are recognized based on the surface electromyography signal,plantar pressure signal,and attitude signal.Distinguish between locomotion movement patterns and motionless movement patterns is based on the angular velocity signal of the attitude sensor.The phase-dependent LDA algorithm is utilized to recognize four locomotion movement patterns of walking,running,stairs ascent,and stairs descent.The optimal phase and the optimal feature set are determined.Based on the plantar pressure signal,two motionless movement patterns including standing and sitting are recognized.Finally,the virtual lower limbs based on Android are created.Based on the Euler angle signal of the attitude sensor,the template data,and the phase difference of the left and right foot movements,the postures of the virtual lower limbs are updated in real-time to make the virtual lower limbs follow the movement of human during walking.In the process of human walking,the interface of the mobile phone application can display the activation of muscles,the predicted time of the gait cycle,the gait phase,and information of multi-sensor in real-time.