Reimagining the road to self-driving cars
Summary
- Alex Kendall founded Wayve in 2017 while completing his Ph.D. at Cambridge, proposing an end-to-end neural network approach for autonomous driving that contrasted with the industry's modular stack methods.
- Wayve's technology relies primarily on cameras & AI to learn driving tasks through supervised & reinforcement learning, avoiding the need for high-definition maps, lidar, or hand-coded rules.
- The company faced significant skepticism & ridicule from industry giants like Waymo & Tesla, with Kendall being mocked by Elon Musk & dismissed at industry meetings for his contrarian strategy.
- Wayve chose to train its vehicles in London's chaotic, unpredictable environment rather than controlled settings, believing mastering hard scenarios first would enable easier adaptation elsewhere.
- Despite starting with only five cars & limited data compared to rivals driving millions of miles, Wayve developed algorithms to achieve performance from scarce data & reduced technical debt.
- By mid-2022, Wayve demonstrated improved reliability & fewer errors per mile, leading to a $1 billion funding round in 2024 led by SoftBank with investments from Microsoft & NVIDIA.
- The company now employs roughly 1,000 people, has tested vehicles in over 500 cities globally, & is in talks with nearly every major global automaker outside China.
- Nissan has committed to incorporating Wayve's driver-assistance software into its vehicles starting in 2027, & Wayve is set to operate robotaxis in London with Uber.
- Wayve's end-to-end approach is gaining industry traction as competitors like Tesla & Waymo shift toward unified neural networks for autonomous driving.
- Kendall aims to expand Wayve's platform beyond automobiles to factories, hospitals, farms, & homes, positioning the company as an operating system for intelligent machines.
Wayve founder Alex Kendall on end-to-end machine learning and embodied AI for self-driving cars, and his plan to rival Tesla and Waymo.
- Keywords
- AI Stack, Trends