I never truly believed that self-driving vehicles would reach the UK, so you can imagine my astonishment when I found myself stepping into one of Wayve’s autonomous cars for a trip around north London a few weeks back.
In June, the firm revealed schemes with Uber to initiate trials of Level 4 fully autonomous robotaxis in the capital starting as soon as 2026, as part of a government initiative to expedite self-driving tests ahead of a possible broader implementation in late 2027. Waymo, owned by Alphabet and firmly established in US cities such as San Francisco, Los Angeles, and Phoenix, also has its sights on London, disclosing plans for its own completely driverless robotaxi service in 2026, marking one of its initial attempts to expand outside of the US.
My doubts regarding whether self-driving cars can succeed in London aren’t without merit. On many fronts, London stands as a robotaxi’s ultimate challenge. The city’s roadways are narrow, convoluted, and notorious for their navigation difficulties, a labyrinth of concrete forged over centuries, intended for horses and carts, not motor vehicles. Constrained streets complicate the avoidance of barriers — potholes, parked cars, you name it — and that’s prior to considering the overwhelming influx of drivers, pedestrians, tourists, cyclists, buses, taxi services, and animals (like wayward military horses) sharing the thoroughfare. And it’s probably wise to avoid discussing roundabouts or inclement weather.
Even if a robotaxi manages to navigate London without a hitch, it also requires the acceptance of Londoners regarding the technology. This could prove difficult. We tend to be quite skeptical and when it comes to AI in vehicles; polls indicate that Brits rank among the world’s most doubtful. There has also been significant hype — and failures — related to this technology in the past, leaving a legacy of skepticism and disbelief that newcomers must overcome. Additionally, the iconic black taxis present another hurdle, as they are known for fiercely defending their turf. When Uber first emerged, taxi drivers frequently caused London traffic to grind to a halt, and that industry is still battling the ridesharing service today. Nevertheless, they don’t seem particularly threatened this time, ridiculing driverless vehicles as “a funfair ride” and “a tourist attraction in San Francisco.”
Wayve’s headquarters didn’t resemble a tourist hotspot from San Francisco. The austere combination of exposed brick and black metal fencing gives Wayve, which began in a garage in Cambridge in 2017 and is still guided by cofounder Alex Kendall, the atmosphere of an unremarkable warehouse. A mere 15 minutes away lies King’s Cross, a once-derelict industrial area that’s now inhabited by firms like Google and Meta, a more typical environment for a company that has garnered over $1 billion from giants like Nvidia, Microsoft, and SoftBank (and is reportedly seeking to raise up to an additional $2 billion).
The cars, a collection of Ford Mustang Mach-Es, didn’t appear particularly futuristic either. The main indication of their intention to replace human operators was a compact sensor box affixed above the windshield, a stark contrast to the prominent protrusions seen on Waymo’s vehicles.
Inside, it felt equally mundane. As we departed Wayve’s grounds, the only feature that stood out was the large red emergency stop button situated in the center console, a reminder that, in legal terms, a human operator must be prepared to take over at any moment. If not for the piercing buzz indicating that the robotaxi had assumed control, I doubt I would have perceived that the driver relinquished any control at all.
It managed the city’s complexities surprisingly well — much better than I anticipated. Within minutes, we transitioned from the quiet side streets near Wayve’s base onto a busier arterial road. The vehicle navigated seamlessly past parked cars and delivery vans, decelerated courteously when food delivery cyclists cut in front of us on electric bikes, and thankfully avoided colliding with any pedestrians who utilized London’s crosswalks more like guidelines than regulations.
The ride wasn’t exactly smooth, though, and it was far from the serene experience I had during my initial Waymo trip in San Francisco this summer. Wayve exhibited a more cautious demeanor than I am accustomed to, reminiscent of my sister’s driving on our initial outings when she received her license a few years back.
That caution is particularly peculiar in London. Friends, taxi drivers, bus operators, and Uber drivers I’ve accompanied all seem to exude an inherent urgency and unyielding confidence that Wayve appeared to lack. I haven’t driven since I earned my license 15 years ago — the Tube makes living in London quite straightforward without a car — yet its hesitance constantly tested my patience. Our path led us past the formidable walls of Pentonville Prison in Islington, and we lagged behind a cyclist whom I was confident I could easily overtake, a maneuver any London driver certainly would have executed.
I later discovered that this caution is an intentional design choice, not a flaw. In contrast to Waymo — which employs a blend of detailed mapping, regulations, sensors, and AI for driving — Wayve utilizes an end-to-end AI model that enables it to drive in a more generalized manner. To put it differently, Wayve operates more like a human and less like a machine. It certainly felt that way; I found myself glancing at the safety driver’s hands, half-expecting to see them regain control. They never did. Other road users seemed to share this belief. A police officer even raised his hand in gratitude as we allowed him a gap to turn into a gas station, though perhaps that gesture was directed at the safety driver.
In theory, this embodied AI strategy implies that you could position a Wayve vehicle anywhere, and it would effortlessly acclimate, akin to how a human driver might when maneuvering through an unknown city. While I’m not entirely ready to put that to the test personally, the team mentioned they’ve recently driven through the Scottish Highlands without any issues.
I later learned that this company, which is aiming for markets in Japan, Europe, and North America, has been traveling worldwide on an AI “roadshow” this year to test its technology in 500 unfamiliar locations. With this understanding, it appears Wayve may not need to take The Knowledge, a rigorous series of tests required of London taxi drivers to demonstrate their memorization of countless streets and locations, allowing them to navigate without GPS (it’s also led scientists to rave about their minds).
This strategy means the technology is additionally developed to respond to the environment in a more fluid manner and react more humanely to unpredictable situations and edge cases that deeply concern autonomous vehicle developers. Throughout my journey, it did precisely that. From roadworks, learner drivers, groups of cyclists, and London buses to a person on crutches edging into the street, it managed each encounter effectively, albeit more slowly than a typical London driver might. The most intense moment occurred when a blind man stepped out with his cane between two parked cars — a scene so fitting that I felt compelled to ask the company if it had been orchestrated (it wasn’t) — but before I could respond, the vehicle had already decelerated and changed its path.
As we returned to Wayve’s facility, I realized I had ceased to contemplate who was driving. It was merely the recurring sound of the blaring buzzer that indicated our safety driver had regained control. My mind, evidently, has come to terms with autonomy, at least in its London interpretation. It’s rougher, less polished, more human. And perhaps that’s the essence of it.
