Transportation Department, UT researchers analyze traffic camera data to identify pedestrian safety hazards

Lauren Girgis

The Austin Transportation Department partnered with two UT-based research centers to analyze data from traffic cameras to identify pedestrian safety concerns. 

The partnership with UT’s Advanced Computing Center and UT’s Center for Transportation Research used already existing traffic camera footage to gather information about how and where pedestrians cross streets. The Intelligent Transportation Systems World Congress will give the group a “best paper” award later this month in Singapore for its 2018 report summarizing their findings.

“The question posed in this research is, ‘Can we use the cameras we already have in place to get us good information on things such as the number of pedestrians that are crossing in a particular location, and also, where they are crossing?’” said Jen Duthie, division manager of ATD’s Arterial Management Division.

The city partnered with the computing center to analyze the data. The mass computing used artificial intelligence to recognize objects, such as pedestrians and vehicles, and develop algorithms to track them, said Natalia Ruiz-Juri, a research associate at the Center for Transportation Research.

“This gives us an opportunity to maybe automate some of that analysis and identify locations that are potentially dangerous before accidents happen,” Ruiz-Juri said. 

William Alexander, transportation engineering graduate student, said research done at the Center for Transportation Research is innovative and will help researchers advance in the field.

“Pedestrians are so important to transportation science and engineering,” Alexander said. “By understanding pedestrian behavior better, we can design pedestrian facilities that
are safer.”

As a result of the data, the department installed pedestrian hybrid beacons in specific areas, Duthie said. Pedestrian hybrid beacons are pedestrian-activated traffic signals that allow walkers to safely cross where there isn’t an intersection. 

“We looked at the pedestrian patterns before the beacon was installed and after,” Ruiz-Juri said. “The idea was to see if pedestrian movements were concentrated closer to the beacon after the beacon was installed. The data so far is qualitative, but it suggests that, yes, it works.”

Duthie said the project will not only installed the beacons but will also continue to use the cameras to evaluate other concerning areas where unsafe crossings may be occurring. 

“With over a thousand signals around town, it can be difficult to always know what’s going on at all of them,” Duthie said. “The algorithm can help focus our efforts on areas that have real needs when it comes to pedestrians crossing.”