External Human-Machine Interfaces for Interactions of Pedestrians with Highly Automated Vehicles in more Complex Road Traffic Scenarios: Can AR-based Interfaces Help?

Abstract

The advancement of highly automated vehicles (HAVs; SAE 4) in recent years presents both opportunities and challenges to interactions in road traffic. Research has shown that especially pedestrians may struggle with the loss of a human driver as they currently rely on explicit communication, e.g., hand gestures, in ambiguous situations of interactions particularly in low speed and short distance scenarios. External human-machine interfaces (eHMIs) may bridge this communication gap. However, research on their effectiveness especially in more complex traffic scenarios involving more than one HAV in interactions with a pedestrian is lacking so far and some challenges remain unsolved. Here, augmented reality (AR) may pose a promising remedy for drawbacks of conventional eHMIs. This experimental study aimed at investigating the potential of light-based and AR-based eHMIs in more complex road traffic interactions of HAVs with a pedestrian. 45 participants in a virtual reality study were instructed to cross a shared space with HAVs approaching from different directions. The HAVs communicated via light-based, AR-based, combined, or no eHMI and yielded to pedestrians or not. The eHMIs, especially those incorporating AR elements, significantly lowered crossing initiation times and mental workload, whilst improving subjective safety and understandability of the HAVs' behaviours.