Research On Adaptive Gait Planning Strategy Of Lower Limb Rehabilitation Exoskeleton

In contemporary society,with the increase of people with disabilities and the rapid aging of our population,lower extremity rehabilitation exoskeleton has become a research hotspot in the international academic and engineering technology community.However,the gait planning technology,one of the key technologies for lower extremity rehabilitation exoskeleton,needs further research.Considering some problems existing in current gait planning technology,such as the weak gait stability,the less adaptability to patient's motion,the paper focuses on the research of adaptive gait planning technology,and applied into the 10 degrees of freedom lower limb rehabilitation exoskeleton system successfully.The major contents are as followed:First,for the 10-Do Fs lower extremity exoskeleton system,the kinematics and dynamics modeling of the lower extremity exoskeleton is finished.Based on the analysis of the human lower limb physiology and walking characteristics,the D-H parameter method is utilized to analyze the forward and inverse kinematics of the exoskeleton swing leg and support leg,respectively,and the kinematics solution results are verified by Simulink.Based on kinematic,a dynamic model of swinging leg was established using Lagrangian equation method.However,considering the limited accuracy of the dynamic modeling,an overall virtual prototype of the lower extremity exoskeleton is established by MATLAB/ Sim Mechanics.Secondly,in order to achieve adaptive stable walking and improve walking stability,a stability gait planning strategy based on ZMP online correction method is proposed.For the lower extremity exoskeleton table-car model,the ZMP online correction method is used to generate the center of gravity(Co G)movement using the improved preview control algorithm and ZMP reference trajectory.Getting the swinging foot trajectory by traditional polynomial interpolation method,the inverse kinematics solution of the exoskeleton hip,knee and ankle joints are obtained.Then,to achieve learning patient's healthy limb gait adaptively and access comfortablity of the planning gait,an oscillator learning gait planning strategy is proposed.This paper proposes an i DE AHopf oscillator learning method.An improved differential evolution algorithm(i DE)is utilized to optimize the adaptive Hopf oscillator.By learning the healthy limb motion,then map to another disable dysfunctional limb,the exoskeleton hip and knee joints are generated.The results are compared with DE AHopf oscillator,adaptive Hopf oscillator learning method for its learning errors and speed.Meanwhile,this paper also proposes an RBF-DMP oscillator gait learning method,which utilizes the RBF neural network to optimize the DMP oscillator,and then is applied into the online gait generation of lower extremity exoskeleton.Finally,in order to achieve the exoskeleton adapt to the patient's movement and walking characteristics during the assisted walking procedure,an adaptive gait planning strategy is designed,which could learn the three-dimension foot swing trajectory.The strategy consists of the gait planning layer and gait controlling layer,in the gait planning layer,the i DE AHopf oscillator is used to adaptively learn and imitate the foot swinging trajectory according patient's walking history;Based on the adaptive step length estimation method,the next step is estimated during the double support phase;discrete step planner generates the next Co G trajectory combining the current Co G position information.In the gait control layer,the kinematics inverse solution of each joint is accessed,and the fuzzy immune PID controller finishes the joint trajectory tracking.The balanced gait adaption technology is applied to tune the joint trajectory to improve gait stability.At last,the lower extremity exoskeleton Sim Mechanics simulation is established to verify the overall effectiveness of the gait planning strategy.