Early tests of a new machine learning system by Google have shown some promise in enhancing traffic signal patterns. The Green Light project models traffic patterns to provide optimal traffic light sequences by utilizing machine learning and artificial intelligence. According to Scientific American, early experiments have led to somewhat improved traffic flow on some congested routes in Seattle. The transportation authorities in Seattle stated that Green Light verified known choke locations, helped detect traffic bottlenecks, and offered practical recommendations.
Google's Green Light trial program, which was introduced in the fall of 2023, is being tested in traffic-heavy metropolises like Rio de Janeiro and Kolkata, as well as in Seattle and twelve other cities, including Hamburg. During these experiments, traffic signal sequences are being modified by local traffic engineers based on recommendations from the system. The initiative's goals are to shorten traffic light wait times, enhance key road and intersection traffic flow, and eventually reduce greenhouse gas emissions. According to Google, preliminary research suggests that CO2 emissions might be reduced by 10% and traffic light stops could be reduced by up to 30%.
At its core, Green Light is an artificial intelligence (AI) model that is customized for every intersection, taking into account elements such as its layout, driving and stopping habits, traffic light timings, and the interplay between signal and traffic systems. The technology creates recommendations based on these patterns, which urban planners can access via a dedicated interface. The project's lack of need for expensive fixed sensors or ongoing on-site monitoring is a big plus. Rather, it makes use of already-existing Google Maps traffic data, which is gathered from moving cars and smartphone users who serve as "mobile sensors."
Green Light's suggestions aren't always reliable, though. The director of the University of Michigan's Transportation Research Institute, Henry Liu, is more circumspect about the technology. Although Green Light was able to cut intersection wait times in Birmingham by 20 to 30 percent, Liu points out that the degree of effectiveness varies depending on the starting point. In Birmingham, for example, traffic lights follow set timetables that are based on antiquated traffic information. The benefits to the environment are also debatable because, according to government data, idling cars and traffic jams only account for around 2% of traffic-related emissions in the United States. In actuality, cars that are traveling faster than the speed limit use a lot more fuel than cars that are stopped at red lights.
Additionally, Green Light fails to take into consideration complex elements that may lead to less useful recommendations, such as crossing bus and bike lanes, trams, or busy pedestrian crossings. A traffic light adjustment in Seattle had to be reversed since it eventually proved to be ineffective. Another test city, Manchester, has traffic engineers who routinely decide to ignore Google's advice since the traffic signals there are sometimes purposefully programmed to favor buses or to make commuters in residential areas factor in extra time. The AI's strategy of reducing stops at crossings may work against it in some situations.