Using Wrong-Way Driving Detection Data to Identify Factors That Affect Wrong-Way Driving Behavior and Compare Detection Technologies

Abstract

Previous research has studied wrong-way driving (WWD) crashes, citations, and 911 calls to understand WWD characteristics, but no research has utilized WWD detection data. In this paper, WWD detection data from 48 toll road exit ramps were modeled to identify the factors that affect WWD frequency. The developed negative binomial model showed that exit ramps with more lanes, a toll booth, and higher crossing street volumes were predicted to have more WWD, whereas having more lanes on the opposing approach to the exit ramp was predicted to reduce WWD. Examination of 726 WWD detections showed that 23% were initiated by left-turning vehicles and 18% were initiated by right-turning vehicles, with significantly more left-turn entries occurring at night than right-turn entries (at α = 0.05). A case study of five ramps showed that exits with an extended median on the crossing road had fewer left-turn entries and exits with right-turn lanes on the crossing road had fewer right-turn entries. Lastly, radar, laser, and thermal WWD detection technologies were compared based on their WWD detections and false alarms. The laser sites had significantly more detections and false alarms than the radar sites (at α = 0.05), with the radar sites having a significantly higher WWD detection rate than false alarm rate. Radar false alarms were mainly caused by large trucks, whereas the laser sites had issues with equipment damage and traffic queues. The results of this paper can help agencies better understand WWD behavior and identify the most appropriate WWD detection technologies.