EEG Signal-Based Movement Control for Mobile Robots

Although wheelchair with joystick control is available, people whose hands are paralysed cannot use the joystick and need other forms of assistance to move. This article presents the design and analysis of a mobile prototype robot control using a single-electrode commercial electroencephalogram (EEG) headset. We examine the possibility of detecting P300 and blink signal for use as an input to control a prototype robot. From the captured EEG signals, P300 and non-P300 are classified using an artificial neural network. In another experiment, we classify signals captured during the intentional blink of the eye and signals where there is no blink. Also, we classify when the user intentionally blinks two, three and four times. From the experiments, we found that P300 cannot be successfully detected with a single dry electrode on Fp1 position. Additionally, we found that signals which contain blink and those which do not contain blink can be classified using an artificial neural network. We also found that different number of blinks can be classified using an artificial neural network. Different number of blinks is used to move forward, turn left and right. The model trained to classify between blink and nonblink signals is used to apply the brake. Experiments performed have shown that using a single-electrode commercial headset and blink of the eye, a user can successfully control the prototype to reach a predefined destination.