ABSTRACT
The EMG signal has additional potential to augment more traditional control methodologies, such as automatic speech recognition. For example, recent work has shown that certain facial EMG activity is highly correlated with speech. Follow-on studies at the University of New Brunswick in Canada demonstrated that a neural network could achieve up to 95% accuracy in recognizing the spoken words “zero” through “nine” based on EMG signals recorded from electrodes mounted in a flight oxygen mask.
EEG recorded from the surface of the scalp represents a summation of the electrical activity of the brain. Although much of the EEG appears to be noise-like, it does contain specific rhythms and patterns that represent the synchronized activity of large groups of neurons. A large body of research has shown that these patterns are meaningful indicators of human sensory processing, cognitive activity and movement control. In addition, EEG patterns can be brought under conscious voluntary control with appropriate training and feedback. Although current EEG-based systems represent fledgling steps toward a “thought-based interface,” significant long-term development is required to reach that goal. EEG-based control research is primarily confined to laboratory systems and is based on two general approaches:
