BRAIN COMPUTER INTERFACE APPROACHES TO CONTROL MOBILE ROBOTIC DEVICES

This paper presents and compares two approaches for brain computer interface to steer a wheelchair, namely a new visual based P300 paradigm consisting of 8 arrows randomly intensified used for direction selection and a motor imagery paradigm for discrimination of three commands. Classification follows Bayesian and Fisher Linear Discriminant approaches both based on prior statistical knowledge. Results in P300 paradigm reached false positive and false negative classification accuracies above 90%. Motor imagery experiments presented about 70% accuracy for left vs. right imagery and imagery vs. non-imagery.