Research On Plantar Pressure Characteristics Of Exoskeleton Knee Walking Robot

The lower extremity exoskeleton walking robot can assist the human walking,enhance the human movement ability,and reduce the wear of human joints.Plantar pressure is an important biomechanical parameter in human lower limb movement,which can reflect human movement information.Accurate acquisition of human motion information is the technical difficulty of lower limb exoskeleton robot control system and the key to realize human-machine coordination control.Focusing on the characteristics of plantar pressure,this paper carries out research on human motion pattern recognition,human gait recognition,plantar pressure prediction,knee joint Angle prediction and so on.Firstly,the exoskeleton knee joint walking robot and the plantar pressure acquisition system were designed as the core of the exoskeleton knee joint walking robot software and hardware system platform,including the exoskeleton robot mechanism,plantar pressure software and hardware acquisition system,high-speed camera synchronous acquisition system and other auxiliary verification equipment.The collected plantar pressure signal was processed and analyzed.A filter was designed to remove the signal noise and extract the plantar pressure characteristics.Secondly,the plantar pressure characteristics of different movement modes and gaits were analyzed according to the synchronous lower limb movement images.The planar pressure threshold was used to recognize human movement patterns and human walking gait.A total of 12 groups of experimental data of two subjects were identified,including 60 times of movement pattern recognition and 784 times of gait recognition.By synchronously collecting images,the error analysis of the start and stop recognition results in pattern recognition was carried out,and the average error time was 320 ms.The error analysis of each gait in gait recognition was carried out by synchronously collecting images,and the recognition accuracy was 99.87%.Then,the BP neural network was built to predict the time series,and the planthar pressure was predicted by using the historical information of planthar pressure.The mean square error(MSE)between the predicted results and the actual value was 0.5180,and the mean root mean square error(RMSE)was 0.6438,which proved that the method could realize the plantar pressure prediction.Based on the prediction of plantar pressure,gait recognition was carried out.The success rate of recognition was 100%,which proves the feasibility of the gait prediction method.Then,for the realization of the plantar pressure Angle of knee joint,knee joint Angle BP neural network prediction model was put forward,by entering a foot pressure forecast the knee joint Angle,the actual value attitude sensor synchronous measurement of knee joint Angle,the root mean square error of prediction results and the actual value was 10.2322,the average error Angle was 7.1265 °,It was proved that the method of knee joint Angle prediction was feasible.Finally,the knee joint motion control platform of the exoskeleton robot was built,and the motor of the knee joint was controlled according to the predicted Angle of the knee joint.The motor rotation Angle was obtained through the encoder.The root mean square error of the motor rotation Angle was 23.4219,and the average error of the Angle was 16.85°.The experimental results verified the correctness of the exoskeleton knee joint robot system and the research method of plantar pressure characteristics.