Classification of visually evoked potentials using extreme learning machine to command an intelligent robot chair with communication aid

In recent years, EEG-based technology has become more popular in producing variety of BMI protocols for wheel chair navigation and communication systems. In this research work, as an initial step towards the development of an intelligent navigation system with a communication aid, a simple EEG data capturing procedure has been introduced using visually evoked potentials. A simple, visually evoked brain wave data acquisition protocol has been developed with seven basic tasks; that can be used for navigating a robot chair and also for communications using the oddball paradigm. In this study, ten participants were participated in the data acquisition protocol. The proposed system records the 8-channel electroencephalography signal while the subject was perceiving the seven different visual tasks. The recorded brain wave signals are pre-processed to remove the interference waveforms and segmented into six frequency bands signals, i.e. Delta, Theta, Alpha, Beta, Gamma 1-1 and Gamma-2. The segmented frequency band signals are used to extract the features using cross correlation algorithm extracted over two consecutive frames. Further, statistical features such as Maximum, Minimum, Standard Deviation and Mean features are formed as features set and used to classify using Online Sequential - Extreme Learning Machine. Further, the classification models are compared and from the results it is observed that mean feature set hits the mean maximum testing accuracy of 85.54 % and Standard deviation feature set hits the mean minimum classification accuracy of 73.47 %.