Citation

Wotring B, Antin JF, Smith RC. Virginia Tech Transportation Institute. Blacksburg, VA: National Surface Transportation Safety Center for Excellence, Virginia Tech Transportation Institute, 2021.

Copyright

(Copyright 2021, Virginia Tech Transportation Institute)

The full document is available online.

Abstract

This effort sought to determine the prevalence of particular visual and behavioral indicators for alcohol intoxication using data collected in the Strategic Highway Research Program 2 Naturalistic Driving Study (SHRP 2 NDS). A list of visual and behavioral cues was identified from previous research and served as the basis for identification. The prevalence of several of these cues reached statistical significance between judged states of intoxication. Some cues include, but are not limited to, lids-heavy, dozing, exhilarated, distracted, talkative, inability to sit upright, yawning, and leaning against window. While the study was able to determine the prevalence of the markers, several limitations temper interpretation. First, a large proportion of trips evaluated occurred between midnight and 4:00 a.m., when drivers are likely to be drowsy and exhibit many of the same visual and behavioral indicators also expected to be present in intoxicated individuals. Thus, impacts of drowsiness may be confounded with those of intoxication. In addition, the same visual cues were used both to determine the degree of intoxication as well as the behaviors most associated thereto, thus resulting in a logical conundrum. The results of this research should be viewed as exploratory work that can aid in the generation of hypotheses for future work.

PROBLEM STATEMENT
Driving while intoxicated (DWI) by alcohol remains a serious traffic safety concern. The percentage of traffic fatalities with a driver presenting a blood alcohol concentration (BAC) over 0.08 has remained relatively steady over the last 10 years at around 29%. In 2018, this resulted in 10,511 individuals who died in crashes with an alcohol-impaired driver (National Center for Statistics and Analysis, 2019).

One way to better comprehend and prevent intoxicated driving, and the crashes it can lead to, is to identify visual or behavioral signs of alcohol intoxication in the absence of a breathalyzer or other definitive source of information. Visual identification of intoxicated drivers may be beneficial to law enforcement officers both from a distance (i.e., identifying characteristic vehicle movements) and at close proximity (i.e., observing characteristic driver behaviors just prior to or during a traffic stop), as well as for those who interact with the public such as bartenders or wait staff. Another potentially important outcome may be the ability for researchers to identify alcohol intoxication via recorded video in naturalistic driving studies (NDS). The current effort explores a novel area: the application of previous research on visual and behavioral identifiers of intoxication to NDS video views of the driver’s face and key measures of vehicle kinematics. The literature review presented below discusses relevant previous efforts to identify intoxication in different contexts in order to determine which identifiers could be applicable to the current undertaking.