Skip to main content

Safer Streets through Big Data

Auburn Engineering is paving the way for new solutions to roadway safety through predictive algorithms and advanced technologies

By Christine Hall

Automobile collisions resulted in more than 37,000 fatalities in the United States in 2017, according to recent projections from the National Highway Traffic Safety Administration. Is it possible to predict when and where accidents will happen? Can the number of yearly traffic fatalities and injuries be reduced to zero? Civil engineering faculty members at Auburn University are using big data and advanced technologies to answer these questions through their research in transportation engineering. 

ANALYZING DRIVER BEHAVIOR

Unprecedented amounts of data are being generated from sources such as ride-sharing services and smartphone applications. With information becoming more prevalent and available than ever before, the resultant mountains of data can be mined to better forecast travel behavior patterns.

Managing the transportation system involves long-range planning, and Associate Professor Jeffrey LaMondia is confident that these databases will revolutionize the process. He’s working to help shape the Next Generation Travel Behavior Data Initiative for the Federal Highway Administration to better understand how, why, when and where people travel.

“It’s the first of its kind that we’re pushing forward,” LaMondia said. “I’m working closely with a number of states and private industries to collect and fuse data from a variety of public and private sources together.”

The objectives of the initiative are to collect, process, estimate and report national state and local travel behavior on an annual basis. With enough data, patterns begin to develop that can be applied to transportation modeling, planning and policy. It’s the ability to predict trouble before it happens, such as maintenance issues and traffic congestion, that can make our streets safer for all users.

“It’s extremely probability based,” LaMondia said. “We use non-deterministic methods to statistically identify whether you might have a high possibility of an incident happening, kind of like a red flag indicating that something may be wrong.”

LaMondia is also teaching a data analytics course that blends the fields of mathematics, computer science, statistics and civil engineering. This newly offered class gives students the skills and tools necessary to understand patterns, discover relationships and develop predictions based on these data sets.

“If you leave your house, you’re going to be affected by the things that we do at Auburn.”

“I come at it from a transportation angle, but we use different engineering examples in the class,” LaMondia said. “It’s really focused on modeling and simulating using these giant data sets. From structural equation modeling to neural network analyses, there’s so much information yet to be discovered that’s relevant to all fields of engineering.”

This background in data science will help future transportation planners allocate the necessary resources to select safe and efficient infrastructure advancements to meet the changing needs of roadway users. 

PRIORITIZING INTERVENTIONS

Highway work zones present scenarios for traffic crashes that might not occur otherwise. Rod Turochy, the James M. Hunnicutt Professor of Traffic Engineering and director of the Alabama Technology Transfer Center, is investigating the severity of work zone-related crashes and the relationships between their severity and other variables.

“We encounter work zones on a regular basis, and I only see that increasing,” Turochy said. “Most of today’s system is due for major replacements and managing these work zones in the future will be imperative.”

Turochy has developed a database of work zone related crashes in Alabama for the past 10 years that includes details from police reports, traffic control inspector reports, and supporting documentation from contractors.

“We’re currently in the process of mining that database to find information that we can use to better design work zones and traffic control,” Turochy said. “We anticipate that this information will enable us to make some decisions to better operate these areas and reduce the frequency and severity of crashes.” 

PINPOINTING THE FOCUS

Huaguo Zhou, professor of transportation engineering, is leveraging datasets to create mathematical models that can predict the risk of wrong-way drivers entering the freeway at exit ramps. He is working with students to identify and monitor individual high-risk locations in Alabama. New data analysis technology can enable us to reconsider the assumptions about wrong-way collisions, according to Zhou.

“There have been several research statements that wrong-way crashes are totally random, but we have found that high-risk locations exist. Instead of putting a camera at each possible location, we can develop a model to select sites and then verify the model using field data,” Zhou said. “We hope this model can be used nationwide. Although wrong-way driving crashes are rare, they are severe. If each state can identify two locations per year to implement our suggested countermeasures, there will be a significant reduction of fatalities.”

Over the past five years, Zhou has been working with the American Traffic Safety Services Association to develop advanced traffic control devices and intelligent transportation technologies to reduce severe and fatal crashes, including wrong-way driving, roadway departure and work zone intrusion crashes. Five case study booklets have been published as a result of this work, one of which addresses the impact of connected and automated vehicles on transportation infrastructures and traffic control devices — one of the earliest publications of its kind to help agencies prepare for accommodating future transportation technologies.

Zhou is also collaborating with Turochy on a project to improve geometric design practices for rural highway intersections using the Naturalistic Driving Study Database. Acquiring data from the nationally funded project, the researchers have access to information from instrumented cars around the country to address the role of driver performance and behavior in traffic safety. The Department of Civil Engineering recently purchased a mini-driving simulator to enhance classroom teaching and assess the risk of poor driver behaviors.

“We’re trying to improve the safety and efficiency of the system, mainly in areas of highway safety analyses and traffic operations,” Turochy said. “There’s all kinds of data being collected, from vehicle movement to looking at what the driver is doing and capturing potential distraction issues. Access to this data gives us the opportunity to really see what’s going on from a driver behavior perspective and determine what drivers are responding to.”

With this information, Turochy and Zhou can uncover and evaluate elements that may contribute to crashes at intersections and determine to what extent driver behavior, road conditions and street signs play a role. The information will support the development of alterations to roadway geometry, signs and pavement markings that can prevent traffic collisions and injuries.

“One of the reasons I was drawn to transportation engineering is because it impacts almost everyone every day,” Turochy said. “If you leave your house, you’re going to be affected by the things that we do at Auburn.”