| Lower limb motor dysfunction caused by stroke and spinal cord injury seriously affects the patients’ normal walking and quality of life.Using rehabilitation robot to carry out scientific and targeted gait rehabilitation training for patients can effectively reduce the burden of therapists and improve the lower limb motor function and normal gait level of patients.Gait planning is an important part of the walking aid control of lower limb rehabilitation robot.The goal is to ensure that the motion trajectory of the robot meets the actual rehabilitation training needs and provide appropriate and natural rehabilitation training for patients.However,the current gait planning methods do not fully consider the individual differences of patients,lack of adaptability to different subjects,and can not be adjusted timely according to the patient information in the actual training process.Therefore,how to provide adaptive,safe and efficient gait rehabilitation training according to the subjects’ physical characteristics,exercise intention and gait pattern is an important problem to be solved.In this paper,a personalized gait trajectory planning model is deeply studied,a gait prediction method based on feature fusion is proposed,and the trajectory is adaptively optimized and adjusted accordingly,and training strategies for different rehabilitation stages are designed to achieve self-adaptation to different individuals.Gait rehabilitation training.The main work is as follows:(1)Personalized gait trajectory planning based on human characteristic parameters.Aiming at the problem that most of the existing gait planning adopt unified gait trajectory and lack of adaptability to individuals,a personalized gait planning method based on human characteristic parameters is proposed.The gait trajectory learning model is established by using the dynamic motion primitive algorithm,and the Gaussian mixture regression algorithm is introduced to improve the generalization ability of the model.On this basis,the neural network is used to establish the mapping relationship between human parameters and gait parameters.The corresponding gait parameters are obtained according to different individuals,the model parameters are adjusted,and the personalized gait trajectory suitable for specific subjects is obtained.(2)Gait trajectory optimization based on personal motion intention prediction.In the actual assisted walking process,its gait trajectory needs to be adjusted and optimized according to people’s intention.An optimized long-term and short-term memory neural network architecture is proposed.The feature extraction network of group prediction is designed to extract the temporal motion features of joint angle signal and electromyogram signal respectively,predict the gait motion intention based on the fusion features of double full connection layers,and then adaptively optimize and adjust the reference trajectory.The motion data sets of flat walking and uphill are constructed to verify the effectiveness of this method in improving the accuracy of gait motion intention prediction,and then improve the adaptability to the actual environment.(3)Assistant control strategy of lower limb rehabilitation robot adapted to personalized trajectory.The gait rehabilitation training strategies for different rehabilitation stages are studied.In the early stage of rehabilitation,the personalized trajectory is accurately tracked based on the inverse sliding mode control method to realize the simulated walking gait rehabilitation training;In the middle and late stage of rehabilitation,the active rehabilitation training strategy based on human-computer interaction is adopted to establish the impedance model of ankle rehabilitation robot.Combined with the optimized trajectory,the motion trajectory is further controlled and adjusted according to the patient’s active motion ability under different gait modes,so as to realize personalized rehabilitation training in line with human gait rules and adapt to different environments. |
PDF Full Text Request