Fig 1.
Frontal view and road geometry of the driving simulator.
A: Participants drove a car on a left lane of a monotonous expressway at night, following a leading car driven at 95 km/h. B: The black thick line indicates the road structure for one driving session. The participant’s car ran from Start to End. The arrow lines with Trial1 -3 indicate the road segments for SDLP measurement, and physiological measures (i.e., PERCLOS, SEM, and EEG). The length of the segment is suitable for driving the car for 5-minutes, with a maintained speed of 95 km/h. KSS1—3 indicate the positions of presenting a synthesized voice of “KSS,” which requests the response of the KSS to the participant.
Table 1.
Typical behaviors for observer rating of drowsiness.
Fig 2.
Time course of drowsiness measures.
Each drowsiness measure averaged across 17 participants, was plotted against trial order. The drowsiness measures are, A: observer ratings of drowsiness (ORD), B: Karolinska sleepiness scale (KSS), C: standard deviation of lateral position (SDLP), D: percentage of eye closure (PERCLOS), E: the percentage of time occupied by slow eye movements (SEM), F: electroencephalographic (EEG) alpha power, and G: EEG theta power. Error bars indicate standard error. Lines connecting markers indicate the trials in one session. Gray area in each plot indicates lunch time.
Fig 3.
Relations between the ORD and other drowsiness measures.
A: KSS, B: SDLP, C: PERCLOS, D: Percentage of time occupied by SEM, E: EEG alpha power and F: EEG theta power. Each marker indicates the data of one trial. Each color indicates a participant.
Table 2.
Correlation coefficients after Fisher’s z-transformation between drowsiness measures.