March 14, 2024
Artificial intelligence (AI) seems to be everywhere these days — including the roads. Drivers in Seattle are now traveling through some traffic lights adjusted by an AI algorithm.
Applying AI to traffic signal timing is the next step in a global trend to analyze mobile data from vehicles and apply it to urban transportation management. The goal is to reduce vehicle idling time, improve safety, and, in the process, cut greenhouse gas emissions (GHG).
Seattle is the first city in North America to experiment with Google Research's Green Light initiative, which uses AI to manage signal timing to improve the flow of traffic.
In a public-private partnership with Google that is cost-free for Seattle, the city is testing those AI tools to determine when to change signal timing at three intersections, according to Mariam Ali, deputy press secretary for the Seattle Department of Transportation. "They make recommendations, and we implement them — but the technology is all theirs," says Ali.
Google uses AI and driving trends from maps on smartphones to model traffic patterns and make recommendations for traffic light changes that can improve traffic flow. These AI-based recommendations work with existing infrastructure and traffic systems, and city engineers can monitor the impact and see results within weeks. "By optimizing not just one intersection but coordinating across several adjacent intersections to create waves of green lights, cities can improve traffic flow and further reduce stop-and-go emissions," Yossi Matias, vice president of engineering and research at Google, wrote in a blog post.
Seattle's transportation department has used AI to adjust signals at the intersections of 15th Avenue and Market Street (above), Greenwood Avenue and 80th Street, and on 53rd Street in the Ballard neighborhood. Google identified these lights for a pilot project based on its traffic analysis. Photo by SDOT/Flickr.
The typical driver in the U.S. lost 51 hours due to congestion in 2022, a 15-hour increase from the year before, according to an annual transportation study. Those delays amounted to about $870 in lost time and about $550 in added fuel costs. In addition to lost time, congestion can affect productivity by causing freight delays or preventing employees from getting to work on time.
"Traffic signals are one of the largest causes of delay along arterial roadways," says Karl Typolt, a transportation systems engineer at Transpo Group in Kirkland, Washington. Transpo Group, a transportation planning and engineering firm, was not involved in the Seattle trial, but Typolt works on the ways that cities can improve traffic flow. "There's only so much green time to allocate to different movements and modes. Agencies have to balance who gets priority at an intersection — vehicles, pedestrians, cyclists, or public transit — while improving safety and reducing carbon emissions."
Transportation accounts for about 29 percent of total U.S. GHG emissions, making it the largest single contributor, according to the Environmental Protection Agency. The agency states that each gallon of gasoline burned produces about 20 pounds of carbon dioxide, which can build up at stoplights and cause pollution hot spots.
Seattle officials started discussions with Google in 2022 after the company completed research to reduce carbon emissions in Tel Aviv and later in Rio de Janeiro. "Since they only use Google driving trends to make their recommendations and don't interact with our systems, we saw it as a very low-risk, high-potential return opportunity," Ali said in a statement.
Using AI to solve traffic problems
Other government agencies looking for help to reduce idling time while also reaching their sustainability goals can turn to AI and other data analytics tools to cut carbon emissions. Google says Green Light is helping to lower emissions and save fuel for up to 30 million car rides a month in the 12 cities on four continents where it operates. And tests conducted in the past two years indicated the potential to reduce stops by up to 30 percent and reduce emissions at intersections by close to 10 percent.
"What AI can do is take a lot of big data and make it small and actionable," says Deepak Ramnath, AI product manager at INRIX, a Kirkland-based data analytics company that works with public transportation agencies. "The question becomes: What is that small and actionable thing you are trying to accomplish, and can AI help with that? Not all cases can be solved by AI."
Google's Green Light project, which works with existing infrastructure, has the potential to reduce stops by up to 30 percent and reduce greenhouse gas emissions at intersections by close to 10 percent.
INRIX is not involved in the Seattle project but has provided transportation data tools to cities including Los Angeles and Boston.
Transportation planners need to determine their strategy for using AI and big data. Transportation management consultants say that AI algorithms can coordinate traffic lights at different intersections to achieve different goals: minimize traffic stops, prioritize different types of traffic (such as emergency vehicles or buses), enhance pedestrian safety, or alleviate congestion which can boost overall transportation efficiency.
Another consideration is cost. Adding signal infrastructure and data management tools can be expensive. "The key to success is making sure there is funding in place, [that] there are projects in place for these larger capital investments, and that they have staff that will use and maintain the tools once they are operational," Typolt says.
Agencies also need to realize that technology changes quickly. AI was once primarily associated with the tech sector, including smartphones, computers, and software development. But its use has exploded, driven by an increase in the availability of data, improvements in computing power, and a growing awareness of the benefits of AI. "What can be solved now with AI is 100 percent different than what [will] be solved five months from now," says Ramnath. "That is both exciting and scary, especially if you're trying to do five-year budget planning."
Consultants suggest that transportation planners test new data tools at a few intersections, similar to what Seattle is doing, to help them get a feel for what is possible. Large, one-off music and sports events are prime testing grounds where traffic planners can experiment with AI and data analytics techniques.
Transit officials in Austin, Texas, used data analysis from two weeks during the city's 2022 SXSW festival to change light timing and smooth traffic flow, saving some drivers up to a minute at some intersections. Photo by Aaron Rogosin.
Officials in Austin did just that in 2022 during the annual citywide music and film festival SXSW, which brings 300,000 visitors and $380 million in economic activity to the city for two weeks in the spring. During the festival's first weekend that year, engineers in Austin's Mobility Management Center (MMC) used INRIX signal analytics to evaluate traffic signal performance in and around downtown, according to Lance Ballard, an MMC manager. After collecting and analyzing data — from traffic lights, as well as anonymous GPS data from vehicles and mobile devices — the mobility team laid out plans to help alleviate logjams during the second weekend of the festival by retiming signals at eight locations around the city.
"The signal changes led to an improvement in delay and a benefit of approximately $10,000 of delay savings for the eight locations where the MMC adjusted timings for SXSW between Weekend 1 and Weekend 2," Ballard said.
INRIX also considered the trial to be successful.
"It was a success story from the perspective that [Austin's traffic engineers] were able to test, make changes, and measure the impact," says Steve Remias, a product manager at INRIX. Overall, in its 2022 annual report, Austin credited using INRIX data sets across the city with realizing $800,000 in annual savings for road users and reducing delays at intersections by 35 percent.
Meanwhile, Seattle's ongoing agreement with Google renews annually, providing ample time to extend AI from the three intersections included now to other areas. Ali says the city also is considering adding a fourth location near Lumen Field, home of the Seattle Seahawks football team, to help alleviate traffic headaches.
The city also proposes to reduce GHG emissions by 50 percent by 2030 to meet its climate change goals, according to the Office of Sustainability & Environment. Since Green Light can be scaled to analyze thousands of intersections at the same time, stretching it across the city may help.
RECOMMENDED ARTICLES
