Smart Cities

These startups want to help cities keep the trains running on time

A look at the next generation of transit tech.
article cover

Illustration: Morning Brew, Photo: EschCollection/Getty Images

· 5 min read

This article is part of our series on smart cities—click here to visit the interactive hub and explore other stories.

If you took the subway recently in New York City, your train probably arrived thanks in part to a nearly 118-year-old signaling technology developed during the early days of the New York subway system. The train operator likely used the same signaling technology to know which track to take to drop you off at your destination and when it was safe to proceed along to the next section of the underground system.

But this technology, advanced as it was upon its installation, is a bit dated in our modern times. As New York’s Metropolitan Transit Authority wrote on its website, “With our legacy signal system, all we know is that some part of a train is in some part of a section of track—these sections can be over 1,000 feet long.”

And New York is not alone in needing tech upgrades, at least according to a handful of companies working on next-gen train tech. We spoke with the startups looking to remake the hardware and software that helps subways shuttle people around cities.

Tracking the tracks

4AI Systems, an Australian startup specializing in AI for rail systems, focuses on supporting train operators.

Joanne Wust, president of the company, told us its software, which is “basically a sensory system for rails,” can work with many different inputs: cameras, GPS systems, lidar, radar, and other types of sensors. The 4AI Systems team trains its custom-built model and the neural networks that fall under it in part with data collected via a trial on the New York City subway’s Canarsie line.

The goal: Parse incoming sensor and camera data to give train operators “time-critical decision-making” tools, Wust said, like alerts about blockages, human heat signatures on the track, people walking too close to the edge of a platform, and more. It can also integrate with train control systems to alert train operators about upcoming signals, such as which track to take. The software is all about real-time alerts as opposed to post-processing work, such as flagging potential future maintenance needs.

So far, the tech is in trial mode only.

Meanwhile, Alstom, a transit manufacturer headquartered in France, specializes in communication between trains themselves. The company aims to modernize communication-based train control (CBTC), which uses a radio network to signal which parts of the tracks are currently occupied by a train’s wheels, and, in turn, prevent other trains from entering occupied sections of the tracks. Essentially, computers in trains communicate with track-side equipment for permission to move forward and instructions on where to go.

Susie Levesque, customer director at Alstom, called the company’s product an “evolution” of newer CBTC systems, eliminating much of the track-side equipment and focusing on trains directly communicating with each other, a process that could speed things up and lower latency, potentially leading to increased capacity. She said Alstom’s system is currently deployed in Shanghai; Paris; Lille, France; and Turin, Italy.

Second sight

Platforms like 4AI Systems’ and Alstom’s could evolve based on the information they’re able to collect, and one way they may be able to advance further is through lidar sensors.

Stay up to date on emerging tech

Drones, automation, AI, and more. The technologies that will shape the future of business, all in one newsletter.

“They’ve been using a version of lidar, distance-measuring equipment, in subways and rail systems for decades,” Brent Schwarz, VP of business development at Luminar, told us, adding that cylindrical devices above tracks measure distance and signal train operators when to stop. But, he said, they’re “kind of primitive. What lidar does is: It takes those systems and takes it to the next level.”

When it comes to subway systems, one of lidar’s main functions would be to enable better real-time train positioning so that transit agencies can safely increase system capacity. In effect, it’s a similar goal to upgrading CBTC—just with a different tool.

Used in conjunction with software, lidar could help not only pinpoint train positions but also collect real-time data on obstacles falling on tracks, flag future maintenance issues, keep an eye on vegetation growing close to tracks, spot people coming too close to the edge of a platform, and more, according to both lidar companies we spoke with.

Lidar can also be used to help create a “digital twin” of a subway system, according to Itai Dadon, VP of smart infrastructure at lidar startup Ouster. The company uses a digital twin of train tunnels mapped out by its sensors to help clients plan system improvements, flag potential issues, and more. Currently, Dadon said, its sensors are used in the Hong Kong metro and in Calgary, Alberta, Canada.

“Think about the tunnels themselves and the operations of the trains—so there are a lot of safety components there on assuring that the rails are not obstructed by any object,” Dadon said, adding that the system’s digital twin can be used to help conceptualize infrastructure maintenance.

For its part, Luminar is working with at least four major transit manufacturers, Schwarz said, but declined to name them. He still expects it’ll take several years for Luminar’s lidar to be widely deployed on tracks.

“Rails take a long time to make decisions,” he said. “Rails are incredibly harsh environments…The trains have extreme vibration—much more than cars, trucks, or even most military vehicles, and they [often] don’t get a lot of maintenance. So getting new technology into rail systems is [a] really, really difficult and very [long] process. Big trains, they expect them to last 50 years…Lidar [sensors are] not going to last 50 years [integrated into a train].”

Update: This piece was updated 11/1 to add additional context to a quote.

Stay up to date on emerging tech

Drones, automation, AI, and more. The technologies that will shape the future of business, all in one newsletter.