Research On The Gait Planning And Pattern Recognition Of BRHL

In the development of intelligent prosthesis,a lot of performance test experiments are needed.However,due to the experimental test of the workload and the requirements of good repeatability,people with disabilities wearing artificial experiments are not realistic.Therefore,research group will bipedal walking robot and intelligent prosthesis combination,is proposed for heterogeneous legs robot,biped robot with heterogeneous legs,BRHL)mode,in order to develop a robot used for testing the performance of the prosthesis.Artificial leg simulation of the health legs of the disabled,bionic legs simulation of the disabled wear smart prosthesis.Artificial legs are used to generate humanoid gait and provide gait following target for bionic leg.This paper studies the following contents of gait planning and pattern recognition of biped robot with heterogeneous legs:(1)Human lower limb gait information acquisitionThe purpose of artificial leg gait planning is to obtain the gait of directional human gait,and it is necessary to study the gait of human normal walking.In this paper,the joint angle data of the lower limbs of the human body is obtained by the OpenSim software.(2)Gait planning for artificial legsThe gait planning of artificial leg was studied by Pattern Generator Central(CPG).The parameters of Hopf model are analyzed based on single parameter analysis.CPG network is built by Hopf oscillator,and the CPG network is described by differential equation.Based on genetic algorithm,the parameters of CPG network are determined,and the output CPG curve of Matlab is adopted.The simulation of robot walking based on ADAMS is carried out,and the feasibility of CPG is verified.(3)BRHL gait information acquisition system to buildThe gait planning of bionic leg is studied,and the method of bionic leg is studied.Through the construction of artificial leg gait information acquisition and bionic leg information feedback system,the bionic leg to obtain artificial leg movement information,and the hardware required to purchase.(4)Artificial leg walking pattern recognitionGait pattern recognition of knee joint based on the ground running and the stairs of the upper and lower stairs.First the gait data using wavelet analysis method for preprocessing,normalization method is adopted to eliminate the singularity in the data,using principal component analysis(principal component analysis,PCA)gait data of feature extraction,finally,the PCA based template matching recognition model of walking.(5)Walking pattern recognition based on neural network technologyA multi sensor information fusion method is adopted for the recognition of walking pattern,and neural network is adopted for the fusion.The recognition rate of walking pattern recognition based on BP neural network is more than 96%.BP neural network technique is used to predict the gait information of the artificial leg during exercise.