Navigation of quadruped multirobots by gesture recognition using restricted boltzmann machines

This article discusses a method that performs gesture recognition, with the objective of extracting characteristics of the segmented hand, from dynamic images captured from a webcam and identifying signal patterns. With this method it is possible to manipulate simulated multirobots that perform spec...

Descripción completa

Detalles Bibliográficos
Autores: Saraiva, Arata A., SANTOS, D. B. S, MARQUES JUNIOR, F.C.F, SOUSA, JOSE VIGNO M., FONSECA FERREIRA, N. M., Valente, Antonio
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2018
País:Panamá
Institución:Universidad Tecnológica de Panamá
Repositorio:Repositorio Institucional de documento digitales de acceso abierto de la UTP
Idioma:español
OAI Identifier:oai:ridda2.utp.ac.pa:123456789/5787
Acceso en línea:http://revistas.utp.ac.pa/index.php/memoutp/article/view/2013
http://ridda2.utp.ac.pa/handle/123456789/5787
Access Level:acceso abierto
Palabra clave:Manipulates; multirobots; restricted Boltzmann Machines
Descripción
Sumario:This article discusses a method that performs gesture recognition, with the objective of extracting characteristics of the segmented hand, from dynamic images captured from a webcam and identifying signal patterns. With this method it is possible to manipulate simulated multirobots that perform specific movements. The method consists of the Continuously Adaptive Mean-SHIFT algorithm, followed by the Threshold segmentation algorithm and Deep Learning through Boltzmann restricted machines. As a result, an accuracy of 82.2%.