Drivers may find they make it through some metro Detroit intersections a lot faster, thanks to an AI project from the University of Michigan.
U-M researchers said they're developing technology to help efficiently retime traffic lights to reduce stops and congestion at intersections.
Craig Bryson is the senior communications manager of the Road Commission for Oakland County. He said retiming traffic signals has always been a challenge. “Traditionally, it was labor intensive and expensive, so it didn't get done that often. For us, we were lucky to do it every five to 10 years,” he said.
Bryson said many counties and municipalities virtually never retimed their lights, despite traffic flow changes. Bryson said they used to send a technician to spend hours at an intersection and count the cars — a slow and costly process. “It’s very flawed because human beings, you know, get tired and they miss cars,” Bryson added.
Now, the University of Michigan Transportation Research Institute is using data from cars traveling through specific intersections to figure out the best timing at each light. The GPS data is obtained anonymously from auto companies.
Zachary Jerome is a postdoctoral research fellow at the institute. He said the GPS data allows them to evaluate a broad range of signals and prioritize which ones would benefit from signal retiming.
After identifying those intersections, they use the data to estimate traffic volumes, Jerome said. “What our system is ultimately doing is leveraging data from a small percentage of vehicles to retime traffic signals that give benefit to all vehicle users,” he said.
Bryson said they’ll still have to physically plug in the new timing at intersections, but eventually they’ll be able to remotely reprogram the traffic lights.
The project has been tested at four signals along 8 Mile Road on the Farmington Hills and Livonia border. The results? A 30% reduction in delays and a 40% reduction in traffic stops, researchers said.
Nine signals were updated along 12 Mile Road, from Vinsetta Boulevard to North Connecticut Avenue in Royal Oak, which led to a 20% reduction in both delays and stops, according to the researchers.
“This is huge as far as reducing congestion. It’s also very important for safety because when people are stopping every signal, they get frustrated and they're more likely to take a risk by going through either a red light or a yellow light,” Bryson said.
Jerome said it currently costs around $4,000 to $5,000 to optimize a traffic signal. With the new system, it could be as low as $2,500, close to half the cost.
“The beauty is that this is applicable worldwide in any traffic signal anywhere,” Bryson said. “I think in the long term, it's inevitable. This will ultimately be the way traffic signals are re-timed around the world at some point in the future.”
“As connectivity increases and more and more opportunities are available with technology and AI, this research will be a huge benefit to future implementations on our roadways,” Jerome said.
The project is being paid for by a $1.4 million U.S. Department of Transportation grant.
