Research On Human-machine Coupled Coordinative Motion Control Method Of Lower Limb Exoskeleton

With the increasing concern of the country and society for population aging and laborintensive industries transformation,various kinds of lower limb exoskeletons for the aged,disabled and human enhancement have shown far-reaching application prospects and research values in various aspects such as medical rehabilitation,industrial production and military fields.For the design and research of lower limb exoskeletons,problems such as low model accuracy,poor control performance,uncoordinated human-machine coupling,inaccurate intent recognition,and unintelligent control strategies still need to be solved.Therefore,this paper aims at the human-machine coupling control performance of the lower limb exoskeleton,and carries out relevant research from the point of view of the accuracy,safety and intelligence of the human-machine coupling control,and finally hopes that the coupled system composed of lower limb exoskeleton and human body can efficiently accomplish the set control objectives.The main research work of this thesis is shown as follows.1.In order to provide a corresponding validation platform for the studied control algorithms and strategies,the first-generation lower limb exoskeleton prototype UEXO-I with electric direct drive joints was built,and the mechanical structure,software and hardware composition,and sensor measurement deviation compensation correction method of UEXO-I were introduced in detail.In order to further verify the prediction of the control algorithm under bipedal walking,the second-generation lower limb exoskeleton prototype UEXO-II was built on the basis of UEXO-I,and its mechanical and hardware and software components were introduced.Finally,relevant validation experiments are designed to verify the actual performance of UEXO-I and UEXO-II.2.To address the problem of unknown kinetic model of the lower limb exoskeleton prototype,a method is proposed to identify the kinetic parameters of the lower limb exoskeleton based on the neighborhood optimization algorithm(NFO)and Huber fitness function,which can effectively suppress the influence of measurement noise on the identification accuracy while ensuring the identification efficiency.Based on the identification results,the corresponding control algorithms are designed for the active and passive control modes.In particular,for the active control of the lower limb exoskeleton with the conductance control architecture,a variable conductance control strategy based on the human step frequency is designed to take into account the stability and flexibility of human-machine coupling.3.Considering the stochasticity and uncertainty of human lower limb gait motion information,the gaussian process(GP)method is used to learn human gait information and patterns and apply them to the lower limb exoskeleton control strategy.Firstly,an online gait prediction model based on deep gaussian process(DGP)is proposed to compensate for the measurement delay of human lower limb posture,and a variable conductance control strategy is designed to further guarantee the human-machine coupling performance according to the prediction uncertainty index.In addition,a sparse gaussian process(SGP)based probabilistic model is proposed for the coupling relationship between the two-legged and two-joint trajectories of walking gait,and the output constraint technique combined with the probability estimation confidence interval is used to ensure the safety and coordination of the human-machine coupled system.4.Since the active and passive control performance of the lower limb exoskeleton is dependent on the corresponding position controller performance,the position controller design for the lower limb exoskeleton is investigated.First,an adaptive backstepping controller based on the set-total error bound is proposed to deal with the effect of model identification error.In addition,considering the unmeasurable joint angle of the lower limb exoskeleton,an extended state observer(ESO)-based backstepping controller is proposed to simultaneously estimate and compensate the joint angular velocity and the total set error.Finally,considering the joint position constraint of the lower limb exoskeleton,the finite-time extended state observer(FESO)-based output-constrained backstepping controller is proposed to guarantee the estimation efficiency and achieve the joint trajectory constraint of the lower limb exoskeleton at the same time.5.Considering the modular lower limb exoskeleton system,each limb exoskeleton is regarded as an independent node and constructed as a whole distributed system by using communication networking.Based on the flexible control characteristics of exoskeleton control,a theoretical study of the above problem is conducted,and a distributed impedance control framework is proposed,under which a distributed impedance controller with radial basis function(RBF)neural network is designed to suppress the influence of node model uncertainty on the control performance.In addition,considering that the exoskeleton helps the human body to accomplish specific control objectives by means of human-machine coupled force/moment,a distributed coupled-action co-operative controller is proposed on the basis of distributed impedance control,and the transition from distributed position control-distributed impedance control-distributed coupled-action control is realized by adjusting the control parameters.