A Ride in a UK Self-Driving Car – Here’s What Happened

The prospect of experiencing a self-driving car in the heart of London felt like something from a distant future, yet there I was, settling into the passenger seat of a Wayve autonomous vehicle for a real-world test drive. This journey came on the heels of major announcements: Wayve, in partnership with Uber, intends to launch Level 4 fully autonomous robotaxi trials in the capital by 2026. This initiative is a key part of a government strategy to accelerate self-driving pilots, with a potential wider deployment eyed for late 2027. Not to be outdone, Alphabet’s Waymo, already a familiar sight in several US cities, has also declared its ambition to bring a fully driverless robotaxi service to London around the same time, marking one of its first major expansions outside the United States.

My initial doubts about the feasibility of self-driving technology in London were hardly baseless. The city presents a uniquely challenging environment for any autonomous system. Its road network is a historical labyrinth of narrow, winding streets originally designed for horse-drawn carriages, not modern traffic. Navigating this concrete maze is difficult enough, but the real test comes from the constant barrage of obstacles, from potholes and double-parked vans to the relentless flow of buses, cyclists, and unpredictable pedestrians. Add in London’s infamous weather and complex roundabouts, and you have a recipe that could easily overwhelm a robotic driver.

Gaining public trust is another significant hurdle. Brits are notoriously skeptical about new technology, and surveys consistently show a deep-seated wariness toward the idea of AI controlling vehicles. Past hype cycles that ended in disappointment have left a legacy of public distrust that new market entrants must overcome. Then there’s the formidable presence of London’s iconic black cabs. The taxi industry has a history of fierce resistance to disruptors, as seen with the initial arrival of Uber. Interestingly, many cab drivers currently seem to view driverless cars as little more than a novelty or a “fairground ride,” suggesting they may not yet perceive them as a direct threat to their business.

Wayve’s headquarters, nestled in a part of London just a short distance from the tech giants of King’s Cross, felt more like an unassuming warehouse than a hub of futuristic innovation. The company, which began in a Cambridge garage in 2017 and is still run by cofounder Alex Kendall, has nonetheless attracted over $1 billion in funding from powerhouse investors like Nvidia, Microsoft, and SoftBank. The test vehicles themselves were Ford Mustang Mach-Es, looking almost entirely standard apart from a discreet sensor box mounted above the windshield, a far less conspicuous setup than the bulky hardware seen on some competitor models.

Inside the car, the experience felt remarkably ordinary. The most prominent feature was a large red emergency stop button on the center console, a legal requirement ensuring a human safety driver can intervene if necessary. Were it not for a distinct buzzer sounding to indicate the AI had taken control, I might not have realized the driver had relinquished command at all. The vehicle performed capably, merging from quiet side streets onto busier thoroughfares with surprising competence. It navigated around parked cars and delivery trucks, yielded politely to food couriers on electric bikes, and, to my relief, safely avoided pedestrians who treated crosswalks as mere suggestions.

The ride, however, was not perfectly smooth. It lacked the serene, almost unnatural calm I experienced during a Waymo ride in San Francisco. Instead, the Wayve car exhibited a noticeable hesitancy, reminiscent of a newly licensed driver. This cautious approach felt out of place in London, a city where most drivers project an impatient confidence. At one point, we trailed behind a cyclist for what felt like an eternity on a road where any local driver would have confidently overtaken.

I later discovered this tentativeness is an intentional design characteristic. Wayve employs an end-to-end AI model that learns to drive in a generalizable way, much like a human would. Unlike systems that rely heavily on pre-programmed maps and rigid rules, Wayve’s AI is designed to adapt to its environment on the fly. This “embodied AI” approach theoretically means you could place one of its cars in any city, and it would learn to navigate the local driving culture. The team even reported successfully testing the technology on the unpredictable roads of the Scottish Highlands.

This year, the company has been conducting a global AI “roadshow,” testing its vehicles in hundreds of unfamiliar cities across Japan, Europe, and North America. This method suggests that a Wayve car would not need to pass “The Knowledge,” the famously rigorous examination that requires London cabbies to memorize the city’s vast street network. The technology is built to handle the unexpected with a more fluid, human-like response. During my trip, it capably dealt with roadworks, learner drivers, and even a person on crutches who suddenly veered into the street. The most tense moment came when a blind man stepped out between two parked cars; before I could even process the situation, the car had already slowed and adjusted its path smoothly.

By the time we returned to the Wayve compound, I had stopped consciously thinking about who was driving. It was only the sound of the buzzer signaling the safety driver’s return to control that broke the spell. My brain had subconsciously accepted the autonomy. This version of self-driving isn’t the flawless, sci-fi perfection one might imagine. It’s rougher, more cautious, and in many ways, more human. And perhaps, for the chaotic streets of London, that is precisely what is needed.

(Source: The Verge)