Spring 2019 Vol. 6, No. 1
Transporting crash modification factors to a future with automated vehicles
Gary Davis
As automated vehicles improve in capability and increase in market share, they will change many aspects of highway safety—including the methods used for making decisions about safety.
For the past two decades, data-based tools in AASHTO’s Highway Safety Manual—known as crash modification factors (CMFs)—have provided highway safety practitioners with an empirical way to predict the safety consequences of highway engineering decisions.
“Simply put, a crash modification factor is a way to measure the effectiveness of a particular treatment or design element,” says Gary Davis, a professor in University of Minnesota’s Department of Civil, Environmental, and Geo- Engineering. “While the application of an appropriate CMF can influence the decision to implement a particular project, the misapplication of CMFs can lead to misinformed decisions—and automated vehicles have the potential to reduce the accuracy of CMF predictions.”
The CMFs in the Highway Safety Manual represent the current prevailing driver and vehicle conditions in the United States. To assess the transferability of existing CMFs to new situations such as vehicle automation, an explanation of how the modification achieves its effect will be needed. However, there is currently little guidance on how such explanations might be posed and tested. In a Roadway Safety Institute project, Davis sought to shed more light on how CMFs work by uncovering how one particular CMF—pedestrian hybrid beacons—could modify the likelihood of pedestrian crashes.
Photo: City of Burnsville
A pedestrian hybrid beacon, designed for use at midblock crossings and uncontrolled intersections, is activated when a pedestrian wanting to cross pushes the call button. The signal then initiates a yellow-to-red lighting sequence consisting of steady and flashing lights that directs motorists to slow down and come to a stop. To explore this issue, Davis and his research team first created a micro-simulation to develop an explanation of how pedestrian hybrid beacons could modify crash likelihood. Since existing studies indicated that they can affect both pedestrian and driver behavior, researchers included both possibilities in their model. Based on the results from the simulation, their working hypothesis is that pedestrian hybrid beacons achieve their crash reduction effect in large part by modifying pedestrian behavior. If valid, this explanation has implications for the transferability of the associated CMFs, because sites with a large number of inattentive pedestrians would gain the greatest safety benefits from pedestrian hybrid beacons.
Perhaps most important, this research illustrates the process for assessing how other CMFs might work. “To explain a crash modification factor you need a framework for stating hypotheses and then deriving testable predictions from those hypotheses,” Davis says. “A good starting point is roughly consistent estimates for the target CMF from higher-quality studies. Then, you need to gain insight into the relevant crash events by reviewing crashes that have been investigated and reconstructed in detail.” Once these elements are in place, Davis explains, this research demonstrates how to construct a model that can make predictions about crash events, which can be compared to data to either confirm or refute the explanatory hypotheses. “A well-confirmed explanation then guides decisions about applying an estimated crash modification factor to new situations,” he says.