Drowsiness Detection on Electroencephalogram using Evolutionary Algorithms

C. Conceição, J. Costa, and A.C. Rosa (Portugal)


Drowsiness Detection, Electroencephalogram,Evolutionary Algorithms.


This work presents an effective real time system to detect drowsiness, based on the analysis of electroencephalogram signal bands. It is a low complexity system, with the capability of being personalized for each user and completely independent in the detection process. Tests have been done with pre-recorded MWT data, featuring ten different subjects each with up to four nap attempts, resulting in an average anticipation of two minutes and thirty-four seconds, enabling the implementation of early warning systems. Four real-time tests were performed showing detections when the subjects were instructed to relax. This real time detection method allows a warning to be issued, making it suitable for practical implementation on risk situations. Its low complexity analysis over EEG, along with the anticipation results, present a practical implementation of drowsiness detection device.

Important Links:

Go Back