UT supercomputer tasked with finding Austin traffic solutions

A supercomputer at UT's Texas Advanced Computing Center can identify people, vehicles and traffic lights in raw traffic camera video. (Courtesty TACC)
A supercomputer at UT's Texas Advanced Computing Center can identify people, vehicles and traffic lights in raw traffic camera video. (Courtesty TACC)

AUSTIN (KXAN) — As Austin traffic continues to grow, the city has a unique tool at its disposal to find solutions to traffic woes: UT Austin’s Stampede 2 supercomputer.

For the past year, the city has been collaborating with researchers at the Texas Advanced Computing Center. They’re using raw video and sensors from the preexisting traffic cameras at major intersections all around Austin, including spots like 38th and Lamar, Third and Lavaca and William Cannon and Westgate. The computers can immediately identify people, cars, buses and traffic lights. Then the computers gather data about their interactions.

“As I think a lot of people know, traffic in Austin is a problem” said Niall Gaffney, director for Data Computing at TACC. “Everybody’s had a near miss or a problem at an intersection, so I think that’s where people will very quickly be able to see the benefit of this.”

The team Gaffney is on built a database by going through the video. Now they’re looking to glean information from it the city can use. So far they’ve been counting how many vehicles move down a road and how many close calls occurred between cars and people walking. The city is also looking to find out who is breaking traffic laws and to identify problem spots where people may consistently be driving the wrong way down one way streets, for example.

Gaffney explained that previous traffic research wasn’t nearly as efficient or inexpensive.

“Before this it was a lot of — I don’t want to call it guess work — but it was a lot of speculative work and a lot of research that went into designing things and testing them in the field,” he said. “We can now go through and understand what’s going on in problem places and come up with solutions that might not have been possible in the past.”

A spokesperson for Austin’s Transportation Department explained that the city is already trying to increase mobility around Austin through making improvements to traffic signal timing. The city hopes the data they gather through TACC’s artificial intelligence research can help them make changes immediately. Additionally, the city believes this research can help them in their aim of ending traffic fatalities.

“We are optimistic this technology will serve as a tool for long-term planning and design for our future projects,” a ATD spokesperson said.

The Stampede 2 is the top open science computing system in the world and the 12th fastest computer in the world, Gaffney said. He explained that to run these computations on your personal computer could take weeks. But with the processing power of the supercomputer, they can track all those relationships from traffic data instantly. They don’t even have to download the video to process the information.

“We’re not going to solve downtown traffic next week — I wish we could — but we hope in the next year or two to give people insight so we can make plans for the future,” he said.

But Gaffney added that their research can only help to identify problems, drivers won’t start seeing changes until public officials decide to spend based on their findings.

“Simply having the data doesn’t bring in the money to make the changes, but it will let us spend that money more efficiently as we understand what’s actually wrong better than we did in the past,” he said.

Gaffney also noted that a majority of the funding TACC receives is actually federal dollars. Stampede 2 was funded with a $30 million award from the National Science Foundation. Gaffney believes it will take more federal funding to keep research like the traffic project going for years to come.

In a release from UT Austin, Natalia Ruiz Juri noted that self driving cars will ultimately lead to even more changes for these already crowded roadways.

“Video data will play a key role in understanding such changes, and artificial intelligence may be central to enabling comprehensive large-scale studies that truly capture the impact of the new technologies,” Ruiz Juri said.

KXAN.com provides commenting to allow for constructive discussion on the stories we cover. In order to comment here, you acknowledge you have read and agreed to our Terms of Service. Users who violate these terms, including use of vulgar language or racial slurs, will be banned. If you see an inappropriate comment, please flag it for our moderators to review.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s