Goodbye Gridlock: How Technology is Making Traffic Signals Obsolete

May 8, 2017

Next time you’re stuck at a red light for what feels like an eternity, console yourself with this thought: the era of traffic signals is coming to an end. Smart cars and connected cities offer a whole new way of managing traffic, meaning that soon you’ll be able to sail through every intersection without worrying about the lights.

The Trouble with Traffic Signals

Intersections are the trickiest part of traffic management. Each intersection contains two or more competing flows of traffic, and some system is required to prevent sub-optimal performance, such as long backups, and catastrophic failures, such as collisions. Automated traffic signals, as we know them, began to appear in the 1920s, offering a cheap and simple solution to the intersection problem.

Unfortunately, traffic signals don’t work very well. Backups are common, millions of gallons of fuel are wasted while vehicles idle at intersections and, worst of all, red-light running accidents cause approximately 800 deaths and 200,000 injuries in the U.S. each year. However, traffic signals at intersections have been the best we can do. Until now.

Traffic Management in the Information Era

Previously, the problem with traffic management was a lack of real-time information: we didn’t know how many cars were on the road, where they were going or at what speed they were travelling.

This is changing in the 21st century. For example, most GPS systems now have live traffic information which is based on data from other users. So, if you get stuck in a backup or gridlock, your GPS will tell other GPS users to take a different route.

Early experiments are underway to include traffic light data in such systems. One trial in South Carolina experiments with Wi-Fi beacons near each traffic signal, broadcasting data that can be picked up on a driver’s smartphone. An app then tells the driver what speed to maintain in order to make the green light. The project has only been attempted on a small scale, but in simulations it predicts that the number of cars stopped at red lights may drop by a factor of 1,000.

As cars become smarter, they will be better able to interact with their environment and each other. Within the next decade, self-driving cars will become common and this offers two potential options to manage flow in an optimal way with zero collisions.

One vision is a peer-to-peer model in which self-driving cars communicate with each other over Wi-Fi and make their own decisions about how to proceed through an intersection. Imagine two cars on a potential collision course: they broadcast their positions and speed toward each other, calculate a way for both vehicles to proceed safely, and adjust their speed accordingly. Both vehicles pass through the intersection, with no need for a traffic signal.

In busier networks, such as cities, traffic management may work better if it is centrally controlled. This would mean replacing traffic signals with a network of Wi-Fi beacons that broadcast instructions to each vehicle, telling them which direction to take and the speed at which to travel. A central server can then manage the journey of each vehicle, ensuring that traffic in the city is fully optimized.

This centrally controlled model is essentially the same model used in air traffic control. Aircraft are data-rich machines, filled with sensors that transmit data to the pilots and the ground, making it possible to perform analytics and create sophisticated models. Researchers at MIT are looking at how to adapt those systems for vehicles on the ground.

Controlling Flow in Other Scenarios

Manufacturing is also catching up on the big data revolution. Assembly lines and road intersections are, conceptually, quite similar — you need to optimize throughput, avoid waste and prevent catastrophic failures, the breakdown of a machine, for example. Industrial Internet of Things (IIoT) makes it possible to build smart factories that offer real-time data about every event in the facility, including identifying machines that are in need of repair before they fail.

Smart factories generate production data clouds, just as smart cities and self-driving cars will generate traffic data clouds. Analyzing that data to create optimal systems is no mean feat, and until we can do that, we will have to keep sitting in traffic until the light turns green. But rest assured, for traffic signals, their days are numbered.