Detecting Driver Distraction

Eskenazi, Maxine; Black, Alan W.; Keller, Timothy A.; Shah, Suruchi; Hu, Ting-Yao · 2018 · ROSA P / Technologies for Safe and Efficient Transportation. University Transportation Center

archive: archived pipeline: cataloged verified

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

Summary

This research report addresses the critical safety issue of distracted driving, which remains a leading cause of crashes and fatalities despite legislative bans on cellphone use while driving. The authors propose that automated systems capable of detecting driver distraction and issuing warnings could prevent serious accidents. The primary objective of this project was to establish a controlled data collection environment to study distracted driving, with the long-term goal of developing algorithms to automatically identify distracted states. The study was motivated by the impracticality of collecting sufficient naturalistic data from thousands of drivers to capture rare, specific combinations of driving conditions and distracting activities. To overcome these data collection challenges, the researchers utilized the OpenDS driving simulator, which allowed for precise control over driving events and distraction triggers. The experimental design involved a specific route featuring hairpin turns, which were selected to induce high cognitive load and create dangerous situations when combined with distractions. The study recruited 50 graduate students from Carnegie Mellon University’s School of Computer Science. Each participant drove the same course for approximately 15 minutes. Distractions were introduced via a "Wizard of Oz" system that triggered phone calls, emails, and text messages at specific points in the itinerary. These interruptions varied in cognitive load, ranging from simple queries to complex memory tasks. The simulator logged vehicle data, backward-facing video, and timestamped phone interactions, which were synchronized to create a comprehensive dataset. The researchers hand-labeled instances of distraction based on third-party observation of audio and video recordings. This labeling process identified moments where drivers exhibited distracted behavior in response to the controlled interruptions. The study also administered the Cognitive Failures questionnaire to participants after their driving session to assess individual cognitive traits. The resulting database, which includes synchronized driving metrics, video, and distraction labels, was made publicly available on GitHub. The significance of this work lies in the creation of a robust, controlled dataset that enables the study of specific interactions between driving conditions and cognitive distractions. By using a simulator, the researchers eliminated the need for extensive annotation of naturalistic data and ensured that specific causal relationships could be isolated. This dataset provides a foundation for developing and testing automated distraction detection systems, contributing to the broader field of safe and efficient transportation technologies. The study demonstrates a feasible approach to gathering high-quality data for modeling driver distraction, addressing a key gap in the development of active safety systems.

Key finding

Fifty drivers were recorded for an average of 14 minutes each on a simulated route with controlled phone-call and text interruptions, producing a publicly released distraction-detection dataset.

Methodology

simulator

Sample size: 50

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 bulk_ingest_rosap on 2026-05-23 (7 acquisition events logged).

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.

• distraction detection algorithms
• visual
• cognitive
• external distraction
• auditory
• mind wandering

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).

• Methodological Resource: tool software, measurement protocol
• Theoretical Contribution: conceptual framework