AdvertiseDonate
Jersey CityNew YorkNewarkPhiladelphiaBaltimoreBostonWashingtonBuffalo
Digital Wire Desk · 31 Hudson Yards, 11th Floor, New York, NY 10001

Rivian acquires UK startup to convert 400,000 daily fleet clips into 3D test environments

The 15-person firm, founded by former Tesla and Wayve engineers, will join Rivian's autonomy group to help train its robotaxi software using real-world driving data.

By Benjamin Rousseau ·

Illustration for: Rivian acquires UK startup to convert 400,000 daily fleet clips into 3D test environments
Illustration generated for this article, not a news photograph.

Rivian's push to build a fully autonomous vehicle has accelerated with the acquisition of a small UK firm specialized in generative AI. The move aims to transform raw camera footage from Rivian's existing fleet into complex, three-dimensional test scenarios for its self-driving software.

The company bought Third Dimension, a 15-person startup founded in 2024. TechCrunch first reported the transaction, though Rivian has not released the purchase price. Third Dimension previously described itself as a tool for building 3D environments for game engines, raising $7 million in a seed round led by Felicis in October 2024.

The founders bring specific experience from major autonomy players. CEO Tolga Kart worked on Tesla's Autopilot simulation for over two years before leading engineering at Parallel Domain. Cofounder Piotr Sokólski spent four years at Wayve developing photorealistic neural simulators for self-driving cars.

All members of the team are moving into Rivian's autonomy and AI group, which is directed by James Philbin. This hire also increases Rivian's operational footprint in London. Philbin told TechCrunch that simulation is central to training and validating their driving systems, praising the team's ability to turn real-world data into detailed 3D environments.

The new process relies on data from owners of the R1S, R1T, R2, and commercial vans who opt into camera sharing. Philbin stated that this fleet sends back more than 400,000 driving sequences daily. While most of this data is uneventful, the rare events that challenge autonomous systems, such as a mattress in a fast lane or a child running between parked cars, are difficult to find in time.

Engineers can now recreate these rare scenes in 3D and adjust variables to test the software thoroughly. This approach mirrors strategies used by competitors. Waymo unveiled a world model built on Google DeepMind's Genie 3 in February, and Tesla has demonstrated a neural world simulator for its Full Self-Driving software.

The acquisition arrives as Rivian faces tight deadlines for its robotaxi plans. In March, Uber agreed to invest up to $1.25 billion in Rivian to deploy R2 robotaxis, with $300 million paid upfront. The remaining funds are tied to milestones, including 10,000 driverless R2 vehicles on Uber's network in San Francisco and Miami starting in 2028.

Consumer timelines have also shifted. The point-to-point driving feature for invited Autonomy+ owners is expected later this year, requiring testers to sign nondisclosure agreements. CEO RJ Scaringe initially promised eyes-off driving for 2026 in January 2025, but the target has moved to 2027. Level 4 capability is planned for 2028 on the Gen 3 hardware shown at AI & Autonomy Day.

Source: Electrek

More Tech · Report a business · Got a Story?

Get Digital Wire Desk in your inboxYour daily read on the news. Free, a few times a week, unsubscribe any time.

Related stories

Get Digital Wire Desk in your inboxYour daily read on the news. Free, a few times a week, unsubscribe any time.