Background - Personal life style and work demands are leading to more sleep-deprivation than ever before, which can have a negative impact on health, safety and performance, and even lead to deadly consequences. In particular, drowsiness and fatigue have detrimental effects on driving performance and road safety. Aims - The goal of this study was to investigate the characteristics of eye movements and blinking as a correlate of drowsiness, and their relationships to behavioural and physiological measures of vigilance. Methods - Eye tracking data were collected using infra-red-based systems in two different experiments: a sustained vigilance task (SVT) and a simulated driving task (SDT). A total of 22 subjects participated in this study (15 subjects in the SVT and 7 subjects in SDT). In the SVT experiment, reaction times to the psychomotor vigilance task visual stimuli were used as the baseline for evaluation of the drowsiness detection technique. In contrast, electroencephalogram (EEG) signals were used in the SDT experiment to assess the performance of the eye-tracking-based methodology for drowsiness detection. A set of 25 specific features were extracted from eye tracking data in both experiments, where a non-linear support vector machine (SVM) classifier was employed for binary classification of the state of vigilance. Results - The evaluation results revealed that the state of vigilance was detected with a high average accuracy (ranging from 83% to 93%) in different scenarios/sessions considered for this study. Discussion & Conclusion - Altogether, the results of this study show the potential of the proposed machine learning based methodology forreliable and non-intrusive assessment of drowsiness. These results verify a highcorrespondence between extracted eye tracking features and behavioural/physiologicalmeasures of vigilance (here, reaction time/EEG). Ultimately, this research would lead todevelopment of ubiquitous and automated real-time detection of the state of vigilance in driverswith the goal of improving road safety.
展开▼