The Safety Of Exoskeleton System For The Paraplegic

The exoskeleton for the paraplegic has the attributes of medical devices and high requirements for safety.Therefore,the safety research of the exoskeleton for the paraplegic has attracted much attention,in which the key issues involved are the evaluation of the balance and balance control of the human-exoskeleton system.Aiming at the safety of human-machine system,this dissertation proposed balance evaluation indexes,balance control strategies,and fall protection strategies to improve the safety of the humanexoskeleton system.The main contributions are as followed.Aiming at the problem that the traditional indicators cannot evaluate the dynamic balance of the system accurately,an enhanced stability pyramid index(ESPI)is proposed,which integrates the position,velocity and acceleration of the system's center of mass(Co M)and other information to calculate the urgency of the system imbalance,and represents the system's balance with scalar and vector simultaneously.Simulation results in Gazebo show that this index(with a maximum value of more than 140,while indicating the direction of balance change)is better than traditional one(with a maximum value of less than 40 and unable to indicate the direction of safe change)in terms of sensitivity and representation dimension.Aiming at the difficulty of maintaining system balance,according to whether the system is disturbed or not,the balance control strategy without disturbance,static and dynamic balance control strategies with disturbance are proposed respectively.Among them,the dynamic balance control strategy uses ESPI as the balance evaluation index.The research results show that the balance control strategy without disturbance can control the exoskeleton system to identify and walk up stairs with a height of 12-14.5 cm automatically,and the static balance control strategy with disturbance can restore the balance of the simulated system after being disturbed while standing still within 400 ms based on the center-of-mass prediction.The dynamic balance control strategy with disturbance can adjust the gait parameters of the exoskeleton automatically to maintain the sagittal plane balance during the walking process of the system.Aiming at the problem of no effective fall protection when the system falls,to reduce/avoid fall injuries in critical parts,passive fall protection strategies based on a airbag automatically triggerd by the inertial measurement unit,active fall protection based on human fall experience and soft actor-critic reinforcement learning algorithm are proposed.Among them,four indicators of impact damage,vertical kinetic energy,head impact and control cost are defined as the reward items of reinforcement learning algorithm.The research results show that the passive protection strategy can be fully triggered within 800 ms to reduce the head injury of the system;the experience-based protection strategy can reduce the head injury to 0 when falling back and forth;the protection strategy based on the reinforcement learning algorithm can be applied to the fall protection of the simulated system with the weight of 40-90 kg in the tilt range of 0-15°from front to back and 0-10°from left to right.The above research have been experimented on an exoskeleton for the paraplegic.The balance evaluation indexes and balance control strategies can evaluate and control the balance of the human-exoskeleton system,and the fall protection strategies can reduce injuries of critical parts.At the same time,this dissertation builds an cloud-brain platform based on the exoskeleton for the paraplegic,and accomplishes the safety evaluation of the human-exoskeleton system with the help of the foot pressure and isolated forest.