Researchers in the United Kingdom Used Virtual Reality Technology to Assess Pedestrian Crossing Behaviors with Simulated Automated Vehicles.
Leeds, United Kingdom
Will pedestrians cross the road before an automated vehicle? The effect of drivers’ attentiveness and presence on pedestrians’ road crossing behavior
Summary Information
As automated vehicles (AVs) are deployed, it is important to understand the dynamics of how other road users will interact and respond to them. Interactions between AVs and vulnerable road users such as pedestrians are of particular interest, given that AVs may be considered unusual or untrustworthy by users who have not previously interacted with them. An international team of researchers conducted an experiment using virtual reality (VR) technology to examine how the presence and behavior of a driver in the AV may affect pedestrians' road crossing behavior and perceived risk.
Methodology
Twenty pedestrian participants were asked to cross a single-lane road in an immersive VR environment presented using a head-mounted display, under three different driver behavior scenarios with a repeated measures design:
- a driver in the AV who behaved attentively
- a driver in the AV who appeared to be distracted
- no driver visible in the AV
Participants were instructed to cross between two approaching vehicles if they felt safe to do so. The time gap between vehicles was set to either 3.5 s or 5.5 s. In half of the trials, the second vehicle (the AV) yielded to the pedestrian, coming to a stop 2.5 m before the crossing. In the other half of trials, the vehicles continued driving at a constant speed of 30 km/h without yielding.
After participants engaged in the VR simulation, they were given a survey to gauge the level of trust they had in AVs and their perceived level of control in the scenario. Safety margin was measured based on the amount of time between the pedestrian reaching the opposite side of the road and the passing of the vehicle.
Findings
- The most influential factor in pedestrian crossing behavior was found to be the vehicle's motion cues (gap between vehicles and yielding behavior).
- The pedestrians perceived their risk to be significantly lower when drivers appeared attentive, compared to the other conditions. Since drivers in AVs might engage in other tasks while the AV is operating, there is a need to investigate communication with pedestrians in this particular scenario.
- The safety margin for pedestrians was lower under the shorter time gap condition. Since there was less time for crossing, this resulted in the AV being closer to the pedestrian, especially when the AV did not yield.
