Vehicle automation may improve road safety because most crashes are caused by human error. However, reliable autonomous vehicles in which the ‘driver’ never has to take over the wheel are still in the distant future. At first, cars will only drive themselves under certain conditions. When these conditions are not met the car will prompt the driver with a
take-over request (TOR).
In this study, a simulator was used to investigate whether drivers could spot latent hazards in the road and traffic scene immediately after manual driving had become obligatory (that is, after a TOR). A latent hazard is a traffic situation in which precursors of a hazard are visible. The number of latent hazards spotted was assumed to be an indicator of the driver’s situation awareness. Latent hazards may or may not materialize into acute threatening situations in which a crash is very likely. The simulator drive contained eight latent hazards (that did not materialize). The default mode was automated driving; however, when participants approached a latent hazard a TOR was prompted. Fifty participants were randomly assigned to one of two groups: in the 4 s group the TOR was prompted four seconds before participants had to resume manual driving, and in the 6 s group the time allowed was six seconds. All participants played a computer game on a tablet while the simulator car was in automated mode. Their gaze directions were recorded with an eye tracker during the entire trip of approximately 45 min.
On average, participants in the 4 s group gazed at 29% (SD = 16.5) of the latent hazards, and participants in the 6 s group gazed at 47% (SD = 17.8). A generalized linear model for binomial data revealed that this difference was significant (X2 (1, N = 43) = 11.23, p < .01). This result indicates that drivers need time to develop situation awareness when they have to resume driving. It could be that, after having been out of the loop, drivers have to construct a mental representation of the traffic situation before they can recognize latent hazards.