| There are a large number of special people with lower limb dyskinesia in our country.At the same time,as we are entering an aging society,the trend of aging population is increasing,and the population of empty nest elderly has gradually increased too.The care and nursing problems of such people have become increasingly prominent.The lower extremity exoskeleton robot can significantly improve the motor ability and quality of life of people with lower limb dysfunction,and is one of the most promising ways to solve the problems of lower limb disability and elderly mobility.Exoskeleton robot is a human-centered human-machine coordination intelligent system.To achieve the goal of assisting mobility,the wearer's motion intentions should be fully perceived to provide appropriate power assisting strategies.According to the analysis and summary of the current research status,it can be found that the commercial exoskeleton robots are applying to various fields such as medical and logistics industry,however,the wearer's intentions have not fully perceived.Besides,the human-exoskeleton coordination strategy and assisting strategy are rough.Bioelectric signals such as electromyogram and electroencephalogram can reflect the original movement intention of the wearer,which is an important way to achieve natural human-exoskeleton interaction.However,there are relatively few researches on the applying the bioelectric signals to the coordination strategy of exoskeleton robots.Thus,it is urgent to further study.In the application of lower limb paraplegia rehabilitation,the fixed gait trajectory is adopted mostly and the wearer need proactively coordinate with the exoskeleton by long-term training.In the application of elderly assistance,the fixed torque assistance is adopted mostly with limited functions and insufficient adaptability to daily life such as multiwalking tasks.Considering the limitations such as fixed strategy,time lag,and with single function,it is difficult to adaptively couple with the wearer,and it is difficult to achieve functions such as free walking,adaptive assistance,and multi-scenario application.In order to solve the above problems in human-exoskeleton coordination,this thesis designs two kinds of human-exoskeleton cooperative control frameworks for assisting walking of paraplegia patient and the elderly respectively.The researches on human-machine-interface,gait planning and coordination strategy,intention recognition are conducted as follows:(1)In terms of human-machine-interface,sEMG and EEG signals are applied to the lower limb exoskeleton robot,and the human-machine interfaces based on sEMG and EEG are constructed respectively.In the application of sEMG signals,a natural human-machine interface based on upper limb sEMG signals was designed for paraplegic assisted exoskeleton by using the coordinated relationship between the wearer's upper limb movements and exoskeleton robot movements to enhance human-machine synergy performance.The sEMG signals and the status of exoskeleton robot was used to identify the wearer's intentions of standing,walking,and stopping,instead of push-button triggering.The experimental results verified the feasibility.In the application of EEG,independent component analysis and channel selection based spatial filter were designed to remove EOG,ECG and other signal noise.Based on the filtered signals,the intention of start and stop was identified.The Effectiveness of the method is verified by the recognition experiments.(2)In terms of motion planning and coordination,an autonomic transfer of center of gravity(COG)of human-exoskeleton system control framework and gait planning method are proposed for a 4 DOF under-actuated paraplegia assisted lower extremity exoskeleton.The proposed method is aiming at reducing the difficulty of COG transferring and increase the safety when using the exoskeleton robot.By designing a state-machine coordinated control framework,which decouples the simultaneous motion of the wearer and the exoskeleton,the effects of interaction between the wearer and the robot is reduced.An inverted pendulum model based online gait planning method and gait trajectory optimization method are designed to generate a safe gait that under the mechanical and electrical constraints of the exoskeleton robot.The proposed gait planning method is able to transfer the COG of the human-machine system automatically in 4 DoF underactuated exoskeleton robot.The proposed method is verified by experiments of simulations,healthy volunteer and disabled volunteer.(3)In terms of human walking feature cognition,three discrete gait phase recognition methods are compared,and a continuous gait phase estimation framework for multi-walking tasks is proposed for elderly assisting exoskeleton robot.In the aspect of discrete gait recognition,the differences of walking gait between different wearers are studied,and a kernel recursive least squares algorithm is introduced to construct a discrete gait phase recognition model using only angle sensors.The kernel recursive least squares algorithm based model is compared with commonly used support vector machine model and multi-layer perceptron neural network model.In terms of continuous gait phase recognition,a novel continuous gait phase estimator that based on an adaptive oscillator network is designed to estimate the continuous gait phase under multi-walking tasks,which overcomes not or slow convergence problem of traditional adaptive oscillator during gait task switching.Considering the stability and simplicity of the sensor in multi-walking tasks,the gait task classifier only uses the angle of the hip joint as the sensing source,and the effectiveness of the proposed method is experimentally verified.In summary,this work studies the natural human-machine interface,gait planning strategies and gait phase recognition methods towards the elderly and disabled assisting lower limb exoskeleton robots,which provides a basis for enhancing the collaborative control effect of the lower extremity exoskeleton robot. |
PDF Full Text Request