Uma interface para geração de comandos em tempo real para um robô móvel com rodas baseada em fluxo ótico e reconhecimento de características faciais

• Ano de defesa: 2015
• Autor(a) principal: Oliveira, Pietro Martins de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual de Maringá; Brasil; Departamento de Informática; Programa de Pós-Graduação em Ciência da Computação; UEM; Maringá, PR; Centro de Tecnologia
• Palavras-chave em Português: Interação humano-computador; Robô móvel; Cadeiras de rodas inteligentes; Brasil.; Intelligent wheelchair; Wheeled mobile robot; Human-machine interface; Brazil.; Ciências Exatas e da Terra; Ciência da Computação
• Link de acesso: http://repositorio.uem.br:8080/jspui/handle/1/2580

Rehabilitation and improvement of living conditions of people with special motor needs offers many research possibilities. This study has the goal of developing a new interface based on optical flow and facial expressions recognition, aiming to enable people who are able to move muscles above the neck to command an electric wheelchair. This interface uses a webcam directed to the face of the user, in order to capture a sequence of images, which are processed aiming to recognize facial patterns that serve as commands for the system to perform some action in real time. This action is sent to a robot that represents the wheelchair. The system is divided into two modules, the first named Computer Vision Module MVC and the other is the Robotic Module MR. The first module applies computer vision techniques to recognize facial expressions and head movements, interpreting the intentions of moving from the user, and thus, sending commands to the robot. The second module is a wheeled mobile robot RMR which has its degree of maneuverability equal to (2,0), which are the same degrees of an wheelchair. This robot receives the commands generated by the MVC by means of a bluetooth communication interface, and then performs the movements that a wheelchair would do in a real situation. Tests were conducted involving various types of environment lighting, and also people with different skin tones. The experimental results show that it is possible to control the movements of an RMR by means of the analysis of images that captures the head movements and facial expressions. Indoor environments showed high rates in the correct recognition, and also very low false command recognition rates.