Automated Vehicle-Pedestrian Interaction: The effects of communication strategies and repeated exposures

This research investigated pedestrians’ crossing decisions and attention allocation, as indicated by head and gaze movements, in interactions with automated vehicles (AVs) over repeated exposures, aiming to provide insights for developing safe and effective AV communication strategies. The work addressed questions relating to (i) drivers’ kinematic cues in different contexts and their impact on pedestrians’ crossing decisions for proposing AV implicit communication strategies, (ii) pedestrians’ attention allocation behaviour in front of AVs employing explicit communication strategies, and (iii) the impact of repeated exposures on pedestrians’ adaptation in crossing decisions and attention allocation behaviour in response to implicit and explicit communication strategies.

To address these questions, a series of experiments were conducted in virtual environments using a CAVE-based pedestrian simulator to explore AV-pedestrian interactions across varying contextual factors. Additionally, a distributed simulation setup was developed, connecting the pedestrian simulator to a motion-based driving simulator, enabling real-time interactions between both actors in a controlled and repeatable environment. Results revealed that drivers’ kinematic cues, such as braking and lateral movements, served as effective implicit communication strategies, significantly influencing pedestrians’ crossing decisions and surpassing the impact of infrastructure cues like zebra crossings. Furthermore, pedestrians’ attention allocation patterns in front of AVs were similar to those observed with conventional vehicles. However, the presence of explicit communication methods from AVs, such as external human-machine interfaces (eHMIs) or augmented reality (AR), reduced pedestrians’ head-turning and gaze behaviours, indicating lower attentional demands and effectively conveying AV intent. Repeated exposures to these implicit and explicit communication strategies revealed a learning effect, with pedestrians adapting their crossing decisions and attention allocation behaviours over time.

This thesis concludes by providing a comprehensive understanding of AV-pedestrian interactions through novel experimental approaches and measurements, offering insights for designing effective implicit and explicit communication strategies for AVs.