Return to Projects List

[NTC2015-MU-R-07] Identification of Potentially Hazardous Roadway Network Locations Using Microscopic Observational Vehicle Data and Macroscopically Modeled Reaction Time


The state of the practice in safety analysis for identifying hazardous locations and determining appropriate countermeasures is documented in the Highway Safety Manual and is founded on statistical analysis of crash occurrences.  With the proliferation of technologies for gathering detailed, trajectory-level data from vehicles in the transportation system, investigation into identifying potentially hazardous locations through non-crash driver behavior and non-crash events is well-motivated.  Furthermore, the macroscopic flow model equivalent to the so-called Gazis-Herman-Rothery (GHR) car following model provides a yet untapped means to estimate the reaction time necessary for asymptotic stability using the data from fixed-point sensors that measure flow, speed, and occupancy.  As with the microscopic observational data, this empirically derived reaction time should also be investivated as a means to identify hazardous location from archived traffic condition data rather than relying solely on observed crashes.

This collaborative research effort will determine if either of both of these approaches can be exploited as valid and robust indicators of hazardous locations without the need to wait for crashes to occur. The potential safety benefits are significant.  Even though the availability of microscopic observational data is likely to become widespread as we move into the world of connected vehicles, the research will aim to correlate the macroscopically derived reaction times with microscopically observed driver behaviors, such as hard decelerations and/or extreme avoidance maneuvers.  If a strong correlation can be demonstrated, pro-active identification of hazardous sites using archived traffic condition data could emerge as an important alternative or complement to traditional statistical crash analysis.