Probe and connected vehicle data for enhancing traffic management and road safety: A study of work zone crashes, dynamic message sign efficiency and smart work zone public information
Date
2024-08
Authors
Okaidjah, Dorcas
Major Professor
Advisor
Day, Christopher
Wood, Jonathan
Sharma, Anuj
Quinn, Christopher
Lutz, Robyn
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
As traffic congestion and road safety remain paramount concerns, utilizing advanced technologies and data-driven approaches has become critical in transportation engineering. The objective of this research is to examine the use of crowdsourced probe data, including segment speed data and connected vehicle data, to enhance road safety and improve traffic management. Five studies were undertaken that explored novel applications of these datasets.
Segment speed data is in widespread use by agencies to monitor congestion, but its effectiveness as a predictor of crashes has not been studied much. The first study examines traffic flow characteristics in work and non-work zone environments to understand their implications on road crashes at the segment level. Segment speed data is used to develop performance indicators that are combined with related factors to develop a mixed effect linear regression model of crashes. Significant correlations were found between congestion, certain travel metrics, and crash rates, particularly in rural interstates and inactive work zones.
The second study is the first to use connected vehicle data to investigate the effectiveness of dynamic message signs (DMS). Three separate analyses were undertaken. An initial analysis incorporated measurements of changes in speeds from 26 DMS locations and showed a statistically significant impact of DMS messaging on vehicle speeds. This was followed up by a broader analysis incorporating data from 48 locations and effects on speed under different types of message content. The results showed that most message categories had a statistically significant impact on speeds, generally reducing them, with varying amounts of impact. An additional analysis of DMS sign impact examined the impact of message content by considering the message sentiment. Large language models were utilized to classify the message sentiment as positive, negative, or neutral. Messages with positive sentiment and neutral undertones reduced speeds slightly, as determined by the mixed effect model.
The final study explores the reliability of segment speed data and connected vehicle data sources for driving queue warnings in smart work zones. The study compares these against ground truth sensor data to evaluate the suitability of crowdsourced data for smart work zone applications. Study findings demonstrate that connected vehicle data has poor data coverage during overnight periods but is able to yield performance measures with low numbers of missed calls, exhibits low latency in detecting congestion onset, and may hold promise for providing queue warning applications if the number of false calls can be addressed by improving market penetration or improved methodology. Segment speed data had better coverage but longer latency and had many missed calls.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
article