Rivian Buys Third Dimension to Simulate Rare Driving Scenarios

A mattress in the fast lane or a child darting between parked cars are rare events for a human driver, but they are critical failures for autonomous software. Rivian is buying a 15-person UK startup to stop waiting for those moments to happen naturally. The company acquired Third Dimension to convert real-world fleet footage into digital test tracks where engineers can alter variables and stress-test the system.

The startup was founded in 2024 by Tolga Kart and Piotr Sokólski. Kart spent two years on Tesla’s Autopilot simulation team and previously led engineering at Parallel Domain. Sokólski spent four years at Wayve building photorealistic neural simulators. Third Dimension raised a $7 million seed round in October 2024, led by Felicis, and listed autonomous vehicle companies as a target market.

The entire team is moving into Rivian’s autonomy and AI group, which is led by James Philbin. The deal expands Rivian’s London presence, as most of the new staff are based in the UK. Philbin told TechCrunch that simulation is central to how the company trains and validates its driving systems. He noted that Third Dimension has built technology that turns real-world data into detailed 3D environments.

The process relies on data from owners of the R1S, R1T, R2, and commercial vans who opt into camera sharing. Philbin said the fleet sends back more than 400,000 driving sequences every day. While most of that footage is uneventful, the team will isolate the rare, high-difficulty clips. Engineers can then rebuild those scenes in 3D and change conditions, such as weather or traffic density, to test the software's reactions.

Rivian is not alone in this strategy. Waymo unveiled its Waymo World Model in February, which uses Google DeepMind’s Genie 3 to generate scenarios its vehicles have not yet encountered. Tesla has also demonstrated a neural world simulator for its Full Self-Driving software. However, rebuilding actual street scenes from real customer data offers a high-fidelity alternative to purely synthetic generation.

The move comes as Rivian tightens its timeline for driverless operations. In March, Uber agreed to invest up to $1.25 billion in Rivian to deploy R2 robotaxis. Only $300 million of that sum is paid up front. The remainder is tied to autonomy milestones, starting with 10,000 driverless R2 vehicles operating in San Francisco and Miami beginning in 2028.

Consumer-facing features are also on a shifting schedule. Rivian’s point-to-point driving feature is due to reach a small group of invited Autonomy+ owners later this year, with testers required to sign nondisclosure agreements. CEO RJ Scaringe promised eyes-off driving for 2026 in January 2025, but that target has moved to 2027. Level 4 capability is planned for 2028 on the Gen 3 hardware shown at the company’s AI & Autonomy Day.

Rivian did not disclose the purchase price for Third Dimension. The acquisition adds a specialized simulation team to the autonomy division without a large financial outlay relative to the overall robotaxi investment. The new workflow allows engineers to process the daily influx of fleet video and isolate specific edge cases for rigorous testing.