Analyzing Pedestrian Behavior in Augmented Reality - Proof of Concept

With recent advancements in head-mounted displaying technologies, virtual reality pedestrian simulators have become a common tool for traffic safety research. In contrast to field studies, test track studies and traffic observations, simulators enable researchers to analyze pedestrian behavior in a safe and controlled environment. However, creating the necessary virtual environments is time-consuming, especially in terms of meeting today’s expectations regarding graphical level of detail and realism. Furthermore, VR experiments often lack a body representation or require additional sensors to create an avatar. Due to the laboratory setting, VR simulators might fail to convey the feeling of standing on an actual street. In addition, simulators on the one hand and real-world testing on the other hand leave a methodological gap on the reality-virtuality continuum. This paper presents a novel approach for an augmented reality pedestrian simulator. With this simulator, the participant experiences virtual vehicles, augmented on a real scenario, allowing for safe and controlled testing in a realistic setting. In a between-subject design, 13 participants experienced a gap acceptance scenario with virtual vehicles, while 30 participants experienced the same scenario with real vehicles in the same environment. These participants were instructed to initiate a street crossing if they considered that the gap between the two experimental vehicles was safe to cross the street. Results indicate similar, but also offset behavior for both conditions. Lower acceptance rates and later crossing initiation times could be observed in the augmented reality condition. Still, it was shown that augmented reality renders a promising tool for pedestrian research but also features limitations depending on the use case.