Key facts
- Autonomous-driving startup Wayve has raised $2.8 billion from investors including Nvidia, Mercedes-Benz, and Nissan.
- Wayve's AI driving system uses end-to-end machine learning, aiming to mimic human learning.
- The company plans to deploy its technology in Stellantis robotaxis for Uber's ride-hailing network.
- Wayve's system is designed to be compatible with a wide array of sensors and AI chips.
- Waymo's recent expansion has contributed to renewed investor interest in the autonomous driving sector.
Autonomous-driving startup Wayve is attracting significant investor interest, having secured $2.8 billion from investors and strategic partners including Nvidia, Mercedes-Benz, and Nissan. The London-based company utilizes an end-to-end machine learning approach for its AI driving system, which aims to translate sensor data into driving decisions in a manner similar to human drivers. This contrasts with traditional methods that combine AI with software coding and high-definition maps to establish preset rules for vehicle responses.
Wayve's approach is comparable to Tesla's, which also transitioned to an end-to-end model. However, Wayve's system is designed to be compatible with a wide range of sensors and AI chips, potentially allowing it to license its technology to various driverless-car developers globally. CEO Alex Kendall stated the company's goal is to enable full self-driving for any vehicle, brand, and location.
The intensifying competition in the autonomous-driving industry, partly fueled by the expansion of Alphabet's Waymo, has rekindled investor enthusiasm. While many developers are incorporating end-to-end learning, the 'black box' nature of these systems makes interpreting their decisions challenging compared to rule-based approaches. Wayve engineers argue that pre-programmed systems can be brittle in unusual situations, whereas human drivers adapt conservatively when faced with the unknown.
Waymo, while using end-to-end AI, maintains that a conventional, rules-based approach is still necessary for safety at scale. Nissan, a Wayve customer, is closely evaluating the technology, with its tech chief noting the difficulty in understanding how Wayve's system makes decisions. Experts suggest that end-to-end models may accelerate commercial deployment but do not necessarily guarantee superior safety over traditional methods, with some estimating it could take a decade to achieve widespread driverless safety in the U.S.