A driverless vehicle sounds like a transport technology. To a town planner, it is also a land-use question.
If a vehicle can move without a human driver, where does it wait? Where does it pick up passengers? Does it need to park near a destination? Does it circle empty between trips? Does it complement a train or compete with it? Does a shopping street need fewer parking bays and more loading space? Do apartment buildings still need the same parking ratios? Does a depot become more important than a garage? Who gets priority at the kerb when robotaxis, buses, delivery vehicles, cyclists and pedestrians all want the same ten metres of street?
Those are planning questions because technology becomes urban form when repeated at scale.
The American Planning Association’s 2026 foresight work identifies the evolving relationship between autonomous vehicles and public transit as a trend planners should watch. In September 2026, the United States Department of Transportation also released a national automated-vehicle strategy for 2026–2030, while the National Highway Traffic Safety Administration continued work on safety standards and oversight. The technology is therefore no longer a distant thought experiment, but neither is its urban outcome predetermined.
The important planning question is not “When will every car drive itself?” It is “If more vehicles can operate without a human driver, what incentives, street rules and land-use decisions will determine whether towns become easier or harder to live in?”
Automation changes the cost of movement
Much of urban form is shaped by the cost of moving people and goods. Cost does not mean money alone. It includes time, attention, labour, parking, risk, inconvenience and the value of scarce street space.
A conventional private car requires a driver while it moves and a place to store it when the trip ends. A highly automated fleet vehicle could change both conditions. The passenger might use travel time differently. The vehicle might leave rather than park. A commercial operator might reposition it toward expected demand. An unoccupied vehicle might travel between a drop-off point and a cheaper storage or charging location.
That sounds efficient until the system is viewed from the street.
If vehicles drive empty between paid trips, vehicle-kilometres can rise even if passenger-kilometres do not. If automated travel feels easier, people may accept longer journeys or live farther from work. If fleets make door-to-door travel cheaper, some riders may shift from high-capacity public transport. If parking near destinations becomes less necessary, valuable land can be released—but the demand for pick-up, drop-off, staging, cleaning, maintenance and charging can grow.
The technology removes one constraint and exposes another.
The kerb becomes the battlefield
The kerb is a thin strip of land with too many jobs.
It can hold parking, bus stops, taxi stands, freight loading, bicycle parking, outdoor dining, trees, stormwater infrastructure, accessible drop-off space, waste collection, emergency access, micromobility docks and temporary construction zones. Automated mobility adds another claimant: high-frequency passenger pick-up and drop-off.
That means the autonomous street is partly a kerb-management problem.
A town that allows every automated vehicle to stop wherever its passenger wishes could produce friction at precisely the places where activity is highest. A town that creates designated bays can reduce conflict but may require passengers to walk farther. A town that prices kerb access can influence dwell time and discourage unnecessary waiting. A town that reserves space for buses and people with disabilities can protect public goals against the convenience of private fleets.
The critical point is that the kerb cannot be governed by software convenience alone. It is public space.
This connects with TPW-0034 — The Parking Equation. Parking policy is already a way of allocating street and land resources. Automation does not abolish the allocation problem. It changes its form.
A vehicle that does not park still occupies space
It is easy to imagine a future in which autonomous vehicles eliminate parking. That is too simple.
Parking is one form of vehicle storage. A fleet still needs somewhere to wait when demand is low. Vehicles still need cleaning, charging, inspection, repair and software servicing. Some may stage near entertainment districts at night, employment districts before the morning peak, airports during arrival waves or residential areas before school and work trips.
Storage may move rather than disappear.
That creates a planning choice. Should depots be central, reducing empty travel but consuming expensive land? Should they be peripheral, saving land costs while increasing empty kilometres? Can charging be integrated with existing car parks, logistics sites or service areas? Can parking structures be designed for later conversion if demand falls?
The answer will differ between dense centres, suburbs, industrial areas and small towns. The important thing is to plan the storage system explicitly instead of pretending that “driverless” means “spaceless.”
Parking minimums become even harder to justify as fixed truths
Parking requirements have traditionally been written as ratios: so many spaces per dwelling, office floor area, shop or seat. Those ratios already struggle to reflect differences in transit access, household car ownership and mixed-use environments.
Automation adds uncertainty.
If shared automated fleets reduce private car ownership in one district, a new building with a rigid parking requirement may lock expensive floor area into an obsolete use. If automation instead encourages more car travel, removing parking without alternatives could create new pressure on surrounding streets.
A more resilient approach is adaptability. Parking decks can use level floors, adequate ceiling heights and structural grids that allow later conversion. Development rules can set maximums, shared-parking arrangements or district-level strategies rather than assuming one permanent ratio. Mobility plans can be reviewed over time.
The planner is not required to predict the exact future of vehicle ownership. The planner can reduce the cost of being wrong.
Public transport is the hinge
Whether automated vehicles improve or degrade a town depends heavily on what they do to public transport.
A bus or train can move many people through limited street or corridor space. A car, automated or not, still carries a small number of passengers in a relatively large physical footprint. Automation can improve vehicle utilisation, but it does not repeal geometry.
This leads to two very different futures.
In the first, automated vehicles act as feeders. They connect homes to rail stations, serve low-density areas where fixed-route service is difficult, provide mobility for people unable to drive and fill gaps at times when conventional transit is sparse. High-capacity transit remains the spine.
In the second, automated vehicles compete for the easiest and most profitable trips. Riders leave buses and trains, fare revenue weakens, service declines, and the road fills with many small vehicles that individually feel convenient but collectively consume more space.
The technology does not choose between those futures. Policy does.
Street priority, pricing, fleet regulation, transit investment, station design and land use all influence whether automation becomes a connector or a competitor.
The empty-vehicle problem
An unoccupied automated vehicle is still traffic.
This matters because automation can make empty movement operationally easy. A vehicle can reposition toward demand, travel to cheaper parking, return home after dropping someone at work or circle while waiting.
From a passenger’s perspective, that may be convenient. From a network perspective, it can be wasteful.
Planning therefore needs metrics beyond passenger trips. Empty vehicle-kilometres, deadheading share, kerb dwell time, occupancy and road-space consumption become important. A fleet with high paid-trip utilisation can still create congestion if repositioning is excessive.
Pricing can change incentives. Distance-based road charges, congestion pricing, kerb fees or charges for empty travel are possible tools in some jurisdictions. The design challenge is to make the price correspond to the public cost without creating an inaccessible system for people who genuinely benefit from automated mobility.
Automation could stretch the town
One of the deepest planning effects may occur far from the vehicle itself.
If travelling by car requires less attention, the perceived burden of a long commute may fall. A passenger can read, work, rest or communicate while moving. If that makes distant housing more attractive, development pressure can extend outward.
The result could be a paradox. A technology promoted as efficient could encourage more dispersed land use, longer trips and greater infrastructure demand.
This is why transport technology must be examined with land-use policy. Growth boundaries, zoning, infrastructure sequencing and transit-oriented development can either reinforce compact access or allow travel convenience to translate into sprawl.
The relevant comparison is not between an automated car and a human-driven car on the same trip. It is between two different towns that may emerge after millions of households change where they are willing to live.
The station area has to be protected from its feeder
Stations are natural places for automated pick-up and drop-off. They are also places where pedestrians, buses, cyclists and taxis already converge.
If every feeder vehicle tries to reach the station entrance, the final hundred metres can become the most congested part of the journey.
Good station planning therefore separates functions. High-capacity modes receive the strongest spatial priority. Pedestrian approaches remain direct and safe. Pick-up zones may be placed slightly away from the entrance. Access routes can be managed so vehicles do not cross the heaviest pedestrian flows. Waiting vehicles can be held remotely and called forward only when needed.
The goal is not to maximise vehicle convenience at the station. It is to maximise station throughput and human access.
That principle extends TPW-0015 — Transit-Oriented Development. The station area should be organised around access to high-capacity transport, not surrendered to the feeder modes that arrive there.
First responders reveal the difference between a demo and a city
Urban systems are tested by exceptions.
A vehicle may navigate ordinary traffic impressively and still create problems when a fire engine blocks a junction, police redirect traffic, a signal is dark, a street is flooded, a road worker gives hand instructions or an ambulance needs an improvised path.
In July 2026, the U.S. National Highway Traffic Safety Administration publicly called on automated-vehicle developers to address interactions with first responders after identifying a pattern of incidents. That is a useful planning reminder: a street is not merely a machine-readable lane network. It is an emergency workspace.
Planning for automated streets therefore includes incident-management protocols, pull-over behaviour, emergency geofencing, communication standards and physical designs that preserve access when conditions depart from normal operation.
The edge case is part of the city.
Street design cannot assume perfect machines
It is tempting to redesign streets around the promised precision of automated driving: narrower lanes, smaller gaps, higher throughput.
Some efficiency gains may eventually be possible, but planning should resist premature dependence on perfect performance.
For many years, towns are likely to contain a mixed fleet: pedestrians, bicycles, motorcycles, human-driven cars, partially automated vehicles, highly automated vehicles, buses, freight vehicles, maintenance crews and emergency responders. Weather, construction, temporary events and unusual behaviour will remain.
A street designed only for an idealised automated fleet can become fragile during the transition.
Robust design should therefore prioritise legibility, safe speeds, forgiving geometry and clear allocation of space. Automation should adapt to the town’s safety objectives, not require the town to become a laboratory corridor.
Pedestrians must remain first-class road users
Automated driving systems depend on prediction. Pedestrians are difficult because people do not always behave like vehicles.
A child may run back for a ball. An older person may pause midway. A tourist may look the wrong way. A cyclist may avoid a pothole. Two people may negotiate priority through eye contact that an automated system cannot reproduce in the same way.
The wrong response is to make human movement more rigid so machines have an easier environment.
If streets respond to automation by fencing pedestrians, lengthening crossings or criminalising ordinary informal movement, technical convenience has displaced public purpose.
The safer principle is that automated systems operating in public space must be able to accommodate the ordinary variability of human life within reasonable rules.
Accessibility is a major opportunity
Automation could materially improve mobility for some people who cannot drive or find conventional transport difficult.
That opportunity is not automatic.
A vehicle that can drive itself may still be inaccessible to a wheelchair user. A booking app may exclude people without smartphones. A pick-up zone may be too far from a building entrance. Audio, visual and tactile interfaces may not work for all users. Pricing may place the service beyond the reach of people who would benefit most.
Town planning therefore needs accessible kerbs, adequate boarding space, safe paths between buildings and pick-up points, and integration with paratransit and public transport. Regulation needs to define service obligations rather than assuming the market will discover universal access on its own.
The vehicle can be autonomous and the journey can still be inaccessible.
Freight may automate before private travel does
Automation is not only about robotaxis.
Freight corridors, distribution centres, ports, depots and last-mile systems can adopt forms of automation under more controlled conditions. Driverless yard vehicles, automated highway segments, delivery robots and highly automated trucks create different planning pressures.
Distribution facilities may value direct access to suitable corridors and charging infrastructure. Last-mile transfer points may become more important. Residential streets may need rules for small delivery devices. Loading bays may require digital reservation systems. Industrial land may need to accommodate new service and energy demands.
This is where The Logistics Layer meets the autonomous street. Goods movement is a land-use system as much as a transport service.
Energy becomes part of the mobility map
Many automated fleets are expected to be electric, though automation and electrification are technically separate.
If large fleets charge in concentrated locations, electricity demand becomes spatial. A depot may require major grid capacity. Charging during peak periods can stress local infrastructure. Vehicles may need to rotate through chargers in ways that affect fleet availability and land requirements.
Planning therefore cannot treat charging as an afterthought. Grid capacity, substation space, depot location, fire safety, drainage, queuing and access all enter the site-selection problem.
This is another example of the recurring rule of town planning: when a new system scales, its invisible infrastructure eventually becomes visible.
The data bargain will matter
Automated fleets can generate extraordinary amounts of operational data: routes, speeds, stops, conflicts, near misses, demand patterns and road conditions.
Some of that data could improve planning. It could identify dangerous junctions, demand peaks or poor kerb design. But privately operated systems may consider data commercially sensitive. Fine-grained trip records also create privacy risks.
Towns therefore need a data bargain: enough reporting to protect public safety and manage public streets, with clear rules for aggregation, privacy, retention and commercial confidentiality.
The new TPW-0045 — The Data Gap is relevant here. A town should not become dependent on a mobility dataset it cannot audit or may suddenly lose.
Do not confuse vehicle safety with system safety
A vehicle can satisfy safety standards and still participate in an urban system that performs badly.
Suppose automated vehicles reduce crashes per kilometre but increase total kilometres dramatically. Suppose they improve mobility for affluent travellers while weakening transit used by everyone else. Suppose they reduce parking demand in the centre but create large peripheral depots and more empty travel. Suppose they make long commutes easier and accelerate sprawl.
None of those outcomes can be judged by vehicle engineering alone.
System safety includes road deaths and injuries, but town planning adds congestion, emissions, accessibility, land consumption, fiscal cost, social inclusion and resilience.
The correct unit of analysis is not only the vehicle. It is the town.
Geofencing will create invisible boundaries
Highly automated systems often operate within defined operational domains: particular roads, speeds, weather conditions or mapped areas.
For planners, that means service geography can be encoded in software.
A fleet may serve the airport but not an outer neighbourhood. It may operate in a wealthy district with clear roads but avoid complex informal streets. It may stop service during heavy rain precisely when some users need it most.
These boundaries can become a new form of accessibility inequality if left entirely to operator economics or technical convenience.
Public authorities may therefore need to monitor service coverage, minimum obligations and exclusion patterns. A digital boundary can matter as much as a physical one when it decides who receives mobility.
Pricing is planning
The price of automated travel will shape the town.
If empty movement is free to the operator, there is little incentive to minimise it. If kerb occupancy is free, vehicles may wait in the most valuable places. If road capacity is free at the peak, demand can expand until congestion returns. If transit fares rise while robotaxi prices fall, mode share may shift even when the collective result is inefficient.
Pricing tools can align private decisions with public costs, but they must be designed carefully. A blunt charge can burden people with few alternatives. A targeted charge can distinguish congested places and times, vehicle occupancy, empty running or premium kerb access.
Planning and pricing are therefore partners. The map decides where activity may occur; the price influences how intensely scarce capacity is consumed.
The town should decide what success means before deployment scales
A technology pilot often measures technical performance: disengagements, collisions, response time, service uptime.
A town needs a broader scorecard.
- Did total vehicle-kilometres rise or fall?
- How much travel occurred empty?
- Did bus and rail ridership change?
- Did road injuries change for pedestrians and cyclists?
- Did accessible mobility improve?
- Did kerb conflicts increase?
- Did households reduce car ownership?
- Did parking land become available for better uses?
- Did development spread farther from jobs and services?
- Did low-income neighbourhoods receive comparable service?
- Did emergency response become easier or harder?
- What new infrastructure costs did the public sector absorb?
If success is defined only by the number of automated rides, the town has handed the objective function to the operator.
Pilot the rules, not only the vehicles
Because the technology and behaviour are uncertain, towns can use pilots to test governance.
A pilot district can test designated pick-up zones. A station can test remote holding areas. A city can compare kerb-pricing methods. A fleet can be required to share defined safety and operations data. A service area can include coverage obligations. A temporary rule can limit empty circulation during peak periods.
The point is to learn not just whether vehicles can operate, but which public rules produce acceptable urban outcomes.
That is an important distinction. Technical pilots answer “Can this run?” Planning pilots answer “Under what conditions should this run here?”
Design for a mixed transition
The most plausible planning horizon is not a sudden switch from human driving to full automation. It is a mixed period.
During that transition, infrastructure should avoid unnecessary technological lock-in.
Road markings should remain useful to people. Signage should remain human-readable. Pick-up zones should be adaptable to taxis, community transport and deliveries. Parking structures should be convertible. Communications systems should avoid dependence on one vendor. Street geometry should remain safe even if automation underperforms.
Flexible infrastructure is a hedge against uncertain adoption.
A planning checklist for automated mobility
- Define the public objective. Safety, accessibility, emissions, reduced parking, transit support or some combination?
- Protect high-capacity transport. Decide where buses and rail receive priority.
- Map kerb demand. Identify conflicts among passengers, freight, accessibility, trees, cycling and emergency use.
- Measure empty running. Do not treat unoccupied kilometres as invisible.
- Plan depots and charging. Include grid capacity and service access.
- Protect pedestrians. Do not redesign human behaviour merely to simplify machine prediction.
- Set accessibility requirements. Include vehicles, apps, boarding areas and fares.
- Plan for incidents. Coordinate with police, fire, ambulance and road-maintenance agencies.
- Require useful data. Define reporting, privacy and retention rules before dependence develops.
- Review land-use effects. Monitor whether travel convenience encourages outward growth.
- Keep assets adaptable. Design parking and kerb infrastructure for conversion.
- Use pilots deliberately. Test governance as well as technology.
What should not change
Every technology cycle creates a temptation to rewrite the city around the technology.
But some planning objectives are more durable than any mobility platform.
People should be able to reach jobs, schools, healthcare, shops, parks and one another safely. Streets should support public life as well as movement. Children, older adults and people with disabilities should not be designed out. High-capacity corridors should move large numbers efficiently. Housing should not be pushed ever farther from opportunity simply because travel has become easier for some.
Automation should be evaluated against those goals.
The town is not an accessory to the vehicle.
The autonomous street is really a governance test
The fascinating part of automated mobility is the machine. The consequential part is the system around it.
A driverless vehicle can be exquisitely engineered and still produce a poor town if incentives reward empty circulation, public transport weakens, kerbs become chaotic and development stretches outward. The same technology can be useful if it feeds transit, expands accessible mobility, reduces parking land and operates under street rules that price scarce capacity intelligently.
That is why town planning matters before the technology becomes ordinary.
Once millions of daily trips, property decisions and infrastructure investments adapt to a mobility system, the pattern becomes difficult to unwind.
The autonomous street should therefore be planned as a public system with private machines inside it—not as a private technology that the public street must learn to tolerate.
Three transition scenarios planners should test
Because no town knows exactly how automated mobility will scale, scenario planning is more useful than a single forecast. At minimum, three distinct pathways should be tested because they produce very different land-use outcomes.
Scenario one: privately owned automation. Households replace conventional cars with highly automated private vehicles. Parking at home may remain important, but destination parking can decline if vehicles return home or reposition elsewhere. Empty vehicle travel may rise sharply. Longer commutes may feel easier, increasing pressure for outward growth.
Scenario two: shared automated fleets. Vehicle ownership falls and on-demand fleets become common. Household parking demand may decline, but kerb demand, depots, charging sites and fleet-management facilities grow. The decisive question becomes whether shared vehicles feed high-capacity transit or substitute for it.
Scenario three: freight-first automation. Passenger travel changes slowly while logistics automates quickly. Industrial land, distribution centres, highway interfaces, charging infrastructure and last-mile transfer points experience the strongest pressure. Residential street design changes later.
A robust plan should not depend on only one of these futures. The useful investments are those that perform reasonably across several: adaptable parking structures, well-managed kerbs, strong public transport, connected street networks and flexible depot sites.
Airports, hospitals and event districts are useful stress tests
Some urban places compress enormous demand into short periods. Airports produce arrival waves, luggage and waiting. Hospitals require emergency access, patient drop-off, shift changes and accessible boarding. Stadiums and event districts release thousands of people at once.
These locations reveal whether an automated-mobility plan actually understands capacity. A fleet that works beautifully in ordinary dispersed demand can fail when hundreds of vehicles are summoned to the same kerb within minutes. Queue spillback can block buses, emergency vehicles and pedestrian crossings. Empty vehicles searching for passengers can make the peak worse.
The planning response is spatial and operational: remote holding areas, geofenced pickup zones, timed access, pedestrian priority, transit staging, dynamic kerb allocation and clear emergency routes. The lesson then transfers to town centres, schools and entertainment streets.
Cyber failure becomes a local land-use dependency
Automated mobility also depends on software, communications, maps, positioning, fleet servers and charging systems. A cyber incident or network outage can therefore become a physical street problem.
Town planners do not need to become cybersecurity engineers, but they should ask continuity questions. Where can vehicles fail safely? Can a depot operate manually? Can emergency services override normal access rules? Do critical facilities have alternative transport? Can streets function if digital kerb reservations disappear for several hours?
This matters because infrastructure becomes fragile when its digital layer has no physical fallback. The public realm should remain understandable and usable when automation is degraded. Human-readable signs, safe stopping areas and conventional emergency procedures are redundancy, not nostalgia.
Land value will move before buildings do
If automation changes perceived travel time, parking demand or station access, land markets may respond before physical redevelopment occurs. Sites near high-quality pickup zones or fleet depots can gain value. Large parking lots may acquire redevelopment potential. Remote housing may become more attractive if commuting feels less burdensome.
Planners should therefore watch property signals as well as vehicle statistics. A transport technology can quietly alter development pressure long before annual traffic counts show the full effect. Monitoring permits, land transactions, parking conversions and housing growth can reveal whether the mobility system is stretching or compacting the town.
The deepest planning question remains the same: does the technology make useful destinations easier to reach while consuming less public space and creating fewer external costs, or does it merely make more vehicle movement easier?