The Surprisingly High-Stakes Fight Over a Traffic-Taming ‘Digital Twin’

Imagine driving through Los Angeles in the year 2040. There’s a mix of self-driving and human-controlled vehicles on Santa Monica Boulevard. A serious collision slows traffic to a crawl. But then a special orchestration of traffic signals flips on, parting the sea of cars for an ambulance to throttle through the streets.

This traffic engineering fantasy may be inching to reality, as companies such as IBM, Microsoft, Google, and HERE Maps develop what’s known as “digital twin” technology. The term describes a virtual simulacra of something in the physical world—whether it’s a car engine, a casino floor, or the street network of a major city—that visualizes real changes as they occur, and is “smart” enough to model possible scenario outcomes. In the L.A. example, imagine that a downtown city worker viewed a traffic simulation seconds after the car crash and approved a recommended route for the ambulance, alerting all those connected self-driving vehicles to move aside.

But if the phrase “digital twin” strikes up images of a pixelated doppelgänger dogging your commute, you’re not necessarily wrong to feel creeped out. And you might not have to wait very long to find out if any of those fears are justified: Next week, transportation officials from 13 major American cities will discuss (among other items) whether to collectively to build towards such a model.

“Going forward, each city must manage its own Digital Twin, which will provide the ground truth on which mobility services depend,” states the bylaws of the Open Mobility Foundation, a new nonprofit that counts city leaders on its board of directors.

Launched in late June, the Open Mobility Foundation describes itself as a “public-private forum” to help local governments gain control of their roads from private mobility companies, using big data and open-source code. A central part of OMF’s mission is to govern the the new mobility data standard, commonly known as MDS, unveiled by the Los Angeles department of transportation last year. Currently, MDS pulls in rich, real-time status information about dockless scooters and shared bikes. Many other cities, including Miami, Seattle, Portland, San Francisco, Austin, Minneapolis, and others that have joined OMF, have adopted it.

At their first meeting this coming Monday in Louisville, board members will vote on whether to adopt a set of bylaws that were largely authored by LADOT. A section called “design principles” states that OMF’s work will be “based on the ‘digital twin’ model […], which specifies that municipalities own and control a definitive digital data model of urban mobility.” Having a virtual replica of real-world mobility flows—for scooters and bikes now, and for ride-hailing cars, AVs, and drones in the future—would allow local governments to both trace the movements of individual vehicles, and control them to some extent.

But this notion of a traffic command system is ringing alarm bells in some parts of the small-but-very-chatty world of transportation technology. For one, mobility companies aren’t all thrilled with the idea of cities achieving the power to redirect their vehicles. The concept is also serving as a lightning rod for technologists who work closely with cities, privacy advocates, and some public officials. Everyone wants safer, smoother, more manageable streets. But there are competing visions of how to get there. For a story that starts with scooter data, the stakes here are surprisingly big.

A turning point in the Great Data Wars

The idea for MDS dates back to 2016, when Seleta Reynolds, the general manager of LADOT, was hearing predictions about how autonomous cars might transform cities. To regulate her streets away from becoming a “zero occupancy vehicle” nightmare, she’d need the vehicles to communicate their whereabouts in a consistent manner. In May 2018, the city awarded a contract to a consulting firm called Ellis and Associates to develop a new data language for that purpose.

This was as a major turning point for city officials. For years, Uber and Lyft have prevented cities from accessing information about vehicle locations, car counts, timestamps, and routes. Local governments are eager to use this data to improve traffic flows and regulate cars-for-hire, which have been shown to contribute to congestion and draw riders off public transit. Ride-hailing companies have preferred to set their own terms, claiming that cities lacked the technical skills to process the data they wanted, as well as a vision for applying it. Some metros, like Seattle and New York, have succeeded in capturing data, but Uber and Lyft have also gotten state laws passed to preempt similar attempts.

In 2018, with scooters all but falling from the sky across the U.S., cities resolved to stay ahead of them. Though it had been intended for the cars of the future, MDS also looked like a way to get a handle on the new conveyances that were tripping and delighting Angelenos in equal parts. In September, the city launched a scooter pilot that required participating companies comply with the new standard. Bird and Lime—wary of sullying their relationships with the city, as their ride-hailing predecessors had done—jumped aboard. “At LADOT, our job is to move people and goods as quickly and safely as possible, but we can only do that if we have a complete picture of what’s on our streets and where,” said Reynolds, after announcing the first wave of companies that had agreed to comply. “That’s what this partnership is all about.”

Soon, other cities dealing with their own scooter surge flocked to adopt the new data format. By December, 10 U.S. cities—Austin, Detroit, Kansas City, Miami, Minneapolis, Portland, Seattle, San Francisco, Santa Ana, and Santa Monica—required micromobility providers to share data using MDS. They also flipped on a program that L.A. had built, called “Provider,” that lets companies send the city near-real-time data about individual vehicles’ trips.

A man on an electric scooter rides through the “Four Ladies of Hollywood” gazebo on Hollywood Boulevard in Los Angeles. (Richard  Vogel/AP)

But not everyone was a fan. Uber, in particular, has consistently pushed back on the city’s somewhat casual approach to protecting the privacy of all this data being gathered. “We would really love to see a global standard for planning and enforcement,” said Melanie Ensign, a security and privacy communications officer at Uber. ”The challenges we have are a result of the haste in the way LADOT created MDS.”

The ride-hailing titan complained that the city was collecting too much disaggregated, detailed route data that could be potentially be used to identify riders. Numerous studies have proved this as far back as 2014, when researchers identified the actor Bradley Cooper’s taxi rides from a publicly available data set released by New York City’s taxi commission. And a law enforcement agency, whether its’s the local police department or ICE, could also potentially gain access to identifiable information through a records request. Similar scenarios have already occurred.

Los Angeles eventually issued a set of high-level data privacy principles, and has said that it will protect MDS data from outside requests. But this didn’t do much to satisfy critics. Of course, Uber itself has a reputation for loose data protections and unethical user tracking, but they were joined by other dissenters. “LADOT needs to address the serious privacy and civil liberties issues implicated by the Provider API before moving forward with any further stages of this policy,” stated a joint letter from the Electronic Frontier Foundation and the Open Technology Institute this year.

Los Angeles pushed ahead anyway, and turned on a second program, called “Agency,” that allows it to communicate directly back to bike and scooter companies, alerting them if their vehicles are in the way of a street closure or out of a parking zone. “It’s firmly within my ability to manage and operate the transportation networks, and that’s where we want to stay,” said Reynolds.

But as the city flexed, Uber, Lyft, and Bird squawked. The companies sent a letter of support for AB 1112, a California bill that would block cities from collecting trip-level data through a platform like MDS and regulating micromobility companies across numerous dimensions. “Provider has broader concerns when it comes to privacy,” said Ensign. “Agency is where we run into questions of whether cities should have the ability to dictate where people can travel in city.” For example, should a city have the power to disable a dockless scooter because it’s the wrong spot? Agency isn’t being used that way currently, Reynolds said. But it could.

The threat of state preemption is enough to give any city official pause. In the first half of 2019, Reynolds decided to transfer the governance of MDS to an outside entity with a more formal decision-making structure than Los Angeles alone. In late June, the Open Mobility Foundation was formally announced, with funding from the Rockefeller Foundation, advisors from the open-source standards group OASIS, and corporate members including Bird, Spin, and Microsoft. (Uber and Lyft weren’t invited to join, Wired reported.) A press release announcing OMF describes a mission “to promote safety, equity and quality of life,” at a time when “the number and type of vehicles using the existing public right-of-way rises dramatically in cities across the country.”

One of the first orders of business at Monday’s OMF board meeting will be to vote on whether to adopt the new bylaws, including the parts about digital twins. “The board can decide to jettison it all,” Reynolds said. But what leaders will be mulling goes far beyond little scooters.

The rise of the digital twin

An early proof of concept for digital twin technology came in 2014. Researchers from the University of Washington announced that they had created “a self-organized and scalable multiple-camera tracking system that tracks humans across the cameras” by pairing Google Street View with a clutch of smart surveillance cameras, trained on city streets. They had demonstrated the possibility of building a close-to-real-time, changing visualization of mobility flows, mapped onto a Google’s panoramic photograph of the world. Machine-learning software rendered any seams invisible.

Since then, the term and technology behind the digital twin concept has gained traction. Microsoft, IBM, Google, Descartes Labs, HERE Maps, and other companies are engineering AI-powered simulacra of brick-and-mortar environments. Some use cases are totally quotidian, like office space designers rearranging digital desks and chairs to prepare an actual floor-plan. And it’s not a totally new concept for some cities—in India, for example, several cities are using digital twin software to manage water and power infrastructure. Portland, Oregon, recently began testing Sidewalk Labs’ Replica, a software that uses realistic-but-fake datasets to model transportation flows.

To date, though, none of these approaches have quite reached the level of direct oversight envisioned in OMF’s bylaws. It takes the current powers of MDS and its various data-exchange programs to a much higher level. A traffic planner of a centralized control system would be alerted to changing conditions in real time, as notified by the real-time streams of data constantly redrawing their digital terrains. Built-in AI would inform an engineer about the best decision to make. OMF offers a couple of examples of how this super-advanced SimCity would run traffic simulations:

As a virtual, ‘living’ equivalent of the city’s many systems, a Digital Twin allows a DOT to model possible strategies to plan for and mitigate problems before and as they occur, and to implement a solution which has been virtually tested in many simulated scenarios to minimize risk. For instance, if a stadium lets out 20,000 patrons at 9:30 pm after a given event, what is the best temporary, one-way street configuration? And for how long should the temporary configuration remain in effect? City planners and operators can use dashboards which provide access to different planes of the physical and virtual worlds to gain the insights needed for effective decision making.

This kind of omniscience is a far cry from the obstructed views of urban movement that some mobility companies now provide to cities. Those are batches of highly aggregated data, such as Uber’s trip data visualization platform, Movement. “These systems represent partial, private and conflicting views of the world that are at odds with the needs and priorities of the city and its residents (and may contradict or attempt to countermand each other as well),” the OMF bylaws state. “Going forward, each city must manage its own Digital Twin, which will provide the ground truth on which mobility services depend.”

Some of this sounds great—a city that’s able to rush ambulances through traffic by rerouting other drivers with a few keystrokes might save lives. Still, some officials and data experts wonder about the excess of individualized information that a digital twin would require a city to own, heightening the privacy concerns that are already clouding MDS. OMF’s bylaws mention some broad privacy principles, but critics—among them, leaders of rival mobility data companies—say that that’s working backwards: Step one should be to hammer out how cities are going to take care of protecting citizen identities, not to keep scaling up their surveillance powers.

“I don’t even understand how that vision is compatible with any notion of privacy,” said William Henderson, the CEO of Ride Report, a startup that worked closely with Los Angeles and other cities to provide mobility data management software. “The vision itself is certainly not the one that we share, and not one that all the cities share.”

Regina Clewlow, the CEO of Populus, a similar software startup that helps cities translate and analyze mobility data, agreed. “Based on our conversations with cities today, we find that many are primarily concerned with having enough detailed data to inform and enforce parking policies, vehicle caps, and equity requirements—not to establish real-time control systems, at least not for scooters,” she said.

There is truth to that. Mainly, local governments are using MDS data descriptively, like checking in to see where bikes and scooters are clustering so that they can improve infrastructure, or to make sure companies are complying with various requirements. Robin Hutcheson, director of the Minneapolis Department of Public Works and one of OMF’s directors, extolled the “data dashboard” that her city procured, which helps her verify that scooters are concentrated in poorer neighborhoods, one of the city’s requirements for permitting.

That’s pretty different from establishing real-time control systems for scooters. And it’s telling that no other city except L.A. has adopted “Agency”—the program that allows them to directly communicate back to companies when a scooter is parked incorrectly or need to be cleared away. They’re also stepping more cautiously down the path of telling companies what to do through a digital interface.

Still, some cities are open-minded to the idea of a virtual command center. “We’re not on the same points as L.A. on a lot of things,” said Noah Siegel, the interim deputy director of Portland’s bureau of transportation. In contrast with L.A., Portland’s city council recently released a set of privacy principles that will inform how all of its agencies handle personal data. “But I think the twinning approach is a strong starting point for discussion. The notion that cities should direct trips based on the system as a whole isn’t outrageous.”

Cities mull the future

In urbanist circles, Los Angeles is often lauded for gaining an upper hand in what’s often described as a battle between public versus private interests. Across cities and industry, there is broad consensus that a) cities should be able to manage the roads they own and that b) data is a powerful way to do so. Even with its blemishes, MDS, as a data-sharing standard, seems to be a broadly good thing.

Indeed, it sets the stage for the future: If and when autonomous vehicles arrive en masse, a common framework could make it easier for vehicle makers to send collision reports and car counts to authorities. Theoretically, it could also help cities set up more complex versions of congestion pricing, in which drivers might be charged fees to enter certain popular zones at specific times based on how many cars are already there. Policy wonks tend to say that such a scheme represents the best hope for cities to reduce the carbon emissions rising from the roads. Some kind of digital bird’s-eye-view to help with all of this could be useful.

But L.A.’s haste to implement these far-reaching technologies might spoil those pay-offs by dismissing the most important questions, critics say. For example, could aggregate data answer many of the same questions that planners are after? Uber clearly thinks so. AB 1112, the data-collection-killing bill in California, was put on hold, but there is a good chance it will return. And, is a digital twin that’s built to let cities actively manage streets—as opposed to the more passive approach that most cities currently have—really the best model? That probably deserves an open, transparent dialogue between public officials, companies, and trained academics, Clewlow said.

Indeed, to some minds, the potential issues go beyond data protections. Major ideological differences should be sorted out first, said Kevin Webb, the co-director of SharedStreets. “The bigger question is, what’s the purpose of the technology we deploy in cities?” he asked. “Are we using technology to manage streets or to manage people? All of the concerns about privacy hinge on that question.”

For their part, the cities joining OMF say that they hope to learn about what their peers are doing with data, and to have a role in shaping important conversations about the future of urban mobility. “Reasonable minds can differ over certain facts, but I don’t think that should distract from larger visions,” said Siegel in Portland. “As public leaders, we need to have public regulations, and MDS is one tool that will help us get there.”

Added Hutcheson of Minneapolis: “We’re excited about the opportunities that MDS has provided, and we want to be able to learn more.”

What’s most important for Reynolds is that Los Angeles is no longer the de facto decision-maker behind the technology driving the streets for decades to come. Now it’s up to other cities to help decide which way to turn.  “I wanted there to be a place where the owners of the transportation infrastructure are also the caretakers,” she said. “We want to prevent people from monopolizing the streets, or the choices our citizens have. I didn’t want to get too far ahead in L.A. to say, ‘This is the authoritative way.’”

Powered by WPeMatico