Wayve's robotaxis hit the road: London's first commercial service goes live

The robotaxis are real now. In August we covered Wayve and Uber winning London’s first robotaxi licences; at the start of September the service actually went live. From 3 September, Londoners requesting an ordinary UberX, Comfort or Electric ride can be matched — at no extra cost — with a Wayve-powered car driving itself. For Wayve, it is a milestone worth marking: its first commercial public service anywhere in the world, after years of research and testing.

It is a deliberately modest start. The fleet is small, fewer than twenty Ford Mustang Mach-Es, and every car still carries a trained safety driver ready to take over — this is supervised autonomy, not the fully driverless operation that comes later. What makes it notable is how the cars drive. Wayve’s AV2.0 system leans on its end-to-end learned approach: it drives from cameras and sensor data processed in real time, without the high-definition maps or geofencing that most rivals depend on. If that approach scales, it is a cheaper and more flexible path to autonomy than mapping every street in advance — and London, with its density, weather and improvisational driving culture, is about as hard a test as exists.

The competitive timing is pointed. London is now the second European city, after Zagreb, where Uber offers robotaxis, and Waymo has confirmed its own commercial launch in the capital this year. That sets up a genuine contest between two philosophies on the same streets: Waymo’s heavily-mapped, sensor-rich approach against Wayve’s mapless, camera-first one. For a change, the UK is not watching this play out from the sidelines — the home team is a London company.

The caveats are the honest ones. A sub-twenty-car fleet with safety drivers is a pilot, not a transport revolution, and the hard steps are still ahead: expanding the fleet, widening the operating area, and — the real test — taking the safety driver out. But the distance from “licensed” to “carrying paying passengers” is one a lot of autonomy companies never cross, and Wayve has now crossed it in one of the world’s toughest cities. Whether its bet that a single learned model can generalise where hand-built maps cannot holds up is the question the next year will answer.