Search: onr:"swepub:oai:research.chalmers.se:44595b81-51dc-451b-8e59-fd44b3b5307c" >
Analysis of SHRP2 D...
Analysis of SHRP2 Data to Understand Normal and Abnormal Driving Behavior in Work Zones
-
- Flannagan, Carol A. (author)
- University of Michigan
-
- Selpi, Selpi, 1977 (author)
- Chalmers tekniska högskola,Chalmers University of Technology
-
- Boyraz Baykas, Pinar, 1981 (author)
- Chalmers tekniska högskola,Chalmers University of Technology
-
show more...
-
- Leslie, Andrew (author)
- University of Michigan
-
- Kovaceva, Jordanka, 1980 (author)
- Chalmers tekniska högskola,Chalmers University of Technology
-
- Thomson, Robert, 1964 (author)
- Chalmers tekniska högskola,Chalmers University of Technology
-
show less...
-
(creator_code:org_t)
- 2019
- English.
- Related links:
-
https://research.cha... (primary) (free)
-
show more...
-
https://research.cha...
-
show less...
Abstract
Subject headings
Close
- This research project used the Second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study(NDS) to improve highway safety by using statistical descriptions of normal driving behavior to identify abnormal driving behaviors in work zones. SHRP2 data used in these analyses included 50 safety-critical events (SCEs) from work zones and 444 baseline events selected on a matched case-control design. Principal components analysis (PCA) was used to summarize kinematic data into “normal” and “abnormal”driving. Each second of driving is described by one point in three-dimensional principal component (PC) space;an ellipse containing the bulk of baseline points is considered “normal” driving. Driving segments without-of-ellipse points have a higher probability of being an SCE. Matched case-control analysis indicates that thespecific individual and traffic flow made approximately equal contributions to predicting out-of-ellipse driving. Structural Topics Modeling (STM) was used to analyze complex categorical data obtained from annotated videos.The STM method finds “words” representing categorical data variables that occur together in many events and describes these associations as “topics.” STM then associates topics with either baselines or SCEs. The STM produced 10 topics: 3 associated with SCEs, 5 associated with baselines, and 2 that were neutral. Distractionoccurs in both baselines and SCEs. Both approaches identify the role of individual drivers in producing situations where SCEs might arise. A countermeasure could use the PC calculation to indicate impending issues or specific drivers who may havehigher crash risk, but not to employ significant interventions such as automatically braking a vehicle without-of-ellipse driving patterns. STM results suggest communication to drivers or placing compliant vehicles in thetraffic stream would be effective. Finally, driver distraction in work zones should be discouraged.
Subject headings
- NATURVETENSKAP -- Matematik -- Sannolikhetsteori och statistik (hsv//swe)
- NATURAL SCIENCES -- Mathematics -- Probability Theory and Statistics (hsv//eng)
- NATURVETENSKAP -- Data- och informationsvetenskap -- Datavetenskap (hsv//swe)
- NATURAL SCIENCES -- Computer and Information Sciences -- Computer Sciences (hsv//eng)
Keyword
- Work zones
- crashes
- data analysis
- safety
- topic model
- statistics
- methodology
Publication and Content Type
- rap (subject category)
- vet (subject category)
To the university's database