When Right-of-Way Becomes Algorithm: How Autonomous Driving Rewrites the Underlying System of Urban Traffic

In many cities around the world, autonomous driving is no longer a concept car in a showroom. It appears in ride-hailing orders in Phoenix and San Francisco, in driverless taxi fleets on the streets of Wuhan, and on autonomous bus routes in Shenzhen. But the truly notable change is not happening at the vehicle engineering level; it is happening on the city side: roads are beginning to be re-annotated, intersection signals are beginning to be reprogrammed, and regulators are beginning to set rules for a kind of road user that has no driver.

Over the past decade, the industry narrative revolved around sensors, computing power, and algorithms. Over the next decade, what determines whether this technology can move from designated zones to an entire city may well be urban governance capacity itself.

The Bottleneck Has Shifted from the Vehicle to the City

The core question of early autonomous driving was “Can the car understand the world on its own?” As perception and decision-making capabilities stabilize in limited scenarios, the constraints begin to shift outward: whether a stretch of road permits driverless vehicles, whether intersections have roadside perception units, whether edge computing nodes can push blind-spot information to vehicles at millisecond latency, whether HD maps are continuously updated—none of these can be solved unilaterally by automakers.

This also explains why vehicle-road cooperation, roadside perception, and edge computing appear simultaneously in intelligent transportation plans in multiple cities. Cities are no longer merely providing asphalt and lane markings; they are rebuilding a readable infrastructure for machine drivers. The California Department of Motor Vehicles’ tightening and restoration of driverless operation permits, and the National Highway Traffic Safety Administration’s establishment of an automated-driving incident reporting system, both show that regulation is becoming part of the pace of technology.

Why the Public Sector Is Beginning to Step In

Transportation is one of the largest public resource allocation problems in cities. Congestion, accidents, land occupied by parking, and public transit coverage are essentially all questions of how limited right-of-way is divided. Autonomous driving, for the first time, makes real-time adjustment of this allocation logic possible: fleet dispatching, dynamic pricing, demand forecasting, and area-based flow restrictions can all shift from static planning to continuous optimization.

For city managers, the appeal is two-way. On one hand, autonomous feeder services can fill the gap between rail transit termini and low-density communities; such routes have long been difficult to sustain because of high driver costs. On the other hand, the vehicles themselves become mobile data collection terminals, continuously transmitting road conditions, traffic flow, and even air quality back.

Dubai’s Roads and Transport Authority has set the strategic goal that by 2030 a significant share of trips will be handled by autonomous driving, and Singapore has long included autonomous driving in supplementary research for its public transport system; both point to the same judgment: this is not a consumer-electronics-style product, but infrastructure that needs to align with urban public goals.

Data Ownership Is Harder to Discuss Than Technology

When vehicles continuously collect street scenes, pedestrian trajectories, and road geometry information, a question that did not exist before arises: Whom do these data belong to? To the operating company, to the vehicle owner, or to every city it passes through?This question has already surfaced in multiple jurisdictions. Accident attribution requires complete event data records, but companies tend to restrict disclosure on grounds of trade secrets; cities want traffic data for planning but lack enforcement power over private fleets; vehicles operating across borders may also trigger data localization requirements. Through its type-approval framework for autonomous vehicles, the EU is trying to embed compliance requirements before market entry rather than remedy them afterward.

It could be said that autonomous driving has turned “data governance” from an abstract issue into a concrete administrative problem encountered at every intersection.

Streets Will Be Reallocated

If vehicles can be shared, automatically dispatched, and need not wait in city centers, then the demand for urban land for parking will decline over the long term. For many central urban districts, this means a rare opportunity to free up space—parking lots could be converted into housing, public space, or green space.

But the short-term picture may be more complex. Cheaper travel may induce new travel demand—the phenomenon of induced demand that recurs in transportation research. The congestion and enforcement controversies that emerged in San Francisco during the expansion of robotaxis show that friction between technological expansion and street capacity is real. What ultimately determines the outcome is not vehicle capability, but whether cities simultaneously adjust right-of-way, pricing, and public transit priorities.

Three Rollout Paths Are Diverging

Globally, several distinctly different models have emerged for advancing autonomous driving.

The U.S. path is characterized by corporate leadership and passive adaptation by cities. Operators deploy intensively in a few cities, while regulation gradually tightens or adjusts under pressure from accidents and public opinion.

China’s path is more characterized by coordination between city governments and platform companies. Beijing, Shanghai, Shenzhen, and other places have established demonstration zones, bringing road opening, testing permits, and demonstration operations under unified management. This moves faster, but also depends more on sustained local government investment and policy continuity.

The Gulf states’ path is driven by state capital and top-level planning, writing autonomous driving directly into transportation transformation strategies and advancing it through large-scale projects.

The differences among these three paths will ultimately manifest as different data ownership structures, different liability allocation mechanisms, and different urban spatial outcomes.

Long-Term Risks Are Not on the Technology Curve

There are three levels of risk more worthy of attention.

The first is cybersecurity. When roadside units, signal systems, and fleet dispatch platforms are interconnected, the attack surface expands from a single vehicle to an entire road segment and even city-level platforms.

The second is algorithmic fairness. Differences in pedestrian detection, protection of vulnerable road users, and performance under low light and severe weather are not merely engineering metrics; they are fundamentally a matter of allocating public safety resources.

The third is urban stratification. If autonomous driving services prioritize high-value areas while low-density, low-income communities long lack connections, the technology may instead reinforce existing spatial inequality.

JudgmentAutonomous driving will ultimately not be remembered as an “upgrade of the automotive industry,” but more likely as “cities beginning to possess programmable right-of-way.” It requires cities to simultaneously possess three capabilities: the infrastructural capacity to turn roads into a legible environment, the governance capacity to turn data into public assets for negotiation, and the political capacity to make trade-offs between efficiency and equity.

Competition among future cities may no longer be only about whose rail lines are longer or whose ports are busier, but also about whose transportation systems are easier for algorithms to understand and easier for the public to hold accountable.