A method for predicting the risk of virtual crashes in a simulated driving task using behavioural and subjective drowsiness measures

Development of single-channel electroencephalography signal analysis model for real-time drowsiness detection : SEEGDD

Drowsiness detection is essential in some critical tasks such as vehicle driving, crane operating, mining blasting, and so on, which can help minimize the risks of inattentiveness. Electroencephalography (EEG) based drowsiness detection methods have been shown to be effective. However, due to the non-stationary nature of EEG signals, techniques such as signal transformation and sub-band extraction are increasingly being used to automatically classify awake and drowsy states. Most of these procedures require high computation time. In this paper, analytical and single-feature computation are used to propose a single-channel EEG-based drowsiness detection method to overcome this. Physionet sleep dataset and the simulated virtual driving dataset were used to test the proposed model. When compared to existing work, the proposed approach yields better results.

Assessment of Driver's Drowsiness Based on Fractal Dimensional Analysis of Sitting and Back Pressure Measurements

The most effective way of preventing motor vehicle accidents caused by drowsy driving is through a better understanding of drowsiness itself. Prior research on the detection of symptoms of drowsy driving has offered insights on providing drivers with advance warning of an elevated risk of crash. The present study measured back and sitting pressures during a simulated driving task under both high and low arousal conditions. Fluctuation of time series of center of pressure (COP) movement of back and sitting pressure was observed to possess a fractal property. The fractal dimensions were calculated to compare the high and low arousal conditions. The results showed that under low arousal (the drowsiness state) the fractal dimension was significantly lower than what was calculated with high arousal. Accumulated drowsiness thus contributed to the loss of self-similarity and unpredictability of time series of back and sitting pressure measurement. Drowsiness further reduces the complexity of the posture control system as viewed from back and sitting pressure. Thus, fractal dimension is a necessary and sufficient condition of a decreased arousal level. It further is a necessary condition for detecting the interval or point in time with high risk of crash.