Visual Detection of Driving While Intoxicated. Project Interim Report: Identification of Visual Cues and Development of Detection Methods
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 interim report, conducted by Anacapa Sciences, Inc. for the National Highway Traffic Safety Administration, addresses the low arrest rate for driving while intoxicated (DWI), estimated at only one in 2,000 offenders. The study was motivated by the need to enhance on-road detection capabilities by identifying visual cues that effectively discriminate between intoxicated drivers and those driving while sober (DWS). The research aimed to determine which visual cues occur frequently enough to be useful and possess the highest discriminability, thereby increasing detection probabilities above chance levels. The methodology involved a multi-step approach. First, researchers reviewed existing literature on alcohol’s effects on driving functions—specifically steering control, velocity control, time-sharing of attention, and information processing. Second, they analyzed 1,288 DWI arrest reports from nine police agencies to identify reported cues. Third, an on-the-road detection study was conducted in Charlotte, North Carolina, and Fort Wayne, Indiana. Trained observers accompanied police officers during patrol, recording 643 instances of driving behavior deviations. In each instance, officers stopped the vehicle and measured the driver’s blood alcohol concentration (BAC) using portable breath testers. This design ensured a broad range of BAC levels (39% below 0.05, 23% between 0.05 and 0.10, and 38% above 0.10) to accurately assess cue discriminability. The analysis of 1,681 cue occurrences led to the development of a DWI Detection Guide containing 23 specific visual cues, which accounted for 93% of observed events. Each cue was assigned a value representing the probability that the driver’s BAC was 0.10 or higher. For example, "Stopping (without cause) in traffic lane" indicated a 70% probability, while "Driving with vehicle defect(s)" indicated a 30% probability. The guide also included rules for adjusting probabilities when multiple cues were observed. Key findings indicated that while the potential number of cues is large, most detection events are explained by a small subset. The study also found that patrol strategy significantly influenced cue frequency, whereas environmental conditions had relatively little impact. The significance of this work lies in the creation of a practical, probabilistic tool for law enforcement to improve DWI detection efficiency. By providing officers with a standardized method to interpret visual cues, the guide aims to reduce false detections and increase the apprehension of intoxicated drivers. The authors concluded that while the guide facilitates the application of research findings, a field test is required to evaluate its effectiveness before widespread implementation. The study establishes a foundational framework for understanding the indirect relationship between alcohol intake, impaired driving functions, and observable visual cues.
Key finding
A set of 23 visual driving cues accounted for 92 percent of observed deviations and enabled the development of a probabilistic detection guide that estimates the likelihood of a driver having a BAC of 0.10 or greater based on observed behaviors.
Methodology
on_road
Sample size: 643
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).
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | rosap | — | — | 2 | 2026-05-23 |
| archive | success | — | — | — | 1 | 2026-05-23 |
| extract | success | cached | — | — | 2 | 2026-06-10 |
| clean | success | — | — | — | 1 | 2026-06-01 |
| chunk | success | — | — | — | 1 | 2026-06-01 |
| embed | success | — | — | — | 1 | 2026-06-02 |
| enrich | success | — | — | — | 1 | 2026-05-23 |
| promote | success | — | — | — | 1 | 2026-05-23 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 3 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 19 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
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.
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: measurement protocol, validation psychometrics