Abstract
We previously showed that, in daytime, drivers responded to stopped and slow-moving lead vehicles gradually, rather than suddenly when optical expansion rate reached .006 rad/s. Here we examined whether this occurs in (simulated) nighttime driving. We recorded expansion rate and lead vehicle’s time on screen at six driver inputs (from early- release accelerator to late- slam on brakes) to stopped and slow-moving vehicles. Two groups of participants responded differently to slowmoving vehicles; one group released the accelerator, and one group applied unanticipated-level braking. Responses to stopped vehicles occurred when expansion rate reached .005 rad/s, close to the .006 rad/s reported for immediate hazard perception. Time on screen was longer in daytime than nighttime for slow-moving vehicles for later inputs, suggesting that daytime drivers see the vehicle sooner and better estimate how much time they have to respond. Effects of cell phone conversation and expectancy were not significant.
Of the 1,457,155 police-reported rear-end collisions in 2020, it is estimated that 70 percent were with stopped lead vehicles and 30 percent were with slow-moving lead vehicles (NHTSA, 2022; Knipling et al., 1993). Prior research suggests that drivers will not respond to a lead vehicle until they perceive it as an immediate hazard— when the optical expansion rate reaches 0.006 rad/s (Muttart et al., 2005). However, our recent driving simulator study showed that, with daytime scenes, drivers responded gradually in phases rather than suddenly to stopped lead vehicles before reaching this threshold (Oliver et al., 2022). Here we examined whether this same pattern of results would occur in nighttime scenes. We also examined effects of expectancy and simulated cell phone conversation which in the daytime study resulted in later perception of a hazard in less demanding conditions, and no effect, respectively.
Forty participants used a medium-fidelity STISIM DriveTM Simulator that displayed nighttime rural scenes of a two-lane split highway and were instructed to drive as they normally would, obey all rules of the road, and avoid crashes. There were twelve unique scenes which included six critical events: three stopped (0 mph) lead vehicles and three slow-moving (45 mph) lead vehicles. The lead vehicle’s optical expansion rate and amount of time it was displayed on the screen were recorded at six driver inputs ranging from early to late: begins to release the accelerator, fully releases the accelerator, starts to press the brake, comfort-level braking, unanticipated-level, and brake pedal pressed more than 90%. Half of the participants were randomly assigned to complete the last letter task to mimic cell phone conversation.
Results suggested that optical expansion rate for the stopped lead vehicle increased for later driver inputs but was not indicative of a gradual response at night as we had found for daytime. However, time on screen was significantly smaller for the first driver input compared to all other driver inputs suggesting that drivers responded based on time or a variable linearly related to time rather than optical expansion rate.
In addition, two groups of drivers responded differently to the slow-moving lead vehicle: those that responded with earlier driver inputs (only released the accelerator), and those that responded with later driver inputs (depressed the brake pedal).
Perception of an immediate hazard (mean of 0.005 rad/s which approximates Muttart’s 0.006 rad/s) was closely associated with a stopped lead vehicle at nighttime, and drivers responded to it with unanticipated level braking. This contrasts with our daytime study in which perceptual events preceding responses did not correspond to an expansion rate of .006 rad/s.
For the slow-moving lead vehicles and later inputs (braking responses), time on screen was significantly longer in daytime than nighttime suggesting that daytime drivers see the vehicle sooner and are better able to estimate the time they have to respond. The effects of cell phone conversation and expectancy were not significant.
This research is applicable to the forensic analysis of roadway collisions involving stopped vehicles. Specifically, time of day should be considered when examining a driver’s response to a stopped lead vehicle.
