Research On The Key Technologies Of Motion Control For Wearable Lower Limb Walking Assistant Robots

Wearable lower limb walking assistant robots(WLLWARs)are the robotic systems that can be worn by users and can assist the wearers’ lower limbs to complete the walking task.They have broad application prospects in assisting the elderly and the disabled,life service,industrial manufacturing,national defense and military,and other fields.Improving the motion control performance for WLLWARs will be of great significance to improve the social medical rehabilitation service level,and enhance the working ability of personnel in specific industries.However,there are some problems in the current researches of motion control for WLLWARs,such as motion asymmetry,inaccurate tracking and non-compliant interaction.In this work,according to the problems existing in the current researches,two representative WLLWARs,the robotic prosthesis leg and the soft exosuit,are selected as the specific verification objects.Then,the experimental platforms of the selected verification objects are designed and integrated.Based on the experimental platforms,the research on the key technologies of motion control for WLLWARs is carried out to promote the development of WLLWARs to the direction of high coordination,accurate tracking and compliant interaction.The main research contents and results are as follows:(1)Aiming at the problem of lacking of the real-time motion symmetry between the affected side and the healthy side in the task space when a wearer with unilateral lower limb disability is assisted by a WLLWAR,the robotic prosthesis leg is selected as the specific verification object,and the real-time motion planning method based on the neurodynamic optimization is proposed.Firstly,the kinematics and dynamics of the robotic prosthesis leg are modeled.Then,the motion planning problem for reproducing the walking speed of the healthy side to the affected side is modeled as a quadratic programming(QP)problem with constraints,and the neurodynamic optimization method based on the varying parameter recurrent neural network is adopted to solve QP problem in real time.Finally,the experiments verify that the proposed method can not only enhance the motion symmetry of the prosthesis and the healthy leg in the task space,but also meet the requirement of real-time motion planning for the robotic prosthesis leg with limited computing capability.(2)Aiming at the problem that the tracking control accuracy of the WLLWAR is affected by the external periodic disturbances caused by assisting the wearer to complete the repetitive motion task,the robotic prosthesis leg is selected as the verification research object,and the external periodic disturbance-oriented repetitive learning control method is proposed.Firstly,the repetitive learning control method is designed for the repetitive characteristics of the joint desired motion trajectories of the robotic prosthesis leg,which enables the prosthesis to continuously learn the periodic disturbances and impedance dynamics parameters in the process of tracking the joint repetitive desired motion trajectories.Then,the stability of the proposed control method is proved.Finally,the experiments verify that the proposed method can improve the motion trajectory tracking accuracy of the robotic prosthesis leg in the case of the external periodic disturbances.(3)Aiming at the problems that it is difficult to model the human-robot coupling systems of the WLLWAR,and the tracking control accuracy of the WLLWAR is affected by a large number of unknown nonlinear uncertainties in the dynamics of the human-robot coupling system,the soft exosuit is selected as the specific verification object,and the uncertain human-robot coupling dynamics-oriented adaptive fuzzy control method is proposed.Firstly,the kinematics and dynamics of the human-exosuit coupling system are modeled,and the gait detection method for triggering the ankle plantarflexion assistance is designed by analyzing the normal gait of healthy human.Then,the adaptive fuzzy control method is designed to approximate the unknown nonlinear uncertain terms of the human-robot coupling dynamic model in the process of tracking the desired motion trajectory,and then the stability of the proposed control method is proved.Finally,the experiments verify that the proposed method can effectively improve the motion trajectory tracking accuracy of the soft exosuit in the case of the uncertain human-robot coupling dynamics.(4)Aiming at the problem that the mismatch between the WLLWAR’s preset desired motion trajectory and the wearer’s autonomous motion resulting in the lack of human-robot interaction compliance,the soft exosuit is selected as the specific verification object,and the compliant human-robot interaction control strategy based on impedance learning is proposed.Firstly,in order to deal with the lack of adaptive ability of traditional impedance control with fixed parameters to different wearers’ joint impedance characteristics,the compliant human-robot interaction control strategy with internal and external loops is designed.Then,the impedance learning method is designed in the external loop,and the parameters of the target impedance model are obtained by iterative updating.The reference motion trajectory is calculated by combining the preset desired motion trajectory and measured human-exosuit interaction force,and then the reference motion trajectory is tracked through the tracking control method in the internal loop.Finally,the experiments verify that the proposed strategy can make the soft exosuit adaptively adjust the dynamic relationship between the desired position and the interaction force of the human-exosuit system according to the motion characteristics of different wearers,and apply the compliant human-robot interaction to the wearers for assisting walking.