Crossing the line: Impact of pedestrian group behavior on individual crossing decisions in AV interactions

Abstract

Research and development in automated driving are steadily progressing, with an increasing number of vehicles equipped with these systems participating in road traffic. Ensuring safe and robust interactions between automated vehicles and other road users is essential, especially in urban environments where interactions between vehicles and groups of pedestrians are common. Literature shows that external human–machine interfaces on automated vehicles might serve as a possible solution for such communication. Pedestrian groups have not been adequately considered in research on road crossings in front of automated vehicles equipped with external human–machine interfaces. This study investigates whether and how the behavior of a pedestrian group affects a single pedestrians behavior and perception of crossing in front of an automated vehicle equipped with an external human–machine interface. A virtual reality study with 44 participants revealed that pedestrians adjust their crossing behavior based on the behavior of others, leading to both improved efficiency and potential safety risks during crossings. Subjective measures revealed increased trust in automated vehicles with repeated exposure. However, trust remained essentially unchanged across different behaviors of the pedestrian group. After the experiment, almost three-quarters of the participants reported a subjective influence of the pedestrian group. Objectively, participants crossed earlier and more confidently when the group began to cross the street, improving efficiency in scenarios where the automated vehicle stopped. The same behavior raises safety concerns in scenarios where the automated vehicle is about to pass, as participants follow the pedestrian group early. These findings highlight the need for automated vehicle communication systems to consider group dynamics to ensure safety and efficiency.