The Four-Foot Dividend: How Autonomous Vehicles Could Redraw the Width of Every Street

Here is a number that shapes every city on earth and that almost nobody thinks about: twelve feet.

That is the standard width of a traffic lane in the United States. Twelve feet, measured from stripe to stripe, repeated millions of times across millions of miles of asphalt, in every city and suburb and highway interchange from Maine to Hawaii. It is one of the most consequential dimensions in the built environment, and it was calibrated for one very specific reason.

Humans are bad at driving.

The Geometry of Human Error

The modern traffic lane did not arrive at twelve feet through elegant calculation. It arrived there through a century of negotiation between human imprecision and acceptable casualty rates.

NACTO – the National Association of City Transportation Officials – publishes the most widely referenced guidelines for urban street design in North America. Their Urban Street Design Guide recommends lane widths of 10 feet for urban streets and allows up to 12 feet for highways and arterials. The wider the lane, the faster drivers feel comfortable going. The narrower the lane, the more drivers slow down – an effect that traffic engineers call the “friction” of the street environment.

But even NACTO’s 10-foot urban lane is designed around a fundamental assumption: the driver will drift. Human hands on a steering wheel produce constant micro-corrections. We oversteer, we undercorrect, we glance at a phone or a passenger or a storefront and wander six inches toward the centerline. The lane must be wide enough to contain this wandering with a margin of safety on both sides. Add the width of the vehicle itself (roughly 6 to 7 feet for a standard car), and the remaining 3 to 5 feet per lane is essentially a buffer for human incompetence.

Turning radii follow the same logic. The sweeping curves at intersections – those broad arcs of curb that push pedestrians back from the crosswalk and create vast, empty aprons of asphalt – exist because a human driver turning right at 15 miles per hour cannot execute a tight radius without clipping the curb or swinging into oncoming traffic. The radius is engineered for the worst-case human performance, not the best.

Sight triangles, those invisible cones of required visibility at every intersection, are calculated based on human reaction time – roughly 2.5 seconds for an alert driver to perceive a hazard, decide to brake, and initiate braking. The street must be clear of visual obstructions for the distance a vehicle covers in those 2.5 seconds. At 30 miles per hour, that is 110 feet of required clear sight line in every direction.

Every one of these dimensions – lane width, turning radius, sight triangle – is a spatial footprint of human limitation. And collectively, they define the visual signature of every city on the planet.

Remove the Human, Change the Math

An autonomous vehicle does not drift. Its steering corrections are measured in fractions of a degree, executed hundreds of times per second by actuators that do not get tired or distracted. The buffer for wandering shrinks from feet to inches.

Research from MIT’s City Science group, part of the Media Lab, has modeled autonomous vehicle lane requirements and concluded that dedicated AV lanes could safely operate at widths of 8 feet or less – a reduction of 2 to 4 feet per lane compared to current standards. This is not speculation about distant technology. It is a geometric consequence of removing the one variable that all current lane widths are designed to accommodate: the human margin of error.

Turning radii contract similarly. A vehicle that can calculate its exact trajectory through an intersection does not need the generous curves that accommodate human misjudgment. Tighter radii mean smaller intersections. Smaller intersections mean shorter pedestrian crossings. Shorter crossings mean less time spent exposed in traffic.

Sight triangles become largely irrelevant for vehicle-to-vehicle interactions when both vehicles are networked and sensor-equipped. A self-driving car does not need to “see” around a corner in the way a human driver does – it already knows what is there, communicated by the other vehicle’s transponder or the intersection’s infrastructure sensors.

The compounding effect of these reductions is enormous. In a city like Houston, where the street right-of-way can consume 35 to 40 percent of the total urban land area, shrinking every lane by even two feet would release thousands of acres of space – space that is currently asphalt, currently hostile to every use except moving and storing cars.

How Cars Colonized the Street

To understand what might be reclaimed, it helps to understand what was taken.

Peter Norton’s 2008 book Fighting Traffic: The Dawn of the Motor Age in the American City documents a transformation that happened so completely that most people alive today cannot imagine it was ever otherwise. Before the 1920s, streets were shared public spaces. Children played in them. Vendors sold from them. Pedestrians walked wherever they pleased. The street belonged to everyone.

The automobile industry, working through lobbying groups, insurance companies, and compliant city governments, systematically redefined the street as a space exclusively for cars. The concept of “jaywalking” was invented – Norton traces the term to auto industry campaigns in the 1920s that sought to stigmatize pedestrians who used the roadway. Streets were widened. Sidewalks were narrowed. Setback requirements pushed buildings away from the curb. Parking lanes consumed what had been public gathering space.

The twelve-foot lane is the mature expression of this century-long colonization. Every foot of that width was won by the automobile at the expense of every other possible use of the public right-of-way.

The Reclamation Is Already Beginning

You do not have to wait for full autonomy to see the counter-movement. It is already underway.

Culdesac, the car-free neighborhood in Tempe, Arizona, opened its first phase in 2022 as the first purpose-built car-free development in the United States. No private cars, no parking lots, no garages. The streets are designed for people, bicycles, and small electric shuttles. The result is a development where the street right-of-way is roughly half what a conventional subdivision requires – and the reclaimed space is parks, courtyards, and commercial frontage.

Sidewalk Labs, Alphabet’s urban development subsidiary, produced detailed proposals for a Toronto waterfront district (Quayside) that anticipated autonomous vehicle integration from the ground up. Their plans showed streets that could dynamically reallocate space – lanes that serve vehicles during peak hours and become pedestrian plazas during evenings and weekends. The project was ultimately cancelled, but its design thinking – that street dimensions are not fixed physical constants but variables that can change with technology – has influenced urban planners worldwide.

Barcelona’s superblocks program, while not AV-dependent, demonstrates the principle at municipal scale. By restricting through-traffic in clusters of nine blocks, the city has reclaimed intersection space for playgrounds, greenery, and outdoor dining. Research published in Environment International in 2020 estimated that full implementation of the superblocks plan could prevent nearly 700 premature deaths annually through reduced air pollution and increased physical activity.

The Grid Stays. The Proportions Change.

Here is what matters for anyone who thinks about cities as visual systems: the autonomous vehicle revolution, if it comes, will not redraw the street grid. The grid is infrastructure that persists for centuries. The streets of Manhattan will still run north-south and east-west. The boulevards of Paris will still radiate from their Haussmann-era nodes. The chaotic organic fabric of Tokyo will still wind through its ancient river valleys.

But the proportions within that grid will shift. Lanes will narrow. Sidewalks will widen. Intersections will tighten. The ratio of space given to vehicles versus space given to humans will rebalance for the first time in a hundred years. And that rebalancing will change the visual texture of cities in ways that will be visible from above – in satellite imagery, in mapping data, and in the minimalist street network renderings that we produce at streetgrid.art.

A street map of a city in 2045 may have the same geometry as a street map of that city in 2025. The same lines, the same intersections, the same blocks. But the weight of those lines – the visual thickness that represents the width of the street – may be measurably thinner. The grid will be the same. The space between the lines will be larger. And in that extra space, that four-foot dividend multiplied across every lane of every street, something other than asphalt will finally have room to grow.