PPGINF PÓS-GRADUAÇÃO EM ENGENHARIA DA INFORMAÇÃO FUNDAÇÃO UNIVERSIDADE FEDERAL DO ABC Phone: Not available http://propg.ufabc.edu.br/ppginfo

Banca de QUALIFICAÇÃO: CAIO LIMA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : CAIO LIMA
DATA : 01/12/2021
HORA: 08:30
LOCAL: Remoto
TÍTULO:

Arm and Forearm Prosthesis Simulator Using Electromyographic Signals


PÁGINAS: 43
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Elétrica
RESUMO:

Approximately 7% of the Brazilian population has a motor disability, and this audience generally has greater social vulnerability. Aiming at mitigating these impacts, some assistive technologies such as upper limb prostheses have been developed, especially those that use surface electromyographic signals (sEMG).

Although several of these prostheses have a high success rate (above 90\% in some cases), they have a high average price (U$50,000) and generally only discreetly detect positions, thus limiting their functionality. In view of these difficulties, it is intended to develop a low cost simulator of forearm prosthesis controlled by sEMG signals that adequately represents the continuous movement of the flexion and extension movements of the forearm.

To this end, a low cost device was used to capture of sEMG signals (in this case the device used was the Myo, developed by the company Thalmic Labs, which cost around U$200.00) , the signals were properly filtered (IIR high-pass filter with 20Hz cutoff frequency and filter rejects band with cutoff frequency of 60Hz), as well as its windowing (250ms), extraction of its characteristics (DWT-db7 stratified into 4 levels mapped by the MAV parameter) together with specific classifiers (LDA, Decision Tree, K-NN , Gauss Naive Bayes linear-SVM) and its correct integration with graphics and control software (Gazebo and ROS2, respectively developed by emp Resa Open Robotics).

It was found, for the tested cases, that the most suitable classifiers for the classification of sEMG data were LDA, Gaus-Naive-Bayes and K-NN, obtaining on average errors of less than 20\%, also if represented this simulator by a RR planar robot with two degrees of freedom. \textbf{Perspectives:} It is intended, in view of the preliminary results and based on literature studies, to integrate Computer Vision and Neural Network.


MEMBROS DA BANCA:
Presidente - Interno ao Programa - 1761107 - RICARDO SUYAMA
Membro Titular - Examinador(a) Interno ao Programa - 1946319 - DIOGO COUTINHO SORIANO
Membro Titular - Examinador(a) Externo ao Programa - 2123666 - FERNANDO SILVA DE MOURA
Membro Suplente - Examinador(a) Interno ao Programa - 2334927 - ANDRE KAZUO TAKAHATA
Notícia cadastrada em: 04/11/2021 13:27
SIGAA | UFABC - Núcleo de Tecnologia da Informação - ||||| | Copyright © 2006-2021 - UFRN - sigaa-1.sigaa-1