Wayve’s CEO talks teaching self-driving cars to handle US roads
Wayve announced in October that it would begin testing its driver-assist technology in California.
Europe’s buzzy self-driving startup is about to face a new challenge — driving on the opposite side of the road.
London-based Wayve has ambitious plans for global expansion after receiving backing from Microsoft, Nvidia, and SoftBank.
The company announced in October that it would start testing its advanced driver assist technology in California, joining Waymo and Amazon-backed Zoox in driving on public roads in the US.
That will see its vehicles face a similar headache to human drivers traveling across the Atlantic — switching from UK roads, where cars drive on the left rather than the right.
“There’s going to be some new challenges, whether it’s driving on the right side of the road, four-way stop signs, or right turn on a red. These are things that we don’t have in the UK,” Alex Kendall, Wayve’s CEO, told B-17 in an interview.
Wayve’s self-driving software is trained on the streets of London and other UK cities, known for being narrower and more crowded than the wide-open roads of California.
Wayve CEO Alex Kendall said the company’s vehicles will face “new challenges” in the US.
But the startup, which recently raised over $1 billion from investors including Microsoft and Nvidia, is confident that it can make the leap.
Wayve’s fleet of Ford Mach-Es uses an end-to-end AI model that learns how to drive from real-world testing and simulations.
That’s a different approach from companies like Waymo, which rely on radar systems like lidar and high-precision mapping to limit their vehicles to operating in certain areas.
Tesla has shunned lidar to rely exclusively on cameras and AI for its driver-assist systems, with Elon Musk previously calling lidar a “crutch.” Tesla still spent at least $2 million on the technology in the first quarter of 2024, though it is unclear what it was used for.
Kendall said Wayve’s approach allowed the company’s software to generalize and adjust to new rules of the road in much the same way a human would.
He told B-17 that Wayve was already seeing its vehicles learn to read four-way stop signs, weeks after it started testing a fleet of Ford Mach-Es with safety drivers in California.
“That level of behavior took us significantly more time and effort to learn in the UK. So what that shows to us is that we are generalizing,” said Kendall.
Big year for Wayve
The expansion to the US caps a big year for Wayve, whose AI-first approach to autonomous vehicles has caught the eye of everyone from Meta chief AI scientist Yann LeCun to Bill Gates, who praised the company last year after taking one of Wayve’s vehicles for a spin.
The London-based startup has opened new offices in Vancouver and San Francisco, with Kendall saying the latter was partly to be closer to partners such as Microsoft and Nvidia.
Wayve has also struck a partnership with Uber, which Kendall described as a three-way partnership that will see the two companies work with automakers to equip consumer vehicles with Wayve’s driver-assist technology before gradually scaling up to fully autonomous robotaxis that will be deployed on Uber’s platform.
The Wayve CEO said gaining access to data from a network of millions of consumer vehicles across the globe would help Wayve’s tech learn how to drive in different environments and cultures — a crucial step towards building fully autonomous vehicles.
“Ultimately we want this technology to be able to work anywhere around the world. That’s going to require some exposure to data around the world,” Kendall said.
Robotaxi wars
After years of promises and false starts, autonomous vehicles seem to have taken a big step closer to breaking through to the mainstream.
Alphabet-owned Waymo, which runs its robotaxi ride-hailing network in several US cities, recently said it serves around 150,000 paid rides a week, up from 50,000 in Q2.
Despite scrutiny over its Full-Self-Driving technology, Tesla is also set to join the robotaxi wars after unveiling a steering wheel-less “Cybercab” at an event in California in October.
Wayve has taken a different approach. Rather than building its own robotaxi fleet, the company plans to license its software to different automakers. Kendall said this would give Wayve a crucial advantage over firms like Waymo, which currently only operate in certain cities.
“I think the way that the way that autonomy will succeed at scale is through a system that has the intelligence to make decisions itself. And a geofenced approach fundamentally limits the utility of such a system,” said Kendall.
Ultimately, Kendall said that autonomous vehicles will only have their “ChatGPT moment” once they expand beyond specially modified robotaxis into everyday consumer vehicles.
“I would argue that getting this into consumer vehicles is what’s going lead to that experience. It’s not constrained, geofenced, affluent robotaxi models,” he said.
Wayve’s is testing its advanced driver assist technology in California with a safety driver.
Kendall added that Wayve’s “embodied” AI system’s ability to understand the physical world has applications beyond self-driving cars and hinted that Wayve could one day join Elon Musk’s Tesla in entering the robotics field.
“The majority of the work we do is in the physical world. Having intelligent and trusted machines that we can delegate tasks to and free up time for what matters most is something I’d love to enable,” he said.
“I think the first example of that we’ll see is with self-driving, but longer term there’s going to be a huge wealth of opportunities. I’d love Wayve to help get us into this age of autonomy,” he added.
