A Development Of A Assist-As-Needed Control System Of A Lower Limb Exoskeleton For The Elderly

The physical function of the elderly gradually declines with age,and reduced motor function leads to a decrease in their normal daily activities and quality of life.Exoskeleton robots are one of the main measures to ensure the safety and efficiency of elderly people's walking to improve their independent living ability.However,currently available assisted walking exoskeletons for the elderly mostly focus on reducing the user's metabolism and do not consider the user's active participation to maintain their residual locomotor abilities.Therefore,thesis paper proposes a lower limb exoskeleton assist-as-needed assisted control strategy to improve the participation of the elderly wearer,which solves two major problems in the current exoskeleton assist-as-needed assisted control strategy to achieve an assisted walking exoskeleton system based on the wearer's locomotor ability: 1)how to achieve a simple and feasible locomotor ability assessment of the elderly in walking;2)how to avoid the exoskeleton control focusing too much on reducing the trajectory tracking errors while ignoring the differences in gait trajectories between users.The main research work of thesis paper includes.1.To address the problem of low practicality and generality of human joint moment estimation in existing exoskeleton systems,a knee joint moment model for the elderly based on the human center of mass is proposed.The inertial measurement unit is used to obtain the body center-of-mass motion information,and the BP neural network is used to train the model between the center-of-mass motion information and the knee joint moment,and the average root mean square error is 0.12 Nm/kg when comparing the model estimated moment with the gold standard calculation result.2.To address the problem that the current exoskeleton control focuses too much on reducing the trajectory tracking error and ignores the differences in gait trajectories between users,an assist-as-needed strategy based on the trajectory tracking error virtual tunnel was proposed.The reference trajectories and reference torques were selected by statistical method.The mean standard deviation of joint angles of healthy people in the gait cycle was taken as the virtual tunnel boundary at the reference step speed,and the reasonable auxiliary torques were planned by combining the reference torques to estimate the torque tracking error and the virtual tunnel boundary.Through gait phase recognition and adaptive torque controller,the motion state of the human-machine system and the torque command value of the driver can be obtained to ensure the accurate and timely application of the auxiliary force.3.A knee-joint exoskeleton platform was built to experimentally test the above model and control strategy.The electrical control system of the exoskeleton consists of a sensing system for motion perception,a motor drive system for output assistance,and a control system for torque planning and control.The results of the experimental tests show that the exoskeleton sensing system is feasible for torque estimation and gait phase recognition,and the control strategy achieves assist-as-needed assistance characteristics.The effective validation of the assist-as-needed assistance strategy of the lower limb exoskeleton for the elderly proposed in thesis paper on a practical platform lays the technical foundation for the assisted living for the elderly.