Pedestrian-Autonomous Vehicle Interaction: What's the Problem? Autonomous vehicles (AVs) are no longer a distant future concept but are rapidly becoming a reality. While the advancement of this technology promises everyday convenience, it also raises a crucial question: 'Can these innovative vehicles truly operate safely on our roads?' Especially when people react in unexpected ways. A recently published study has further illuminated the complexities of this subtle interaction. Research conducted jointly by the University of Glasgow and the Korea Advanced Institute of Science and Technology (KAIST) revealed that running pedestrians tend to react more dangerously to autonomous vehicle signals than walking pedestrians. The study, titled 'Running into Traffic: Investigating External Human-Machine Interfaces for Automated Vehicle-Runner Interaction,' was published and reported by Canadian Running Magazine and The Daily Texan. This suggests that communication methods between pedestrians and autonomous vehicles need to be examined from a more human-centric perspective, beyond mere technical issues. These findings offer crucial insights into understanding the complexity of the environment in which autonomous vehicles operate, going beyond simple curiosity. The study involved 24 participants in a simulation conducted using augmented reality (AR). To avoid exposing people to real traffic situations, the research team opted to create an augmented reality environment to replicate virtual scenarios. Participants wore AR equipment and reenacted crossing a road at a virtual intersection, experiencing various scenarios while alternating between walking and running. Each participant was instructed to approach the intersection and cross when they deemed it safe, encountering a simulated autonomous vehicle approaching at that moment. The study observed how pedestrians communicated with vehicles using autonomous vehicles equipped with external Human-Machine Interfaces (HMI). External HMIs are light displays placed on the exterior of autonomous vehicles to replace non-verbal cues such as a driver's eye contact or hand gestures. Specifically, vehicles appeared with no signal, or displaying one of two external HMIs. The first was a simple red and green traffic light indicating whether the vehicle was stopping, and the second was a moving blue light indicating the vehicle's next movement. Participants had to interpret these signals to safely cross the intersection. As a result, while walking pedestrians generally acted cautiously, running pedestrians tended to react more dangerously to autonomous vehicle signals. Walking pedestrians interpreted AV signals relatively carefully, but the situation was somewhat different for those who were running. Running pedestrians often perceived autonomous vehicle signals as dangerous, a tendency that was more pronounced when relying on the simplicity of the signals. These findings suggest that in an autonomous driving environment, where pedestrians must rely solely on mechanical signals without the non-verbal cues of a driver, pedestrians in a hurry may face greater risks. This study is not merely a joint effort by the University of Glasgow and KAIST but also holds significant meaning within a broader academic context. It aligns with the arguments of researchers at the University of Texas, who emphasized that safety certification for autonomous driving systems must test for diversity and be continuously evaluated. In other words, the safety of autonomous vehicles cannot be guaranteed by a one-time test; continuous verification considering various pedestrian behavior patterns and situations is necessary. Key Findings from KAIST's Korean Research So, why do running pedestrians struggle more to interpret autonomous vehicle signals? According to the research findings, this can be explained by a complex interplay of several factors. Running pedestrians experience greater time pressure compared to walking pedestrians and may lack the cognitive capacity to adequately process environmental signals while moving at a fast pace. Furthermore, while running, individuals must focus on their body's movement itself, making it highly likely that their attention, needed to interpret vehicle signals, will be divided. Therefore, simple autonomous vehicle signals may not provide sufficient information in such complex situations. This implies that the design of autonomous vehicle external HMIs must evolve beyond simple visual signals to a more sophisticated approach that considers various pedestrian states and behavior patterns. In conventional vehicles with drivers, non-verbal communication with pedestrians occurs through eye contact, hand gestures, and subtle adjustments to vehicle speed. However, since these human elements are removed in autonomous vehicles, a clear and intuitive signaling system to replace them is essential. A system that can effectively operate even for pedestrians in urgent or stressful situat
Related Articles