The Utah Department of Transportation will launch a pilot program next month that uses artificial intelligence to detect problems on Salt Lake County roads.
The Salt Lake City-based software firm Blyncsy announced the partnership Thursday, marking the first official use of its Payver A.I. technology. The program collects crowdsourced, real-time dash camera footage to detect fading painted lines on roads, but CEO Mark Pittman said he plans to expand the A.I. to also detect hazards such as missing traffic signs, potholes and damaged bridges throughout the state and the nation.
UDOT crews are “kind of driving the roads, looking for these issues all the time,” Pittman said in an interview. “We’re trying to get ahead of that and ... reduce maintenance costs, spending less time looking for issues and more time fixing the issues.”
The pilot program will run six months and cost the state $105,000, initially monitoring 350 road miles in Salt Lake County. UDOT will not provide its own video footage to Blyncsy. Instead, the company buys footage from partners that collect video from in-vehicle dash cameras.
“Satellite truck drivers, Uber drivers, regular folks have a dash cam in their car. Your insurance company, they subsidize you to have one,” Pittman said. “We buy those images, and then we run a machine vision algorithm that makes sense of it.”
The initial focus on painted lane lines is important, Pittman added, to support the rise of self-driving cars. Autonomous vehicles need clear road markings to “see” lanes and parking stalls. But the lines need to be repainted frequently, particularly in snowy areas where plows scrape them away.
“We’re trying to get really fine-grained over time, but we’re starting with what everyone cares about,” Pittman said. “Everyone has striping issues. Everyone’s repainting their roads. They’re critical ... for safety.”
Blyncsy will also be partnering with transportation departments in California and New Mexico to pilot its Payver A.I. in the coming weeks. UDOT has also teamed up with Blyncsy in recent years to collect data on travel times and intersection signals, according to Robert Miles, UDOT’s director of traffic and safety.
“They’ve demonstrated to us some very interesting technology,” Miles said, “but we need to be able to utilize it on our system to see if it’s really going to work here.”
John Gleason, a spokesperson for the department, said achieving real time, situational awareness on Utah’s roads is a long-term goal.
“Right now, we have our maintenance workers and the public,” Gleason said. “If they see a pothole or see striping that needs to be addressed, they’ll call it in.”
Miles said that real-time data not only gives the department a quicker and more comprehensive understanding of road condition, but it also could be more reliable than depending on humans.
“If I drive down a roadway and I’ve had a bad afternoon and I can’t see the striping, I’m going to react more negatively to that than I would the next day when I had a good afternoon,” Miles said. “So getting that repeatable, consistent measure is invaluable to us.”