Rivian has acquired UK startup Third Dimension AI to improve how it trains and tests autonomous driving systems.
The US electric vehicle maker will bring the startup’s 15-person team into its autonomy and AI division. Third Dimension uses artificial intelligence to turn real-world images and video into detailed 3D environments that engineers can manipulate to test driving software. Financial terms of the acquisition were not disclosed.
“Third Dimension’s team has built impressive technology that turns real-world data into detailed 3D environments,” said James Philbin, Rivian’s Senior Vice President of Autonomy and AI. “This allows us to recreate challenging driving situations, vary the conditions and improve how our systems respond.”
Autonomous vehicle developers use simulation to train and validate their systems, but conventional virtual environments can struggle to reproduce the complexity of real-world driving. Building simulations from actual driving footage offers a way to create more realistic scenarios while allowing engineers to change individual variables and test the results repeatedly.
Rivian plans to use Third Dimension’s technology to generate synthetic training data and simulate complex driving conditions. The company said the acquisition would strengthen its data and simulation capabilities across its consumer and robotaxi programmes.
The deal comes as Rivian works towards more advanced automated driving. Its existing Universal Hands-Free feature allows drivers to take their hands off the wheel on supported roads, although they must continue to monitor the road and be ready to take control.
The company is also developing autonomous R2 sport utility vehicles for a planned robotaxi service with Uber, which is expected to launch in 2028. The vehicles will require a self-driving system capable of handling journeys without a human driver ready to intervene.
Rivian already collects driving data from customer vehicles whose owners have enabled external camera sharing. Combining that data with controllable 3D simulations could allow engineers to test a broader range of situations than they can gather through ordinary driving alone.
The challenge is ensuring that performance in a virtual environment translates to the road. Simulations can expose weaknesses in autonomous systems, but their results are only useful if the generated scenarios accurately reflect how vehicles, pedestrians and other road users behave in reality.
Founded in 2024, Third Dimension’s technology was originally developed for applications including video game production, with the founders also identifying potential uses in autonomous vehicles and defence.
