Deceleration parameters as implicit communication signals for pedestrians’ crossing decisions and estimations of automated vehicle behaviour

Tian, Kai; Tzigieras, Athanasios; Wei, Chongfeng; Lee, Yee Mun; Holmes, Christopher; Leonetti, Matteo; Merat, Natasha; Romano, Richard; Markkula, Gustav · 2023 · OpenAlex-citations

DOI: 10.1016/j.aap.2023.107173

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 investigates how pedestrians interpret implicit communication signals from automated vehicles (AVs), specifically focusing on vehicle deceleration parameters. The absence of a human driver in AVs eliminates conventional explicit communication methods, such as eye contact, creating potential safety risks and interaction dilemmas. While previous research established that implicit signals influence pedestrian behavior, it remained unclear whether pedestrians actively estimate vehicle intentions and how those subjective estimations correlate with their actual crossing decisions. The authors aimed to fill this gap by examining the impact of implicit signals on both pedestrians’ subjective estimations of vehicle behavior and their objective crossing decisions across varied traffic scenarios. The researchers conducted a simulated experiment using the Highly Immersive Kinematic Experimental Research (HIKER) lab at the University of Leeds. Thirty healthy adults participated in two tasks: a natural road crossing task and a vehicle behavior estimation task. The experimental design manipulated three vehicle behaviors (early-onset deceleration, late-onset mixed braking, and constant speed), three initial speeds (25, 40, and 55 km/h), and two initial time-to-collision values (3 and 6 seconds). In the first task, participants decided whether to cross the road as an AV approached. In the second task, participants viewed video segments of the scenarios and estimated whether the vehicle was yielding or passing, providing confidence ratings. The study utilized visual cues, specifically the change rate of tau ($\dot{\tau}$), to quantify deceleration evidence and segment the scenarios for analysis. The results revealed that the correlation between crossing decisions and vehicle behavior estimations depended heavily on the traffic scenario. Pedestrians’ recognition of deceleration behavior generally aligned with their crossing decisions, supporting the hypothesis that they actively estimate vehicle behavior during decision-making. However, when the traffic gap was sufficiently large, the effects of vehicle speed on crossing decisions and estimations were opposite, suggesting that subjective estimation may not directly drive decisions when the time gap is large. Pedestrians crossed earlier and estimated yielding behavior more accurately in early-onset braking scenarios compared to late-onset braking scenarios. Notably, vehicle speed significantly influenced estimations; pedestrians tended to perceive low vehicle speeds as yielding behavior regardless of the vehicle’s actual intent. The study also demonstrated that the visual cue $\dot{\tau}$ serves as a practical indicator for controlling deceleration evidence. These findings highlight the critical role of deceleration parameters as implicit communication signals in pedestrian-AV interactions. The study concludes that pedestrians rely on visual cues to estimate vehicle intentions, but this process is complex and context-dependent. The discrepancy between estimation and decision-making in large-gap scenarios implies that AV control algorithms must account for how pedestrians interpret speed and deceleration to ensure safe and predictable interactions. The results provide specific implications for the development of AV motion planning and external human-machine interfaces, emphasizing the need for clear, early deceleration signals to facilitate accurate pedestrian estimation and safe road crossing.

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.

• pedestrian behavior perception
• driver vru interaction
• ehmi external hmi
• vru facing ehmi
• braking response
• rail grade crossings