acars that leave the showroom today have a built-in navigation system and (the more expensive models) cameras that can see all around, to help the driver with parking. But soon, in the not too distant future, cars will be sold with a built-in robot driver, who will find his way completely independently with the same cameras and navigation – at least if it is up to Alex Kendall, the co-founder and CEO of Wayve, a young British company that has specialized in autonomously driving cars.
After its founding seven years ago, Wayve took a different direction than most start-ups involved in autonomously driving cars, which do not seem to have made much progress after billions in investments in recent years. For example, Wayve does not use expensive radars, lidars (laser scanners that accurately map the environment) and terabytes of digital map material, but uses ‘ordinary’ home and garden cameras in combination with artificial intelligence (AI).
About the author
Bard van de Weijer is economics editor of de Volkskrant. He writes about the energy transition and its impact on daily life. .
The British company develops AI that learns in the real world by copying human behavior. If the robot car makes a mistake, for example by steering onto the side of the road, the human driver intervenes with a steering correction. In this way, the first version of Wayves artificial intelligence learned to keep course in 2018 on a quiet road just outside Cambridge, the university city where Kendall and fellow student Amar Shah developed the software.
Fall and rise
In a 2019 blog post, Kendall compared the way his AI learns to drive a car to the way children learn to ride a bike, “through trial and error.” Only accompanied by an adult who intervenes if something threatens to go wrong, every child usually learns to cycle in an afternoon. It was exactly the same for Kendall’s robot car, which started the day unaware. “In just over fifteen minutes we were able to teach her to follow the lane,” he writes.
If a human with two eyes can navigate safely through traffic was Kendall’s and Shah’s idea, then the robot car should be able to do the same. After all, many cars already come standard with cameras and powerful GPUs (graphics processors that specialize in video). If this technology can be used for autonomous driving, then it can be introduced at much lower costs and therefore for many more cars, they both thought.
Wayve does not want to develop cars itself, but merely sell the software to existing manufacturers, who can install it in any car, Kendall said. Just as TomTom once started installing navigation systems in Renault’s Clio and Zoe.
Great investment
That Kendall is not a digital dreamer with unattainable visions became clear this week, when it was announced that Japan’s SoftBank, Microsoft and Nvidia are investing almost a billion euros in the young company. An unprecedentedly high amount for Europe, but also on a global scale the billion is one of the top twenty largest investments in AI, according to Bloomberg.
“Vehicles will soon be able to interpret their environment like people, allowing them to make better decisions, which guarantees more safety,” said Kentaro Matsui of the largest investor SoftBank. According to him, AI will lead to the same revolution in mobility as previously happened with a service such as ChatGPT, which communicates in a human way. Matsui has been on Wayve’s board of directors since the investment.
Wayve will now have access to Microsoft’s vast computing platform on which the artificial intelligence can run. Nvidia builds the graphics chips needed for fast data processing.
The smart software not only looks at the road, it also studies PDF documents from the British government, among others, about local traffic rules. We use different data sources, said the director, whose fortune grew overnight to more than 400 million euros thanks to the deal, the British newspaper calculated Evening Standard.
According to Kendall, the reason that these parties invest so much money in the young company is because his system is so-called platform-independent: it should soon be possible to install it in any type of car, regardless of make or model, Kendall said this week in an interview with tech platform Tech crunch. This ‘brand independence’ should make it available to more people at a lower cost, just like TomTom did two decades ago.
Blessing from Tesla
Kendall, who grew up in New Zealand and studied in the United Kingdom, says he now works with ten major car manufacturers, but does not want to name any names yet. The idea is to install the software in millions of vehicles, so that the company receives data from all over the world to train the AI to be a safe robot driver; like a driving school with millions of teaching cars.
Elon Musk, the owner of Tesla, which has been working on self-driving cars for years, has also embraced Wayve’s working method. Musk thinks he can launch a robot taxi on the market this year. “We’ve been doing this since 2017. It was nice to see Tesla choose our approach last year,” Kendall told Bloomberg this week. ‘It’s great that they are joining the club.’
But unlike Musk, Kendall seems to be in little hurry. He refuses to give a date on which his self-driving system will be launched. Safety comes first, he says.
