Using vehicle-based sensors of driver behavior to detect alcohol impairment.

Lee, John; Brown, Timothy; Fiorentino, Dary; Fell, James C.; Traube, Eric; Nadler, Eric · 2011 · ROSA P / United States. Department of Transportation. National Highway Traffic Safety Administration

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 study addresses the persistent problem of alcohol-impaired driving, which contributes to approximately 30% of traffic fatalities despite existing enforcement and educational efforts. The research investigates the feasibility of using vehicle-based sensors to detect alcohol impairment in real time by identifying behavioral signatures associated with changes in driver control inputs, vehicle state, driving context, and driver state. The goal was to develop algorithms capable of distinguishing drivers with blood alcohol content (BAC) levels above the legal limit of 0.08% from those below it, potentially enabling vehicle-based countermeasures to prevent crashes. Data were collected using the National Advanced Driving Simulator (NADS), a high-fidelity simulator, involving 108 volunteer drivers across three age groups (21–34, 38–51, and 55–68 years). Participants drove through representative urban, freeway, and rural scenarios at three BAC levels: 0.00%, 0.05%, and 0.10%. The study design included rigorous screening for eligibility, including health status and drinking history, and controlled alcohol administration to achieve target BAC levels. Sensors captured detailed metrics such as lane position variability, speed variation, steering inputs, and eye movements. The researchers developed and evaluated three types of algorithms—logistic regression, support vector machines (SVM), and decision trees—to classify drivers as impaired or unimpaired based on these sensor data. The results demonstrated that vehicle-based sensors could detect alcohol impairment with an accuracy of approximately 80%, comparable to the Standardized Field Sobriety Test (SFST) used by law enforcement. Specifically, decision trees achieved the highest accuracy at 84.7%, followed by SVMs at 82.3% and logistic regression at 82.0%. Lane position variation was identified as the most sensitive metric to alcohol impairment, showing a linear decrease in performance as BAC increased. The time required for detection varied significantly based on algorithm complexity and driving context; complex algorithms applied to demanding situations detected impairment in as little as eight minutes, while simpler algorithms required up to twenty-five minutes. Detection was notably faster when algorithms utilized variables specific to the driving situation (e.g., lane keeping on rural roads) rather than generic variables. Additionally, individualized algorithms that accounted for baseline driver behavior outperformed generic models. The study concludes that vehicle-based systems using behavioral metrics like lane position and speed variability are viable tools for detecting alcohol impairment. The findings highlight that detection performance is heavily influenced by driving context, suggesting that future algorithm development must account for specific roadway situations and individual driver baselines. These results support the potential for vehicle-based interventions to provide real-time feedback or warnings to drivers, thereby mitigating alcohol-related crashes. The authors also suggest that similar sensor-based approaches could be extended to detect other impairments, such as distraction and drowsiness.

Key finding

Vehicle-based sensor algorithms detected alcohol impairment with approximately 80% accuracy, matching the performance of the Standardized Field Sobriety Test.

Methodology

simulator

Sample size: 108

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.

• alcohol detection systems
• alcohol
• dui enforcement
• drowsy as impairment
• drowsiness detection algorithms
• polydrug

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: validation psychometrics, tool software
• Theoretical Contribution: computational model