A Development Of Following And Human-Powered Augmentation Control For Load-Carrying Lower Exoskeleton

For load-carrying lower exoskeleton,aiming at its three key problems(dynamic mod-els,adaptability for loads and balance): the model-based control algorithm in the swing phase requires accurate dynamics models? the adaptability for variable loads and the bal-ance are not considered in the support phase.Therefore,this thesis focus on the key per-formances of the following for swing phase,adaptability for variable loads and the balance for stance phase to improve the assistance efficiency.Main contributions of this thesis are described as follows:For the dependency of the Sensitivity Amplification Control(SAC)on accurate dy-namic model in the swing phase,a novel Model-based control with Interaction Predicting(MIP)is proposed.Which can predict the interaction with gait descriptors and the inter-action is used as the cost function to optimize the weights of the model-based impedance controller and SAC to lower dependency on accurate dynamic model.The interaction forces resulted from inaccurate dynamic model can be reduced by 88.9%.And the pro-posed algorithm can reduce the metabolic energy of the swing leg by 44.6% compared to SAC.In order to improve adaptability for variable loads in the stance phase,a novel Adap-tive Torque for Variable Loads(ATVL)based on Stylistic Dynamic Movement Primitives(SDMP)is proposed.In which,the torque model based on SDMP is formed according to the human joint torques with different loads,and the adaptable function relationship be-tween the parameters of the torque model and loads is learned by Locally Weight Gaussian Regression(LWGR).According to the oxygen consumption during walking with loads,the proposed algorithm can reduce the metabolic energy by 20.4% compared to without torques.Considering the balance during walking with loads,an Adaptive Torque for Variable Loads based on Balance constraints(ATVLB)is proposed based on the ATVL.Under the constraints of inverted pendulum model,the torques planned by the ATVLB can improve the balance during walking with loads.In the experiments,the double support phase can be reduced by 5.4%.And it reduce the metabolic energy by 21.6% compared to without torques.In order to validate the proposed algorithms,a Human-powered Augmentation Lower Exoskeleton(HUALEX)system is bult with hydraulic actuators.And participated in the national competition of lower exoskeleton for load-carrying walking.In which,the field simulation environment is used for estimating the HUALEX system based on the algo-rithms of this thesis.The algorithms developed in this thesis can be used in application scenarios such as industrial assistance,outdoor hiking and outdoor mountaineering in the future.