Mobile Robot Control by Neural Networks EOG Gesture Recognition

This paper describes the development of a neural networks gesture recognition system whereby one can control a mobile robot by using the components of his brain wave bio-potentials. Such a system may be used as a control device through human eye-movements, facial muscle, and brain wave bio-potentials. Neural networks are trained to classify EOG data into one of two classes corresponding to two cognitive tasks performed by eight training segments. The operator's forehead bio-potentials can be acquired and processed in Cyberlink as mobile robot control source signals. The computer analyzes an operator's the EEG(electroencephalographic) and EOG (electrooculographic) signals in real time. Neural networks analyze user’s EOG signal in order to discern for the presense of a signal and then decide whether it corresponds to a valid command. In the course of EOG analysis, the neural network checks for example, turning the robot. A trained neural network can effectively recognize user intention, left or right based only on the EOG signal. The experimental results suggest that a mobile robot can be operated by human brain wave bio-potentials with neural networks.