Sleep in highly automated driving: Takeover performance after waking up

Abstract

Takeover performance in automated driving is subject to investigation in the context of a variety of driver states such as distraction or drowsiness. New driver states will emerge with increasing automation level with drivers potentially being allowed to sleep while driving a highly automated vehicle. Still at some point during a drive, drivers will be required to or voluntarily take back control of the vehicle. A simulator study was conducted to investigate drivers’ ability to take over the vehicle control after sleeping. In a within-subjects study design N = 25 test drivers completed a drive using a highly automated driving system a) during day time after a full night of sleep and b) early in the morning after a night of partial sleep deprivation. During the second drive, sleep was measured in drivers according to the American Academy of Sleep Medicine (AASM) standard using electroencephalography (EEG). In total, the participants had to handle four takeover requests (TORs) from the system, two while being awake (day drive) and two when being awakened from sleep stage N2 (morning drive). The objective criticality of the situations was assessed performing the Takeover Controllability rating (TOC-rating). The results indicate that the applied takeover time of 60 s was sufficient for drivers to reengage in driving after sleeping. Reaction times were extended by about 3 s after sleep compared to the wake condition. Takeover performance assessed with the TOC-rating however was clearly worse after sleep than after wakefulness which was also reflected in the drivers’ subjective perception of the criticality of the situation. Further research is needed on how to deal with performance impairments after waking up from sleep during automated driving.