Font Size: a A A

Research On Gait Symmetry And Gait Pattern Of Biped Robot With Heterogeneous Legs

Posted on:2014-09-08Degree:MasterType:Thesis
Country:ChinaCandidate:D C SongFull Text:PDF
GTID:2308330473453763Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
Biped Robot with Heterogeneous Legs is a novel robot model, which is integration of the research of biped robot and intelligent prosthesis. It can simulate the situation of the amputee who wears intelligent prosthetics, do a mass of repeatability and variety test study of intelligent prosthesis, and provided a scientific ideal platform for the research on intelligent prosthesis.The paper investigates the gait pattern of BRHL in detail on the basis of the current research of the biped robot and prosthesis, the system components of BRHL and the significance of BRHL study. We mainly focused on three parts:the gait symmetry optimization of BRHL, the gait pattern recognition of BRHL and the establishment of human gait model. The difference of the right side and left side knee joint’s structure will bring bad effect on the BRHL’s gait symmetry. We propose a genetic algorithm and the BP neural network approach to solve the poor symmetry of BRHL’s step length. At first, we transform the right knee joint angle of subject through initialized weights BP neural network and calculate the step errors in the ADAMS simulation environment. And then we use the step errors as the genetic individual fitness evaluation index to optimize the weights of BP neural network. After several iterations, the symmetry of six objects’ step length is within the normal range, which proves the feasibility of this program.In real life, the amputee who wears intelligent prosthetics will encounter a variety of road conditions. Hence, we need intelligent prosthesis can identify the different road conditions and gait pattern according to the motion information of the wearer. In this paper, we use MTi to collect acceleration signals of lower leg. After feature extraction on these signals using wavelet decomposition, we establish nine observed objects’ hidden Markov (HMM) models in five different road conditions. Finally, we verify it with test data, the experimental results show that the proposed method in this paper can identify gait pattern with high accuracy.After the intelligent prosthesis identifies the road condition and gait patterns, another problem is how to adjust prosthetic gait according to the specific circumstances. Firstly, we carry out a quantitative analysis of gait and calculate two gait parameters:step length and step frequency. Then we express the knee angle with fewer characteristic parameters。Finally we use the RBF neural network model to establish mathematical relationship between gait patterns, step length, step frequency, and characteristic parameters of knee angle, and predicted the knee angle in different situations, achieving good results.
Keywords/Search Tags:Intelligent prosthesis, Optimization algorithm, HMM, Gait pattern recognition, Gait adjustment
PDF Full Text Request
Related items