Augmented reality for supporting the interaction between pedestrians and automated vehicles: an experimental outdoor study

Aleva, Thomas K.; Tabone, Wilbert; Dodou, Dimitra; de Winter, Joost C. F. · 2024 · OpenAlex-citations

DOI: 10.3389/frobt.2024.1324060

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenge of improving pedestrian safety in interactions with automated vehicles (AVs) by evaluating augmented reality (AR) interfaces. While external human-machine interfaces (eHMIs) on vehicles have shown promise, they suffer from limitations such as occlusion and an inability to address individual pedestrians. Previous AR research relied heavily on virtual reality or questionnaires, lacking real-world validation. This research investigates whether AR headsets can effectively support pedestrian crossing decisions in an outdoor environment and compares the efficacy of different AR anchoring methods: vehicle-locked, world-locked, and head-locked. The experiment involved 28 participants who wore Varjo XR-3 see-through AR headsets in a closed outdoor area. Participants observed a simulated AV approaching from the right and were tasked with holding a button when they felt it was safe to cross. The study compared four AR interfaces against a no-AR baseline: "Planes on vehicle" (vehicle-locked), "Virtual fence" and "Fixed pedestrian lights" (world-locked), and "Pedestrian lights HUD" (head-locked). The AV exhibited either yielding or non-yielding behavior. Data collection included button press timing, post-trial intuitiveness ratings, post-experiment questionnaires assessing acceptance and preference, and semi-structured interviews. Statistical analyses utilized repeated-measures ANOVAs and paired-samples t-tests. Results indicated that participants subjectively preferred all AR interfaces over the no-AR baseline. The "Pedestrian lights HUD" was statistically more effective than the baseline, leading to higher button press percentages during yielding scenarios, indicating greater confidence in crossing. In contrast, the "Fixed pedestrian lights" performed poorly, scoring lowest on composite acceptance scores and preference rankings. This underperformance was attributed to low saliency; participants perceived the interface as too small and difficult to locate, which delayed their recognition of the signal. The "Virtual fence" and "Planes on vehicle" showed no significant differences from each other but were generally preferred over the fixed lights. Intuitiveness ratings were significantly higher for AR conditions compared to the baseline, particularly for yielding AVs. The study concludes that while AR is favored over no communication, its effectiveness is heavily dependent on design factors such as visibility, anchoring method, and visual attention demands. Head-locked interfaces proved superior in this context because they remained within the user's field of view, reducing the cognitive load of locating the signal. The findings highlight that clear, easily interpretable, and salient AR interfaces are critical for successful AV-pedestrian interaction. This work provides a methodological framework for conducting outdoor AR experiments and offers specific design insights for future wearable safety technologies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• ehmi external hmi
• vru facing ehmi
• hud ar windshield