Whenever people go from one place to another, they don’t think much about the roads and sidewalks that pass beneath them. But this infrastructure, known as the public right-of-way, doesn’t work by magic. It is managed and regulated by specific laws. People don’t own the roads they travel on, but streets and sidewalks provide an easement—a right of use or passage separate from that of ownership.

For example: a single-family homeowner usually owns the sidewalk that flanks a property, particularly if that sidewalk falls behind a tree-planting strip that separates it from the street. By local standard or writ, the homeowner grants an easement to the general public to use the sidewalk, utility companies to use the curb where utility lines run, and so forth. Similarly, cities, towns, and the federal government own the roads, but they grant easement to the public to make use of them for transport. The relationships between private and public landowners is managed by local land use and planning agencies. If a builder wants to construct a new driveway from an undeveloped private property to the public right of way, it needs to get permission to do so from the appropriate public agent. That same agent is responsible for maintaining the roads so that they are usable. Some roads, like freeways, don’t offer any private access and restrict public easement to vehicles.

Unfortunately, America’s roads have seen better days. After a massive investment in new infrastructure since the mid-20th century, streets, roads, and freeways have ossified. New roads are tough to build in established cities, and existing ones are increasingly difficult to improve. When roads do get built, they are usually constructed for new development in suburbs or subdivisions. Fear, local resident entrenchment, and lack of funding has hampered adequate upgrades of roads, too. America’s car cities—Los Angeles, Houston, Atlanta, and Dallas, for example—have endured increased congestion when more and more cars travel on roads built for far fewer.

Even when the roads are clear, they are often in poor condition. The economic crisis of 2008 reduced the tax base that supports public service in the near term, but it also inaugurated a prolonged program of social austerity justified by that supposed crisis. And the longer maintenance and support gets deferred, the more complex and costly it becomes to catch up.

The state of public roads might have a surprising and substantial impact on the role of self-driving cars. Cars need roads to drive on, so the technology companies at the center of the robocar revolution might become increasingly invested in them. But roads themselves might become inaccessible to citizens if driving becomes fully autonomous.

*  *  *

If one thing’s clear about cities of the future, it’s that they will become more and more intertwined with the technology business. Public-private partnerships—mergers of private and public funding and support—are helping make costly, unpopular urban development projects more palatable, including tech-enabled transit. Altamonte Springs, a city of 42,000 residents north of Orlando, Florida, for instance, is set to begin offering Uber public subsidies to service its residents. The initiative, which is worth up to half a million dollars a year to the ride-hailing company for starters, effectively turns it into a private provider of public transit.

Eventually, other, larger cities might choose such arrangements to spare themselves the cost and trouble of maintaining public transit services. Such an outcome wouldn’t be entirely new, either. Before the car took over after World War II, public transit was mostly privately owned and operated in America. A return to privatized transit in the form of car-hailing services would make a poetic victory for the automobile over the train and the bus.

A welcome one, too. Citizens are becoming more and more accustomed to the conveniences of app-based services like Uber and Lyft. Google’s parent company, Alphabet, reportedly has 100 people working on future urban design projects at a new entity, Sidewalk Labs.

It’s not hard to imagine a near future in which municipalities like Altamonte Springs might use the cost savings (or direct investment) of companies like Uber, Lyft, or Google to invest in much-needed maintenance and updates. Once cars become autonomous, the benefit of investments like these will increase even further, since the cost and liability of human drivers can be averted. Better roads mean lower maintenance costs.

Substantial, new infrastructure development might even be impossible without private-sector investment. For example, engineers have been researching future electric motorways that could charge electric vehicles while they drive. These roads would be very expensive in the near-term, but by reducing the weight and complexity of batteries in vehicles, it could pay off in the long term—but only with the leverage of scale. It’s hard to imagine many governments investing in such infrastructure right now, but if tech or automotive companies end up building new fleets of autonomous, electric car-service vehicles in the future, the opportunity to outsource power to the roadways might become appealing enough to “wire” the roads like Google is currently wiring cities for fiber broadband.

Public-private partnerships for roads might begin the erosion of the public right of way. But it’s also possible that autonomous vehicles will all but require limited access to public roads to operate effectively.

Today’s self-driving cars have to be designed and programmed to interact with messy circumstances. Pedestrians, dogs, bicycles, human-driven vehicles, and other obstacles all pose challenges to robocars, and if autonomous vehicles are even modestly successful, avoiding collisions with fallible human drivers will prove a temporary problem.

That transition could take years, and for the moment it is subject to state-by-state rather than federal regulation. But massive private wealth and widespread public approval among tech companies, along with overall regulatory decline in general and explicit federal support for self-driving cars in particular, make it just as likely that robocars will become widespread sooner than anyone expects.

The more self-driving cars there are on the roads, the less complex and more predictable the overall behavior of traffic becomes. One technological model for both semi- and fully-autonomous cars is vehicle-to-vehicle communication. Once installed with GPS, networking, and sensor arrays, cars will be able to communicate with one another directly, allowing for better routing options and increased safety response, such as adaptive accident avoidance. Drivers could use the relief. Thanks in part to smartphone-caused distracted driving—a circumstance created by the same technology companies that might soon offer autonomous vehicles as respite— traffic deaths last year rose at the fastest rate since 1965. Some proponents have predicted that self-driving cars will prevent so many traffic deaths, human-driven cars will be illegal within the next two decades on safety grounds alone.

The rise of automobiles in the 20th century pushed out pedestrian access in cities almost everywhere in America. Future connected vehicles could perform more sophisticated feats than just replacing human fallibility, ones that might exclude human drivers even from operating vehicles alongside robocars. And with that exclusion, traditional vehicles might meet the same fate in this century that pedestrians did in the last.

Consider intersections. Stop signs, traffic lights, and other common flow control tools give human drivers a common set of conventions for piloting their vehicles. But future autonomous cars would have no real need for them. Networked, computer-controlled vehicles can coordinate with another, evaluating traffic conditions and altering their course in real time to avoid collisions. Like the Lexus-branded pods in Steven Spielberg’s Minority Report, future self-driving vehicles might drive at high speed, coordinating with and around one another in a highly choreographed way. The result might look like the labyrinthine traffic in a chaotic city like Delhi, but without the risk of calamity.

If electric motorways become a reality, it’s even possible that they could be built in such a way that roadways could physically or electronically prohibit traditional vehicles from traversing them. If cars can get smart, why can’t roads? And if roads get smart, it’s likely that they’ll express some of that supposed wisdom through new controls and restrictions.

*   *   *

Once autonomous cars reach a tipping point, cities and the private companies that operate robocar services might voluntarily exclude human drivers from the roads. Not only because doing so could save hundreds of thousands of lives per year, but also because ceding the roads to robots might become a more convenient and economical alternative to sharing the roads with them. Given the opportunity, autonomous car services will likely be willing to invest heavily in redeveloping urban infrastructure that would give them an economic advantage in serving the populations that live there. Especially if it would mean a monopoly on local transit. The results could change city life entirely.

Some consequences of widespread robocar transit are clear already. Parking lots and structures are commonly cited future casualties of the future rise of autonomous cars. As Clive Thompson wrote in Mother Jones earlier this year, “many urbanists predict that fleets of robocars could become so reliable that many, many people would choose not to own automobiles, causing the amount of parking needed to drop through the floor.” Nobody likes parking lots, and it’s possible that robocars could inaugurate an era of redevelopment for these urban blights.

But if parking lots are the organs of the urban creature, roads are its circulatory system. They are the primary means by which people move from place to place—whether they do so by car, bus, foot—or even streetcar or light rail. No matter how hypothetical, the potential eviction of human access to roads ought to raise far more eyebrows than the philosophical gimmickry of robocar Trolley Problems.

Surely not all roads would be made robot-only at once. Perhaps freeways and major thoroughfares would do so first. But a city is a network of roads, and the shift from shared to computerized streets would spread. Roads connect to other roads, such that the transition from hybrid passageways to fully autonomous ones would imply an alteration to the urban landscape not yet conceived. It might be easier just to ban people from roads entirely—even secondary roads and residential side-streets. After all, that’s also how the car took over from the train and the sidewalk in the United States.

The possibility of being locked out of public rights of way is a problem somewhat unique to America. Dense European cities like Paris and London are already criss-crossed by public rail transit and relatively small in physical size. It’s easy and even pleasant to walk across Paris or Barcelona. Crossing Houston or Atlanta by foot or rail isn’t just unpleasant; it’s all but impossible.

Unless the nascent renaissance of pedestrian-friendly urbanism expands significantly and rapidly, the self-reliance cars have provided for decades could reverse course. Instead of offering freedom to maneuver, cars would yoke citizens to the services of oligopolistic technology transit companies.

*   *   *

There are reasons beyond pride and tradition to worry about future autonomous car service providers’ impact on urban life. Social justice, for one. There’s evidence that gig economy companies like Uber and AirBnB make it easier to discriminate against racial minorities, disabled people, and other disadvantaged populations.

American streets have long been a site for racial strife, whether through the federal endorsement of housing prejudice known as redlining, via urban planning that further segregated minority neighborhoods from economic centers, or from the white flight and dissent that stymied the development or expansion of public transit networks in the civil rights era. Once wealthy, non-local technology companies have a direct impact on the financing and planning of urban spaces, the negative impact on less white, less wealthy communities could be severe.

The past offers lessons for the future in this regard. In Atlanta, where I live, urban planners of the mid-20th century effectively zoned the city into distinctly separate black and white areas. As public transit and economic development-driven investments like stadia entered the picture, black communities were physically cut off from white ones. Streets were dead-ended and renamed, hampering navigational continuity. The effects of these choices continue today, invisibly. They would become even more entrenched if the  publicly-passable thoroughfares that already segregate the city were made impassible except by robocar. Historically, thoroughfares have already helped isolate neighborhoods by race and class. If those arterials were impassible like freeways, their divisive effect would only intensify.

There are also economic concerns. We already know that Uber and Lyft are overwhelmingly used by the wealthy and well-educated. Those services are still relatively expensive compared to public transit, even if they are often more convenient. With the exception of private property that explicitly limits public easement, roads and sidewalks currently can be accessed free-as-in-beer as much as free-as-in-speech. But in a hypothetical future in which local municipalities enter into arrangements with future versions of Google or Uber, that access might come at a price. It’s easy to imagine that cross-town transit might soon require acceptance of non-negotiable terms of service that would allow your robocar provider to aggregate and sell where you go, when, with whom, and for what purpose.

Other, stranger realities are possible. Imagine if walking across the street required a microtransaction to insure safe passage. Violations might be subject to tickets or fines—although more likely, your local transit vendor would already know where you are thanks to your smartphone, and just debit your metered service plan accordingly. (Cities like Los Angeles and Honolulu, which impose steep fines for jaywalking, offer a prototype. The car industry has a long history lobbying against jaywalking.) And imagine if Google or Apple, which already provide most folks with mapping services, also got into the business of conveying those citizens through cities. Maps might become accessible only by the good graces of car services. Perhaps you could buy a subscription for unlimited access to your block or neighborhood. The materially fortunate might spring for an all-access plan—the only way to see a whole-city street view.

If this sounds preposterous, consider the fact that tech companies tend to spurn negotiation with local communities if they don’t get their way. Just as Uber and Lyft pulled out of Austin to protest an undesirable regulatory hurdle, Google or Apple might restrict access to their mapping services in areas that don’t adopt political positions convenient to their corporate interests.

They might even elect to alter the physical environment accordingly. In America, street signs are yoked to signage built for human drivers, mounted atop traffic signals and stop signs. Once those devices aren’t needed, their attached markings might also disappear. Perhaps tech giants could persuade municipalities to remove street signs and markers to realize cleaner, less distracting urban conveyance via the synergy of app-street-and-car transit networks.

No matter the scenario, one thing’s for sure: When Silicon Valley runs our automobiles, they’ll do so according to the business practices of the technology industry, not according to the principles of local urban planning and civic life.

*   *   *

The communities most resilient to whatever changes autonomous cars might bring to the public right of way are likely to be those that wean themselves from cars in the meantime. Here, American urban planners and citizens alike can learn from recent trends in Europe, where the relative powerlessness of the automobile has allowed cities to suppress and even reverse its advance. As my colleague Adrienne LaFrance reported last year, Oslo is trying to ban cars entirely from its downtown by 2019. And Barcelona hopes to combat pollution and noise with nine-block “superblocks,” neighborhoods that are bounded by car-accessible streets, but which reserve internal passage for citizen use.

With a few notable exceptions, American cities will never be dense and small enough to adopt European approaches to urban planning. But efforts to make sprawling U.S. cities denser, more walkable, and less reliant on automobiles are more popular than they have been since the car and the suburb took over for the train and the city-center after World War II. Increasing safe access to major roads for bicycles and walkers, for example, might offer an unexpected benefit should the robocarpocalypse arrive: Neighborhoods with deliberately-built, well-maintained paths for non-automotive transit will retain the freedom of entrance and egress that the car currently insures everywhere. And a renewed interest in denser, mixed-use zoning and development has created more opportunities for superblock-like communities in car-centric cities like Atlanta and Houston. Even so, the automobile still rules these towns, and especially in the cheaper, more sprawling suburbs and exurbs where a majority of metropolitan-area inhabitants still reside. Avoiding the roboway entirely might prove impossible.

When imagining a future with self-driving cars, the impact on the immediate future seems most urgent. For the moment, safety and liability issues concern the public most. But even if imperfect, autonomous vehicles are already much safer than human-driven ones. When it comes to robots on the roadways, city dwellers would be better off worrying about their future access to the roads in the first place. Eventually, those citizens might enjoy a new kind of freedom, that of avoiding owning and operating automobiles altogether. But in exchange, they might also give up freedom to move around the city without the help of a few, large global technology companies. Then the suburbs might reverse their fortune and become desirable once more. Not as a place to live, but as a place to escape and move about freely, without the intervention of Big Robocar. That is, if you can still get there in the first place.