Watch Out Car, He’s Drunk! How Passengers of Vehicles Perceive Risky Crossing Situations Based on Situational Parameters

Automated vehicles promise enhanced road safety for their passengers, other vehicles, and vulnerable road user (VRU). To do so, automated vehicles must be designed to reliably detect potentially critical situations. Humans can detect such situations using context cues. Context cues allow humans drivers to anticipate unexpected crossings, e.g., of intoxicated night owls in a street full of bars and clubs on a Friday night and, consequently, to decelerate in advance to prevent critical incidents.
We used the “Incident Detector” to identify possible context cues that human drivers might use to assess the criticality of traffic situations in which a car encounters a VRU. Investigated potential predictors include VRUs’ mode of transport, VRUs’ speed, VRUs’ age, VRUs’ predictability of behavior, and visibility obstruction of VRUs by parked cars.
In an online study, 133 participants watched videos of potentially risky crossing situations with VRUs from the driver’s point of view. In addition, the participants’ age, gender, status of driver’s license, sense of presence, and driving style were queried.
The results show that perceived risk correlates significantly with age, speed, and predictability of VRUs behavior, as well as with visibility obstruction and participants’ age. We will use the results to include detected influence factors on perceived subjective risk into virtual test scenarios. Automated vehicles will need to pass these virtual test scenarios to be deemed acceptable regarding objective and subjective risk. These test scenarios can support road safety and thus, greater acceptance of automated vehicles.