Robotaxis Are Moving From Pilots Into City Transport Networks
European cities have spent much of the autonomous-vehicle era watching small demonstrations take place inside carefully selected districts. A shuttle travelled between two predictable stops, a robotaxi operated with a safety driver or a limited fleet served an airport and surrounding business park, allowing manufacturers to prove that the technology could function without forcing transport authorities to decide how thousands of autonomous vehicles should fit into an existing mobility system. During 2026, that distinction has started to erode as commercial operators plan substantially larger European deployments and public transport organisations begin incorporating autonomous services into real networks.
The operational challenge changes with scale. Twenty vehicles can be supervised closely and routed through areas chosen because road conditions suit the technology, whereas thousands of vehicles begin influencing congestion, kerb demand, public-transport ridership and charging infrastructure. Cities consequently need to evaluate robotaxis as a transport mode rather than as a technology trial.
Their strongest urban case may lie outside dense city centres, where conventional public transport often becomes expensive to provide. Buses work efficiently when enough passengers share predictable routes, while lower-density districts generate journeys that are dispersed across time and geography. An autonomous vehicle can offer door-to-door or near-door service without requiring a driver for every individual journey, potentially filling the gap between private cars and fixed-route transit.
That model becomes more useful when transport authorities integrate it with existing services. A robotaxi that competes for passengers along a frequent metro corridor adds vehicles to streets already served efficiently, whereas the same vehicle connecting a suburban neighbourhood to a railway station can increase the usefulness of mass transit by solving the first and last kilometre of the journey.
Pricing will strongly influence which version emerges. If autonomous rides become inexpensive enough to compete with public transport directly, some passengers may shift from buses and trains towards smaller vehicles that occupy considerably more road space per person. Cities could then reduce labour costs in one part of the transport system while increasing congestion elsewhere.
Shared rides provide one response, although consumers have historically shown mixed enthusiasm for sharing vehicles with strangers when a private alternative remains affordable. Operators may need price differences large enough to make pooling attractive, while cities can reinforce the incentive through access rules, kerb charges or priority lanes that favour higher-occupancy vehicles.
Kerb management will become particularly important because autonomous cars do not solve the problem of where passengers enter and leave. A human driver can improvise, circle the block or recognise that a hotel entrance has become temporarily obstructed, while an automated fleet works better when cities provide clearly mapped stopping locations and digital information describing where vehicles may wait.
Large deployments could therefore accelerate the conversion of kerbs into managed infrastructure. Operators may reserve or request stopping capacity dynamically, municipalities can price the most congested locations differently and vehicles can receive real-time instructions before reaching a destination. The same infrastructure could serve taxis, delivery vehicles and accessible transport rather than remaining exclusive to robotaxis.
Public transport authorities also need to decide how much operational control they want. A privately operated robotaxi platform can respond quickly to demand but will naturally optimise around commercially attractive trips, whereas a city may need service in neighbourhoods or at times that generate weaker economics. Contracts can require coverage, accessibility or integration with public fares if autonomous vehicles form part of the official transport network.
Accessibility offers one of the more promising applications provided vehicles are designed for it from the beginning. People who cannot drive because of age or disability could gain substantially more independent mobility, although a vehicle with no human staff needs reliable methods for boarding passengers who require assistance and dealing with unexpected situations once the journey begins.
Remote operators will remain part of that infrastructure even when the vehicle drives itself. An autonomous car may encounter roadworks, police directions or an unusual obstruction it cannot interpret confidently, and a remote support centre can help resolve situations without placing a driver in every vehicle. Scaling fleets therefore changes labour rather than eliminating it entirely, shifting some roles from individual cars into centralised operations.
Charging creates another planning requirement because an electric robotaxi fleet can travel far more kilometres each day than a privately owned EV. Vehicles need to charge without disappearing from service for long periods, which favours high-powered depots and strategically located facilities that can handle concentrated demand.
Cities will also need access to operational data without turning every autonomous journey into a municipal surveillance system. Vehicle counts, empty kilometres, incident rates and stopping patterns can help authorities understand how the fleet affects the network, whereas passenger identities and detailed journey histories require stronger privacy protections.
The arrival of commercial robotaxi fleets will test an assumption that autonomous driving has often encouraged: that better vehicle technology will automatically produce better transport. A city can deploy technically excellent autonomous cars and still worsen congestion if too many travel empty, duplicate public transport or compete for limited kerb space.
Automation becomes more useful when cities decide what role they want the vehicles to perform before fleets become large enough to shape travel behaviour by themselves. The technical challenge has been teaching cars to navigate the city; the urban challenge is deciding which journeys the city should encourage them to make.
