When Automated Vehicles Encounter Vulnerable Road User Groups: From Stated Preference Surveys to Real-World Experiments

With the rapid deployment of automated vehicles (AVs) in urban traffic, interactions between AVs and vulnerable road users (VRUs) have emerged as a new safety challenge. Existing research primarily examines individual VRUs; however, group interactions in real-world traffic, particularly those involving diverse VRU types and varying group scales, remain understudied, potentially hindering the ability of automated vehicles to safely navigate such complex scenarios. To address this gap, this study investigates how group-level behavioral cues (VRU type, group scale, and crossing behavior) and individual-level heterogeneity (psychological factors and socioeconomic attributes) affect VRU-AV interactions, as well as the gap between interaction intentions and actual behaviors. This study integrates stated preference (SP) surveys and real-world experiments in matched group interaction scenarios to capture both self-reported and observed decisions based on data from 44 participants. The results indicate that within-group preferences and larger group scales significantly increase the likelihood of crossing at both individual and group levels, with different VRU types exhibiting distinct decision patterns. Moreover, the effect of perceived behavioral control is reversed by the group context, while external group behavioral cues override the internal social norms. These findings highlight the need for caution when relying solely on SP data in group interaction studies. This study contributes to the theoretical understanding of VRU group-AV interactions and provides practical insights for policy formulation and AV system design.