Sitting in traffic, especially in a busy city like Seattle, can feel like a twisted version of the kids’ game “red light, green light.” You crawl forward in your car a bit and then the signal changes and you wait, frustrated, through another cycle.
While many traffic signals still run on timers that might cause those delays, the future of traffic control at intersections could get a lot smarter and more efficient once those signals start talking to each other and to the cars, says Ali Hajbabaie, assistant professor of civil and environmental engineering at Washington State University’s Voiland College of Engineering and Architecture.
Hajbabaie and his graduate students examine the effects of traffic signals that work as networks, talking with connected vehicles or sensors to better adjust timing. To avoid overload, each intersection is its own network, which then communicates with other traffic signal networks to figure out how to keep vehicles moving in a more effective way and reduce congestion.
“Think about Seattle. There are many intersections and many constraints” to the mathematical problem, says Hajbabaie. “Each intersection has too many decisions to make, so we distribute the problem. Instead of looking at the whole city, which represents a very complex problem, each intersection is a problem that we can solve.”
Based on their models, that system could reduce delays by up to 43 percent. It also relies heavily on connected vehicles that communicate with the signals and each other. In the real world, the results likely wouldn’t be that dramatic, but traffic congestion would significantly improve with around 35 percent of vehicles connected.
Congestion is a pressing and very visible problem, particularly in urban areas around the world. Kirkland-based connected car services company INRIX calculated that in 2018 commuters across the United States spent an average 97 hours last year stuck in traffic, at a cost of $87 billion lost to traffic congestion. That doesn’t even count the environmental or health consequences.
In Seattle, the sixth most congested U.S. city according to the study, motorists spent an average 138 hours in traffic, costing an average $1,932 each. Portland, Oregon, clocked in at number 10 on that list.
Smarter traffic signals wouldn’t solve all traffic congestion, but, Hajbabaie says, more communication between vehicles and signals can make a difference.
“We’ve created that conversation between intersections, so that one might tell another one, ‘I’m going to send this many vehicles to you.’ The other one says, ‘OK, I have this much capacity to receive vehicles,’” he says.
Mehrdad Tajalli, a doctoral student working with Hajbabaie, says the old methods, such as traffic engineers counting the vehicles in an intersection and then adjusting timing on the signal, are not able to cope with the additional traffic or the emergent technologies.
In the connected system, “each intersection can find the optimal timing,” he says. “Signal controllers can also send messages to vehicles to adjust their speed. By combining both signal timing and speed optimization, it’ll reduce travel time by 6–7 percent more.”
Hajbabaie and his students projected what could happen when 100 percent of vehicles are connected and automated. On the computer model, vehicles zip around each other in what appears to be a chaotic scramble. However, the vehicles clearly are telling each other how to slow down, speed up, and maneuver in a way that keeps everyone moving quickly with virtually no wait times or near-crash conditions.
While a connected traffic future may be a decade or more away, Hajbabaie and his students are already examining the effects of a connected and automated vehicle environment through a grant from the Washington State Department of Transportation to prepare for the future.