Police officers — the original traffic safety technology—patrol an intersection in Denver. (Keith / Flickr)

The next-generation intersection helps all modes share the street

Emerging traffic technologies promise to improve safety while moving everyone — not just drivers.

In their quest for safer streets and zero traffic deaths, more and more cities are focusing their attention on the intersection. For all its familiarity and organizational power, the intersection is also a source of great stress and frustration. It can be dangerous: Roughly 40 percent of crashes occur at intersections, and about half of fatal crashes involve turns. The traditional diagram used by traffic engineers for a typical intersection shows no fewer than 56 conflict points for vehicles and pedestrians.

The intersection’s safety concerns have gained attention as people turn to other modes besides cars to get around city streets. Many cities now encourage cycling, walking, public transportation, and other driving alternatives to improve sustainability, equity, and affordability. They’re reducing speed limits for cars to increase safety for pedestrians. They’re converting road space into transit lanes or bike-share stations or pedestrian plazas. In many ways, cities are going back to the future — prioritizing the type of shared street access found in the days before cars.

These trends represent the next phase of urban mobility, but the tools that cities use to manage their intersections haven’t kept up with the times. The good old traffic light does a great job organizing complex car movement and optimizing for travel speed. It wasn’t built for the tasks of improving safety and moving people using other modes.

There’s no simple solution to crowded crossings, but it’s possible to imagine how digital technology that wasn’t available even five years ago carries the promise of a more responsive intersection. Sensor arrays and computer vision can make detection of multiple street users readily achievable, potentially providing a more accurate count of all modes and informing decisions about traffic flow and priorities. A step further out, machine learning is becoming more capable of improving large, complex systems not unlike traffic. On-site processing can reduce the need for expensive fiber installation and enable more security and data privacy.

To understand why these tools represent a step up, it helps to get a better sense of the problem in the first place.

Old tools for new streets

The original intersection technology was the humble traffic officer. When cars first hit city streets, in the early years of the 20th century, the chaos of pedestrians, carriages, and motor cars all fighting for space required the type of real-time judgement, discretion, and adaptability that only a human being could deliver. This personal oversight was the safest, most equitable way for everyone to share the road.

Cars soon won the battle for city streets, and the focus on one mode made it possible to adopt the more simplistic street controls that cities rely on today — primarily, the traffic light. The function (and even the form) of this tool hasn’t changed much in decades. It’s little wonder that when cities today want to manage an unruly or complex intersection they don’t rely on a traffic light at all — they send in a traffic cop.

When cities want to manage an unruly or complex intersection they don’t rely on a traffic light at all — they send in a traffic cop. (Nick Harris / Flickr)

The traffic light’s limitations make it hard for cities to adapt to modern mobility challenges. Fixed-time signals, for instance, are unable to adjust if an intersection is suddenly filled with people jogging or riding a bike on a warm spring afternoon, instead of processing its usual queue of cars in the dead of winter. They’re oblivious to the amount of cars going in each direction, sometimes stopping a long queue for a short one. Sensors for actuated signals can detect a queue of cars, but most can’t detect pedestrians, cyclists, or transit vehicles, let alone prioritize these modes.

Cities use various approaches to address their street challenges, and many don’t involve new technology. Some areas apply urban design upgrades, such as reverting from traffic signals to roundabouts or simple four-way stops or Dutch-inspired woonerfs. These controls are equally oblivious to the street environment, but at least they calm traffic and improve safety.

Other cities have started looking at adaptive signal control options, such as SCATS, SCOOT, ATC-Lite. These adaptive systems have been applied in varied situations with improvements ranging from drastic to marginal. But such systems are rare; most U.S. cities rely on fixed-time signals or centralized control via Transportation Management Centers. At TMCs, operators review a live feed of traffic conditions at an intersection and choose from a series of pre-approved signal plans that may or may not fully address the given problem. The lag in response can be torturous for those affected by it, and the systems still struggle to account for alternative modes.

The suite of next-generation traffic technologies — while still in need of much testing and refinement — at least holds the potential to help cities manage their streets with the rapid decision-making ability of a traffic officer. Their heightened awareness of all travelers gives hope of improving street safety, while also helping to move all people instead of just drivers.

Making intersections safer and fairer

When you reimagine traffic signals with these advanced capabilities, you begin to see some key potential benefits in terms of traffic safety and transportation equity:

Safety. An intersection that knows both the speed and volume of all modes can implement changes that warn people about collisions or hazards. That adaptability could prove especially valuable to cities pursuing a Vision Zero policy of eliminating traffic fatalities.

For instance, tech-enabled signals could slow down vehicle traffic if average car speeds at a given intersection exceeded 20 mph (the threshold at which collisions with pedestrians become more deadly) or if they recorded a high number of near-misses. They could also hold crosswalk signals to leave more time for children, the elderly, or the disabled. If combined with new alert systems, in-car systems, or V2X technology, they could also potentially give drivers a more direct warning of pedestrians or cyclists in the area.

Such lights would be safer for people in cars, too. At an urban arterial with fixed timing, drivers may speed up when the light turns green, only to hit the brakes for a red light at the next intersection. This endless acceleration and deceleration increases the risk of rear-end collisions, and it encourages drivers to go fast in the hopes of making the light. But a corridor optimized for a steady flow could space out cars and give drivers more buffer space to react to a lane change or sudden stop.

People throughput. Traditionally the “success” of an intersection has been judged by its ability to move as many cars as it can as quickly as possible. That definition is changing. In January, the Federal Highway Administration announced that it would start to measure road performance in terms of how many people get moved, not just cars, giving more priority to transit vehicles, pedestrians, and cyclists.

One big advantage of measuring people-throughput is better coordination of existing infrastructure. By giving priority to people over cars, adaptive intersection technology can encourage travelers to shift from driving onto other modes. And (nerd alert) traffic isn’t linear, which means removing just a few cars from the road can have outsized improvements on congestion, at least in the short term.

In addition to accommodating all modes, adaptive systems can help improve commutes. The trip to work is far more frustrating when it takes 20 minutes one day and 40 minutes the next, compared with a steady 30-minute commute, even if the average time spent traveling is the same. By maintaining consistent traffic speeds—even at lower, safer speeds—adaptive intersections can make commutes more reliable and predictable for drivers and transit riders alike.

Again, tech-enabled, adaptive traffic signals alone won’t solve the problems of road safety or traffic congestion in cities. Nor will they eliminate the need for sharp transportation policy, engineering analysis, or physical improvements. If anything they increase that need, since local communities set the priorities for traffic controls to follow. And there are still plenty of other challenges facing these systems: counting pedestrians, cyclists, and transit riders at an intersection accurately with digital tools; securing all the data; and keeping the cost of such technologies within the reach of your average city government.

I know, I know. We’re heading toward a future of flying autonomous vehicles that could make traffic lights obsolete. But until that day arrives, there are plenty of steps cities can take toward safer streets that work for everyone.

Follow what Sidewalk Labs is thinking, doing, and reading with our weekly newsletter, or on Twitter and Facebook.