Using Drivers' Stop/Go Decisions in Driving Simulator to Assess Rear-End Crash Risk at Signalized Intersections

Advanced driving simulators have a potential to be utilized to assess crash risks of driving behaviors, traffic surroundings, and highway designs. This study focused on investigating if drivers' stop/go decisions due to signal change in the simulator's virtual reality can be used to assess rear-end crash risk at the signalized intersection. A signalized intersection with as many important features (including roadway geometries, traffic control devices, intersection surroundings, and buildings) was replicated into a high-fidelity driving simulator. Using the virtual signalized intersection, a driving simulator experiment was conducted to test driver's stop/go decisions at two approaches of the intersection that respectively showed higher and lower rear-end crash risks in the crash history analysis. The experiment results showed that the variability in drivers' stop/go decisions at the higher crash-risk location is higher than that at the lower crash-risk location. Further, through modeling rear-end conflicts based on driver's stop/go probability as a function of potential time to intersection, it was found that the rear-end crash tendency at the two locations displayed by the driver's stop/go behavior in driving simulator is consistent with the finding from the trend in crash history analysis. This research supported that driver's stop/go behavior in a driving simulator can be utilized to assess rear-end crash risk at signalized intersections to seek effective engineering countermeasures to lower crash rates for the high-risk locations.