Traffic lights have been around for nearly a century and in all that time they haven't changed much. Some alternate the times of green, red and yellow in a fixed way, regardless of the traffic; others, more recent, have sensors that adjust the lights according to the number of cars in front of the traffic lights. Henry Williams explained in the Wall Street Journal what some startups and auto companies are doing to ensure that, thanks to artificial intelligence, “city traffic becomes a distant memory”. Extensive program.
But even without numbers and data, anyone who has been in a car knows that traffic lights are useful but certainly improvable tools. Dorsa Sadigh, a professor of computer science at Stanford University, told the Wall Street Journal that there is research that has shown that in a hypothetical city without pedestrians or human drivers – therefore only with self-driving cars – traffic lights they would be useless. Self-driving cars could indeed communicate and regulate each other on how to manage traffic. Since at least pedestrians will continue to be there for a long time, however, they are trying to work out ways to keep traffic lights by making them smarter and more efficient. The idea is to create traffic lights that, in addition to looking at what they have in front of them with a sensor, can communicate with cars and with all other traffic lights, to make traffic more fluid. Traffic lights equipped with artificial intelligence would also allow certain vehicles – such as ambulances, police cars and public transport – to have some sort of right of way.
The first thing to do in order to have an artificial intelligence that works is to provide it with as much information as possible. It is the reason why many startups are collecting every possible data on the traffic of certain intersections or certain cities.
Rapid Flow Technologies (RTC) is working on a traffic management plan called Surtrac, which is already in operation in East Liberty, a neighborhood in Pittsburgh, Pennsylvania. The traffic lights of East Liberty where Surtrac has been installed communicate with each other, disposing of traffic based on the data available to them: Greg Barlow, one of the founders of RTC, said that thanks to Surtrac «an intersection decides how to program based on what your sensors see and what everyone else sees ”and that traffic on East Liberty has dropped 42 percent since Surtrac. In addition to allowing people to arrive earlier, Surtrac also reduces braking, stops and starts, reducing pollution. RTC's next step will be to develop a system to allow drivers to share their intended itinerary so the system can adjust accordingly. Meanwhile, RTC is about to expand into some areas of Atlanta and other cities in the northeastern United States.
The Wall Street Journal also spoke about Vivacity Labs, a startup from Milyon Keynes, in the UK. For now, it doesn't have a system already operating on traffic lights, but the interesting thing is that its sensors «don't just collect information; each sensor is in fact a powerful computer attached to a camera and is capable of analyzing the traffic that the camera sees at the intersection », also making predictions about future traffic. As Williams wrote, apps like Google Maps capture information in real time: they say what the situation is like but have a hard time saying what it will be like in a few minutes. The Vivacity Labs system, on the other hand, allows us to “predict how traffic will develop”.
There are many things to do. Karina Ricks, Pittsburgh's head of mobility and infrastructure, said there is for example a clear difference between a bus with thirty people on board or a bus with one person, and that a really smart system should take that into account. . Another problem is that traffic lights alone don't have enough data: cars also need to let them know where they are and where they are going. Something is already moving: those who drive certain Audi models in Las Vegas, for example, receive information about some traffic lights before they even get there: whether the traffic light towards which they are about to arrive is red or not, and in how many seconds it will change color.
Finally, two curiosities: they have nothing to do with artificial intelligence but they say why almost a century ago they chose to use green, red and yellow. As Quartz explained, they are the three colors with the most wavelengths of the visible spectrum: the most recognizable, in short. The first railway traffic lights used white for “go” and red for “stop”, but it seems that there were problems due to the breaking of the red lenses, which made the lights become white (so that the trains that should have stopped thought instead of having to go). It is said that during the Cultural Revolution in China, Mao Zedong thought of inverting green and red (because red, the color of the revolution, was supposed to invite us to continue, not stop), but nothing came of it.
