Dynamic Gait Modeling For A Lower Extremity Exoskeleton System

Exoskeleton robot is a kind of technology that integrates sensing,control,information,fusion and mobile computing to provide wearable mechanical mechanism for its wearers.At present,exoskeleton robot has been widely used in medical rehabilitation,military fire fighting,industrial loading and other fields.Exoskeletons,which help normal adults build strength and load capacity,have attracted worldwide attention.At the current stage,most of the studies on assisted exoskeleton control methods are limited to the torque control in the exoskeleton support stage and the follow-up control in the swing stage.This hybrid active and passive control method has been basically mature at present.But in complete active control direction,there is no very good solution,so this paper proposes a new exoskeleton robot gait mode,the strategy aims to realize the active control mode for exoskeleton during whole gait cycle,through the human-computer interaction sensing information perception the wearer's movement trend.And through the gait model output appropriate exoskeleton gait curve,achieve good man-machine coupling and flexibility.The main work and achievements of this dissertation are summarized as follows:1.This paper proposes a parameteric gait prediction model.The model is based on Dynamic Movement Primitives(DMP).This method is an efficient,stable and practical dynamic trajectory planning algorithm.By introducing a second-order dynamic system and an attractor model,it first uses the radial basis function to learn the reference trajectory,and then uses the attractor model to modulate the trajectory and output a new desired trajectory with reference trajectory characteristics.The parametric gait prediction model proposed in this paper innovatively expands the model of dynamic movement primitives and enables it to learn multiple trajectory characteristics.At the same time,the principal component sequence of the curve set is proposed by using the singular value decomposition method.The dimensionality of the original curve set is reduced and the most important features are extracted for learning.The algorithm efficiency is optimized and the response time of the system is improved.Finally,the model utilizes the regression algorithm in machine learning to establish the mapping relationship between the walking speed of exoskeleton wearers and the principal component parameters in the parametric gait prediction model,so as to achieve a complete parametric gait prediction model.2.On the basis of the above parameteric gait model,the dynamic parameteric gait optimization model is established.The output of the parameteric gait prediction model is dynamically modulated by adding the coupling term of cooperative motion primitives into the dynamic movement primitives system.At the same time,an iterative learning control algorithm is introduced to iteratively optimize the deviation between the expected curve and the model output curve.The ideal dynamic parameteric gait optimization model is established.The model has the ability to optimize the output curve of the parameteric gait prediction model,which enhances the human-machine coupling between the exoskeleton robot and the wearer,and also ensures the comfort and safety of the exoskeleton wearer.