Ann Arbor Times

A new traffic signal timing system developed by engineers at the University of Michigan is being tested in Oakland County, aiming to reduce the number of stops drivers make at intersections. The system uses GPS data from as little as 5% of vehicles on the road to recalibrate signals every few weeks, which helps prevent unnecessary stops.

The Road Commission for Oakland County has partnered with the university to implement this technology. So far, 13 intersections have been updated with the new system. Over the next six months, up to 40 intersections are expected to be equipped with this technology, which is being prepared for commercialization by a startup called Connected Traffic Intelligence. This rollout is supported by a $1.4 million grant from the U.S. Department of Transportation.

There are about 320,000 traffic signals across the United States, and annual congestion costs related to these intersections total approximately $22.9 billion. These costs include time lost waiting at lights and increased energy consumption due to inefficient signal timing.

“We utilized the GPS data to scan over 1,400 intersections in Oakland County and identify the 30-40 intersections that would most benefit from our retiming updates,” said Zachary Jerome, a postdoctoral research fellow at U-M’s Transportation Research Institute. “Vehicle telematics data provides us with opportunities that were previously unavailable to evaluate traffic signal timing performance across entire traffic networks. It enables us to proactively pinpoint inefficiencies rather than having to install roadside detection systems at every intersection.”

The university’s technology typically uses data from about 6%-10% of all vehicles nationwide as a substitute for traditional traffic detection equipment at intersections. For example, if a tracked vehicle stops about 100 feet before an intersection, it likely indicates several cars ahead in line. In an earlier trial over an 18-month period starting in 2023 at 34 Birmingham (Michigan) intersections, this approach led to a reduction in stops by between 20% and 30%. Road commission officials believe these improvements could also enhance safety.

“This will help us identify intersections where there is congestion and delay,” said Danielle Deneau, director of the traffic safety department at the Road Commission for Oakland County. “We can correlate or compare that to intersections where there are high frequencies of crashes. With that in hand, we can make improvements on the timing of those traffic lights that will hopefully reduce the number of crashes.”

Most current traffic signals operate based on preset patterns scheduled throughout different times of day—a method intended to coordinate flows but often not responsive enough for changing conditions. Updating these patterns typically requires significant time and expense; recalculating timings can take two to six months per intersection and cost up to $4,500 each time—so many municipalities update them infrequently.

Adaptive signal systems have existed since the 1970s but are costly: installing sensors and programming real-time adjustments can cost up to $50,000 per intersection just for installation alone. While Oakland County began using adaptive signals decades ago—now covering about half its total signals—the new U-M system offers potential improvements at lower cost: around $2,500 per intersection annually.

“As our system does not rely on infrastructure-based sensors, it is very scalable and can be implemented for the entire Southeast Michigan so that more drivers can save travel times with lower signal retiming cost,” said Henry Liu, director of UMTRI and professor of civil and environmental engineering.

Funding comes through USDOT’s Strengthening Mobility and Revolutionizing Transportation program supporting advanced technologies aimed at improving transportation efficiency and safety.

Connected Traffic Intelligence has licensed this technology; both Liu and Jerome—as well as University of Michigan—have financial interests in this company.