Lessons learned from pedestrian-driver communication and yielding patterns

Das, Subasish · 2021 · Transportation Research Part F Traffic Psychology and Behaviour

DOI: 10.1016/j.trf.2021.03.011

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 the complex, tacit communication patterns between pedestrians and drivers at non-signalized crosswalks, aiming to improve pedestrian safety and inform autonomous vehicle design. While pedestrian crashes are a significant safety concern, particularly at locations without traffic signals, existing research often isolates individual factors influencing yielding behavior using traditional statistical methods. This approach fails to capture the multi-variable interactions inherent in human decision-making. To address this gap, the authors analyze naturalistic driving data to identify hidden patterns associated with both successful and failed communication events, defined as instances where drivers appropriately yield or fail to yield based on right-of-way rules. The researchers utilized a dataset from the Virginia Tech Transportation Institute’s Safe-D Dataverse, comprising 1,808 pedestrian-driver interactions recorded via data acquisition systems. From an initial pool of 97 variables, the study employed three machine learning models—single decision trees, random forests, and gradient boosting—to determine variable importance. This process selected 16 key features related to driver reactions, pedestrian behaviors (e.g., assertiveness, position, distractions), and environmental conditions (e.g., speed limits, traffic controls). These variables were then analyzed using Taxicab Correspondence Analysis (TCA), a pattern recognition method that identifies co-occurring attribute combinations more effectively than traditional logistic regression for high-dimensional data. The analysis revealed distinct patterns for successful versus failed communications. Successful scenarios were strongly associated with specific combinations of attributes, including pedestrian-driver eye contact, clear facial expressions, assertive pedestrian behavior, and the presence of effective traffic control devices. In contrast, failed scenarios varied significantly by roadway speed limit. On higher-speed roads (35 mph), failures were linked to passive or undecisive pedestrians located far from the crosswalk, where drivers tended to accelerate rather than wait. On lower-speed roads (15 mph), failures were associated with distracted pedestrians, vehicles having the right of way, and the absence of traffic control devices. These findings provide actionable insights for transportation agencies to develop targeted countermeasures, such as infrastructure improvements and user education, to reduce pedestrian crashes. Furthermore, the identified communication patterns offer critical data for developing computer vision algorithms in autonomous vehicles, enabling them to interpret human intent more accurately and make safer yielding decisions in complex, uncontrolled environments.

Key finding

Successful pedestrian-driver communication is strongly associated with eye contact, facial expressions, and assertive pedestrian behavior, whereas failed communication patterns vary by speed limit and involve passive or distracted pedestrians and speeding drivers.

Methodology

naturalistic

Sample size: 1808

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. Discovered via author_sweep_intake on 2026-05-27.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

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

• driver vru interaction
• pedestrian behavior perception
• rail grade crossings
• signaling behavior