Interface homem-máquina para uma cadeira de rodas baseada no movimento ocular e piscadas de olhos

• Ano de defesa: 2007
• Autor(a) principal: Cavalieri, Daniel Cruz
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Human-Machine Interfaces; Myoelectric Signals; Digital Signal Processing; Pattern Recognition; Video-Oculography; Rehabilitation Engineering; Videooculografia; Engenharia de Reabilitação; Reconhecimento de padrões; Interfaces Homem-Máquina; Sinais mioelétricos; Processamento digital de sinais; Interação homem-máquina; Processamento de sinais - Técnicas digitais; Reconhecimento de padrões óticos; Reabilitação; Sistemas homem-máquina; Engenharia Elétrica; 621.3
• Link de acesso: http://repositorio.ufes.br/handle/10/4055

This work has the purpose of developing human-machine interfaces for people with disabilities in order to improve their life-quality, providing communication with a PDA (Personal Digital Assistant) and then controling a robotic wheelchair. Both interfaces have been implemented, one based on the myoelectric signal intentionally generated by eye blinks, and another based on the eye-tracking through video-oculography. In the first case, classical techniques of digital filtering were used, in order to attenuate the noise which corrupts the myoelectric signal. Techniques of pattern recognition have been implemented for processing with eye blinks. Initially, a simplified system for the analysis of activity, based on the variance of the signal, was implemented to detect the blink. Later, with the purpose of classification and reduction of the stochastic effect of the signal, artificial neural networks were used, whose structure is composed of a input layer, an intermediate and an output layer, trained with Bayesian Regularization, Resilient Backpropagation and Scaled Conjugate Gradient algorithms. In order to expand the opportunities of people with severe disabilities and resolve problems found at the interface based on eye blinks, a system for eye-tracking was also implemented. This interface employed techniques for image processing allied to the application of Randon Circular Transformed of Hough. Furthermore, in order to increase the resolution of the system, a Kalman filter was applied to the eye center coordinates which was determined through the centroid of the region of interest from the iris location. Problems with this interface were also evaluated, which are caused by the difference in light intensity and the eye blinks. In both interfaces, results have proved very satisfactory.