
이미지: AI 생성 — METAL LAB
Summary
- Uber and British startup Wayve launched a self-driving ride-hailing service in London on September 3, local time — the first of its kind in the UK.
- Wayve's AV2.0 drives without lidar or HD maps, so it launched running across all of London except airports from day one.
- Uber plans to bring Wayve vehicles to 12 markets worldwide, and named Tokyo as the next city.
Uber and British autonomous driving startup Wayve launched a self-driving ride-hailing service in London on September 3, local time. It's the first time a self-driving vehicle has carried paying passengers in the UK.
Riders who book UberX, Uber Electric, or Uber Comfort in London may now be matched with a Wayve vehicle. There's no extra charge, and the fare is shown up front in the app.
Running across London, except airports
The vehicles being dispatched are electric Ford Mustang Mach-E models fitted with Wayve's "AI Driver" and surrounding sensors.
The service area covers all of London except airports. This isn't a pilot confined to a small zone — it's launching city-wide from the start.
The current phase isn't fully driverless. A trained safety driver licensed by Transport for London (TfL) rides along to supervise. Uber says it plans to move to fully driverless operation once regulators approve it.
The fleet is starting small. Uber says it will scale up gradually based on demand, readiness, and regulatory progress. More than 140,000 London users have already opted in through app settings to be matched with self-driving vehicles.
Uber requires that every autonomous vehicle on its network meet the company's own safety guidelines. Wayve's vehicles cleared that bar before being added to the dispatch pool.

A driver that doesn't memorize the map
What sets Wayve apart from other self-driving companies is its approach to maps. Wayve doesn't use lidar or pre-surveyed HD maps at all.
The company calls this approach "AV2.0." Instead of hand-coded rules, it relies on an AI Driver that learns from experience, the way a human learns to drive, and lets the AI make judgment calls in real time.
Think of the difference as two ways of preparing for an exam. The HD-map approach is like a student who has memorized the entire syllabus — strong within the material covered, but stuck the moment a question falls outside it.
AV2.0 is more like a student who understands the underlying principles. It can figure out how to navigate an unfamiliar side street on its own. And because it doesn't need to send survey vehicles ahead of time to scan a new city, it can expand faster.
Wayve was founded in London in 2017 and has been training its AI on London roads since 2018. Wayve CEO Alex Kendall has described London as "one of the most complex driving environments in the world."
Same app, different car
Uber designed the passenger experience. Inside the vehicle, an interactive screen supporting 64 languages lets riders start the trip and follow the route the car intends to take.
Opening the door on arrival and starting the ride both happen through the Uber app — no separate app to install.
Riders also keep a choice at the dispatch stage: before the car arrives, they can decide whether to keep the self-driving vehicle or switch to a human-driven one.
What Uber is really after
The London launch is the first result of Uber and Wayve's partnership. As part of its investment in Wayve, Uber agreed to bring Wayve vehicles to 12 markets worldwide, with London as the starting point.
Uber has named Tokyo as the next stop, with plans to deploy Nissan Leaf vehicles running Wayve's AI Driver on NVIDIA DRIVE Hyperion later this year.
The switch in vehicle model isn't an issue because Wayve's software isn't tied to any specific vehicle platform or sensor setup. Moving from the Mustang Mach-E to the Nissan Leaf doesn't require rebuilding the system from scratch — and that's part of why the company is confident it can scale quickly.

Uber's broader plan centers on a "hybrid network" — putting self-driving vehicles and human drivers side by side on the same network, so one covers gaps left by the other during off hours or in underserved areas.
The scale is notable. Uber says it now works with more than 30 autonomous driving partners across mobility, delivery, and freight, and plans to broker self-driving rides in 15 cities by the end of 2026. Its stated goal is to become the world's largest broker of autonomous rides by 2029.
Editor's take
What stands out here isn't that you can now hail a self-driving Uber in London. It's that the service launched city-wide without lidar or HD maps.
Anyone who's watched enough self-driving demos knows the pattern: remarkably smooth within a defined zone and along a fixed route, then it falls apart the moment it steps outside that boundary. Expanding that boundary has always been the industry's real cost problem.
Sending out survey vehicles to scan a city and updating maps every time a road changes means paying that setup cost all over again for every new city. Wayve being able to launch across all of London except airports, right out of the gate, comes down to it having sidestepped that cost structure.
It's worth noting that the same shift is playing out across AI more broadly right now. Learning is replacing places where humans once had to hand-code rules, and the moment that happens, expansion costs drop sharply. Self-driving is simply the field where that shift is showing up first in the physical world.
Two things are worth watching for teams following this from Korea. One is the position Uber has chosen. Rather than build self-driving technology itself, Uber has positioned itself as the distribution layer, plugging in multiple partners. As the technology race intensifies, whoever controls demand comes out ahead.
The other is that Wayve emerged with full backing from the UK government. It's unusual to see a company announcement accompanied by joint statements from both the Transport Secretary and the Business Secretary. That's a signal the UK wants to claim self-driving as an industrial policy win, and it's likely to show up in how fast regulation moves too.
London has become the first case of self-driving moving out of pilot zones and into a city's actual transit network. Wayve's AI Driver is now being tested on real roads, carrying paying passengers.





Comments