From Visions to Reality: Investigating the Interplay of Vehicle Kinematics and Light-band eHMI in a Real Vehicle Study

Highly automated vehicles (HAVs) will interact with pedestrians in urban environments. This requires efficient communication tools to ensure mutual understanding. Past research showed that pedestrians mostly used vehicle kinematics to communicate with vehicles, e.g., the vehicle's speed and distance. However, pedestrians required further explicit communication signals when the traffic situation was ambiguous. Light-band external human–machine interfaces (eHMIs) transmit additional explicit communication signals to pedestrians, e.g., the vehicle's yielding intent. To this point, the precise interplay of vehicle kinematics and eHMIs for HAVs has not yet been determined. Nevertheless, previous research showed that combining both means of communication has great potential to increase pedestrian perceived safety and to ensure a safe interaction. Only a little research used real vehicle studies to investigate the interaction between pedestrians and HAVs in a close-to-reality experimental setting. However, this would ensure the transferability of experimental results to future urban traffic. Therefore, this study aimed to address this research gap by investigating the effects of vehicle kinematics, eHMIs, and their interplay in a real-world pedestrian crossing on pedestrians' behaviors and subjective evaluations. In this field experiment, we applied a light-band eHMI on a Wizard-of-Oz test vehicle, an actual vehicle instructed as an HAV. We investigated the effects of vehicle kinematics (early yielding vs. late yielding) and the eHMI status (no eHMI, static eHMI, dynamic eHMI) on pedestrians' crossing behavior and subjective evaluation in a low-speed real-world setting. The static eHMI displayed the vehicle automation status by a static illuminated eHMI. The dynamic eHMI conveyed the automation status and the vehicle's yielding intent. This study focused particularly on the interplay of vehicle kinematics and eHMI status. We assumed that the crossing initiation was shorter when a dynamic eHMI was combined with an early yielding compared to a late yielding in this real-world setting. Moreover, we hypothesized that pedestrians' subjective evaluations are more positive for a well-coordinated interplay of eHMI and vehicle kinematics. The results showed that pedestrians initiated their crossing earlier with dynamic eHMI vs. no eHMI or static eHMI. Furthermore, they perceived a dynamic eHMI as safer and more trustworthy compared to no eHMI or a static eHMI. Combining an early yielding and dynamic eHMI increased participants' perceived safety of the vehicle behavior and trust and improved pedestrians' affective evaluations compared to a late yielding with dynamic eHMI. Overall, this real vehicle study highlighted the importance of implicit and explicit communication signals and their well-coordinated interplay for pedestrians' future interactions with HAVs in a real-world setting.