Research On Design And Application Of A Soft Power-Assisted Upper Exoskeleton

As a typical man-machine interaction system,the power exoskeleton system has a wide application background in the fields of industrial production,medical rehabilitation and military strategy by combining human intelligent decision and power output of robots.As the main component of the man-machine integration system,the exoskeleton needs to have the ability to follow users to ensure the same trajectory.In this study,the soft power exoskeleton system of upper limb was designed and studied to meet the requirements of motion performance and sensing control of upper limb exoskeleton system.This study mainly started from two levels of theory and practice.First,based on the screw theory,a type synthesis approach of hybrid mechanisms with fewer degrees of freedom based on the constraint limb was proposed.Next,from the analysis of the large deformation of the flexible rod,the simplified model of the flexible rod in a plane and the three-dimensional space was put forward,and then the kinematic model of the exoskeleton was analyzed based on the simplified model of the flexible rod.Next,a surface electromyography-force model(sEMG-F)based on surface electromyography signal was proposed to predict the direction of hand force.Finally,based on the kinematic model and sEMG-F model of exoskeleton,a fuzzy control strategy of exoskeleton was proposed for the motion control of exoskeleton.With the focus on the design of a hybrid mechanism and problems of complicated analysis,this research analyzes the mutual relations between motion axes based on the screw theory.A method for constructing a constrained limb between the motion axes is proposed,and expanded to a new approach to convert a parallel mechanism to hybrid mechanism by adding constrained limbs.The mechanism of soft exoskeleton is designed using the same approach.In view of the large deformation of flexible rod,this study proposed the spline pseudo-rigid-body model based on the pseudo-rigid-body model and spline curve and solved all characteristic parameters of spline pseudo-rigid-body model with optimal algorithm.In view of the deformation analysis of flexible rod in three-dimensional space,this study proposed spline Cosserat model and designed a simulation experiment.The results are inputted to a Back Propagation(BP)neural network trainer for simplifying the mapping relation between flexible rod and deformation.In view of the high-order non-linear problem of kinematic model of flexible mechanism,this study combined the analysis of mechanism joint and body structure of soft exoskeleton and established the kinematic model of each joint of exoskeleton on the basis of man-machine integration ideas.Based on the force-deformation mapping of flexible rods,the kinematical model of driving components was analyzed and applied to the kinematic solution of exoskeleton joints.Motion intention recognition is one of the basic technologies to realize man-machine interaction in exoskeletons.In this study,the force direction of the hand was used to characterize the movement intention of the upper limb of the human body.The related experiment and training were designed to deduction the sEMG-F model.The sEMG-F model obtained the intermediate variable by multiplying the collected electromyography signal with electromyography-force matrix through wavelet transform and processing.The hand force direction is obtained by a BP neural network with intermediate variable as input.In view of different usage scenarios of exoskeleton,the power-assisted control strategy and the rehabilitation control strategy of the exoskeleton were presented in this study,respectively.Among them,the power-assisted control strategy realized the motion control of exoskeleton system based on the electromyography model and fuzzy controller,and combined with the gravity compensation torque and the interactive torque of the driving joints.In the rehabilitation control strategy,this study planned the movement trajectory of exoskeleton in the virtual environment mainly based on the improved artificial potential field method and realized the rehabilitation training task of the system combined with the exoskeleton control model.