Driver Impairment Detection & Safety Enhancement Through Comprehensive Volatility Analysis [Slides]

Khattak, Asad J.; Arvin, Ramin; Chakraborty, Subhadeep; Melton, Chad; Clamann, Michael · 2021 · ROSA P / Collaborative Sciences Center for Road Safety

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 addresses the critical safety issue of driver impairment and distraction, which contribute to approximately 35% of transportation-related deaths. The study aims to quantify the association between these factors, driving performance instability, and crash severity using a systems-based approach. The primary motivation is to advance transportation safety by understanding how distraction and impairment limit driver attention, increase reaction times, and elevate workload, ultimately leading to safety-critical events. The methodology utilizes high-resolution data from the Naturalistic Driving Study (SHRP 2), encompassing baseline driving, near-crashes, and crashes. The researchers developed a systems framework integrating driver biometrics (gaze), vehicle kinematics, and environmental data. They employed path analysis to explore the direct and indirect relationships between distraction/impairment, driving instability, and crash intensity. Additionally, the study applied Artificial Intelligence techniques, specifically a 1-Dimensional Convolutional Neural Network-Long Short Term Memory (1D-CNN-LSTM) model, to predict safety-critical events in real-time using 15-second windows of instantaneous vehicle kinematics and distraction profiles. The project also involved experimental data collection in simulated and naturalistic settings, recording physiological responses such as Galvanic Skin Response, Electrocardiogram, and Electromyographic signals alongside vehicle dynamics. The findings reveal that distracted and aggressive driving significantly increase driving instability, which serves as a strong proxy for crash intensity. Distraction was found to increase crash severity both directly and indirectly through this increased instability. The analysis of distraction duration showed that longer durations, particularly involving mobile phones, substantially raise the probability of involvement in safety-critical events, with risk varying by distraction type. Alcohol and drug impairment also significantly increased crash risk. The AI model demonstrated robust performance, correctly predicting 73.4% of safety-critical events with a precision of 95.7% and maintaining a very low false-alarm rate of 0.57% during non-event driving. Model convergence indicated no significant overfitting issues. The significance of this work lies in its comprehensive systems framework that links microscopic driving behaviors to macroscopic safety outcomes. By establishing driving volatility and distraction as leading indicators for crash prediction, the research supports the development of real-time hazard prediction systems. The successful application of deep learning to naturalistic streaming data suggests that monitoring driver biometrics and vehicle kinematics can effectively enhance driver safety and reduce crash severity, providing a foundation for future intelligent transportation systems.

Key finding

Distracted and aggressive driving increase driving instability, which directly and indirectly increases crash intensity, while an AI model using vehicle kinematics and distraction profiles successfully predicts safety-critical events with 73.4% accuracy.

Methodology

naturalistic

Sample size: 9239

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 (6 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.

• telematics crash prediction
• naturalistic crash near crash
• distraction detection algorithms
• sex gender
• visual
• cannabis

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).

• Empirical Findings: crash risk outcomes
• Methodological Resource: dataset resource
• Theoretical Contribution: computational model