Font Size: a A A

Research On Active Assistance Control Of Exoskeleton Based On The Estimation Of Joint Torque And Instability State

Posted on:2022-04-03Degree:DoctorType:Dissertation
Country:ChinaCandidate:S Y QiuFull Text:PDF
GTID:1522306839476564Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
At present,the world has entered the age of aging,and the huge pressure of social medical security and the serious shortage of labor force have become two prominent problems facing the aging society.As a kind of intelligent wearable assistance equipment,the exoskeleton robot has a broad application in the aspects of assisting the elderly and the disabled and motion assistance.However,higher requests on the assistance effect and safety of exoskeleton are needed by the aging society.The assistance effect of exoskeleton needs to be further improved and the walking stability of exoskeleton needs to be guaranteed.Comparing with the traditional motion following control,the active assistance control technology can not only guarantee the motion autonomy of the human body,but also actively provide assistance for the human body.Therefore,active assistance control is expected to further improve the assistance effect and safety of the exoskeleton.However,the current active assistance control technology still has the following two prominent problems: firstly,the hysteresis of human joint torque estimation restricts the further improvement of its active assistance effect;secondly,it lacks the active perception ability of human stability state and the ability to correct motion instability.Therefore,this paper aims to further improve the active assistance effect and safety of lower extremity exoskeleton.This paper focus on the following key issues such as prediction of human joint motion trajectory,estimation of human joint torque,estimation of human instability state and active assist control method of lower limb exoskeleton.Firstly,the research of human joint torque estimation is carried out.Human joint torque is an important manifestation of human motion intention.To accurate obtain the real-time torque information and ensure the realization of the active assist control,the key to estimation of human joint torque is to solve the problem of estimation lag.Therefore,by improving the traditional adaptive oscillator algorithm and dynamic motion primitive algorithm,this paper introduces the online estimation of motion trajectory frequency and proposes the frequency adaptive dynamic motion primitive algorithm(FADMPS)to realize the online learning and prediction of the motion trajectory of lower limb joints when the human body walks stable.Based on the statistical data of body size and limb inertia parameters of Chinese adults,the contraindicated kinetic parameter model of human body and the inverse kinetic model of human lower limbs were established.The real-time estimation of human joint torque was realized based on the FADMPS algorithm and the lower limb inverse dynamics model of human body.The FADMPS algorithm was used to predict the trajectory of human joint torque when walking in advance,which improved the hysteresis of human joint torque estimation caused by the human nervous system and exoskeleton sensing system.Instability state estimation of human body is the basis of active balance assistance control of exoskeleton.In order to solve the problem of fast and accurate estimation of instability state of human body during standing and walking,this paper carried out the research of human instability state estimation based on multi-sensor information fusion.Firstly,a human inverted pendulum model was established to analyze the standing stability and walking stability of human body,and the instantaneous capture point and stable landing point were deduced.Based on this,the basis and conditions for judging the standing stability and walking stability of human body were obtained.Then the Kalman filter algorithm is used to fuse the data of multiple human attitude sensors to estimate the state of the center of mass of human body.The human body instability state estimation method is proposed based on the instantaneous capture point and the center of mass.It realizes the rapid and accurate estimation of the unstable state of human body during standing and walking.The effectiveness of the human body instability state estimation method was verified by the bipedal instability recovery process simulation based on the capture point tracking control.On the basis of above two studies,this paper further carried out the research on active assist control of lower extremity exoskeleton based on the estimation of human instability state.To reduce the energy consumption of human body during stable walking through active assistance,and at the same time to ensure that the unstable state of human body quickly recovered after external interference.In this paper,the framework of the lower extremity exoskeleton active assistance control system is established to selfregulating assistance mode based on the stable state of human body.In this framework,an active walking assistance control based on FADMPS and linear extended state observer is used to accurately track the target assistance torque trajectory of the exoskeleton during stable walking to improve the power helping effect and reduce the energy consumption of human body.Based on the instantaneous capture point virtual stiffness model,the mapping between the instability degree of human body and the active assistance torque of the exoskeleton was established.Through the active balance assistance control,the rapid recovery to stability was guaranteed.The effectiveness of the proposed method is verified by the simulation and experiment of the related active assistance control of the exoskeleton robot.Finally,this paper carried out an experimental study on the lower extremity exoskeleton active assistance control,and tested the instability state estimation of human body,the effect of walking active assist control and the effect of instability active assist control.On the one hand,the effect of the active walking assistance control method is tested by measuring the joint motion trajectory,lower limb muscle activity and human energy consumption in the process of active walking assistance.On the other hand,the actual effect of the active assist control method for instability is tested by measuring and analyzing key indicators such as the motion state of the Co M,the motion state of the instantaneous capture point,the motion state of the swinging foot,the recovery time of instability and the stability margin of human body during the process of instability recovery.The effectiveness of the proposed methods are verified by the quantitative analysis of the experimental results.
Keywords/Search Tags:Lower limb exoskeleton, Joint trajectory prediction, Joint torque estimation, Instability state estimation, Active walking assistance, Active balance assistance
PDF Full Text Request
Related items