Crowdsourced Assessment of 227 Text-Based eHMIs for a Crossing Scenario

Bazilinskyy, Pavlo; Sodou, Dimitra; De Winter, Joost · 2022 · Crossref

DOI: 10.54941/ahfe1002444

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 lack of consensus regarding the optimal phrasing of text-based external human-machine interfaces (eHMIs) for automated vehicles (AVs). While text-based eHMIs offer precise information compared to abstract signals like LED strips, they require focused attention and time to read, raising concerns about their efficacy and clarity. The research specifically investigates which text messages are most compelling for pedestrians in a crossing scenario, how message perspective (egocentric vs. allocentric) affects interpretation, and how language barriers influence comprehension. To evaluate these factors, the authors conducted a crowdsourcing experiment involving 1,438 participants, with 908 retained after data cleaning. The study assessed 227 distinct text-based eHMIs, derived from literature and systematic combinations of verbs and modifiers, presented on images of an AV’s front bumper. Participants indicated their willingness to cross the road on a 0–100 scale and had their response times recorded. The eHMIs were categorized as egocentric (addressing the pedestrian, e.g., “DON’T WALK”), allocentric (describing the vehicle’s state, e.g., “BRAKING”), or both. The study also compared English and Spanish translations of 47 messages to assess cross-national interpretation. The results demonstrated that egocentric messages were significantly more compelling than allocentric ones, driving participants toward clear decisions to either cross or not cross, whereas allocentric messages often left participants in doubt. Egocentric messages also yielded faster response times. Conversely, longer messages were associated with slower responses. Specific allocentric messages, particularly those with negative wording (e.g., “WON’T GO”) or ambiguous intent (e.g., “I SEE YOU”), were found to be confusing, characterized by low compellingness, high inter-subject variability, and slow response times. Additionally, Spanish-speaking participants found eHMIs in Spanish more compelling than those in English, while English-speaking participants struggled to comprehend Spanish messages. The findings imply that eHMI design should prioritize egocentric instructions to ensure clear communication and rapid pedestrian response. The study highlights the risks of ambiguous allocentric messages and underscores the challenge of language barriers in global AV deployment. While text offers precision, the authors note that the time required to read text and the difficulty of standardizing messages across languages may limit the viability of text-based eHMIs, suggesting that simple, universally understood commands or non-textual signals might be more practical for widespread implementation.

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.

• vru facing ehmi
• ehmi external hmi

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Applied Guidance: design guidelines