London-based company building the AI 'driver' for self-driving cars, using a single AI system that learns from real driving instead of hand-written rules.
Wayve builds the "driver" for self-driving cars. Most self-driving systems are made of many separate parts, each programmed with rules and relying on very detailed maps. Wayve takes a different approach. It trains a single AI system on huge amounts of real driving data, so it learns to drive more like a person does, from cameras and experience. The company calls this approach AV2.0. It says the system doesn't need detailed maps and can be adapted to different vehicles, from cars to delivery vans.
The idea started at the University of Cambridge. Alex Kendall and Amar Shah were both studying for PhDs in machine learning when they founded Wayve in 2017, convinced that a car should learn to drive rather than be told how. Alex Kendall is now the company's chief executive. Wayve moved its base to London, near King's Cross, and also has offices in Silicon Valley and Vancouver.
Wayve says its aim is to pioneer embodied AI for a smarter, safer world. "Embodied AI" means AI that doesn't just process information but acts in the real world, like a driver steering through traffic or, one day, a robot. Its stated goals are to turn years of research into dependable software for the car industry, get its first product into many vehicles, and use the technology to help cut road deaths and the cost of accidents.
Rather than building cars, Wayve sells its software to carmakers. Its product, the Wayve AI Driver, is designed to be added to vehicles at scale, and its website names Mercedes-Benz and Microsoft (supercomputing for training its models) among its partners. Wayve and Uber also run supervised self-driving rides in London. Its research unit, Wayve Labs, explores how machines can perceive, reason and act in the physical world. Projects include GAIA, which generates driving videos to test rare situations, and LINGO-1, an AI that explains why the car is driving the way it is.
Wayve describes a culture built around tackling hard problems, high standards, honest feedback and learning from failure. Scientists and engineers hold regular research forums and reading groups and test ideas quickly on real roads. Many have Master's degrees or PhDs, but the company says it also values people who took different paths. Typical roles include machine learning engineer and applied scientist. The company says it welcomes several interns each year, who join a team and experience full-time work, so keep an eye on its careers page for current openings.
Website: wayve.ai
PURPOSE helps school and university students explore 62 sectors and over 1,000 entry-level careers in the UK, then connects them directly with employers.
Explore PURPOSE