Physicl raises $10 million to make robot-training simulations more realistic

Physicl has raised $10 million in seed funding from Inovia Capital to help world model developers and robot builders create more realistic simulations for training physical AI systems. The company is applying technology originally developed to produce 3D product renderings for online retail.

Founder and CEO Alex de Vigan spent the past decade building nfinite, a platform that helped brands and retailers create 3D models of products for online displays. As generative AI changed that business, he saw another use for its technology: representing 3D environments with accurate geometry, materials and physics.

Robot developers have found video-based training limited for tasks involving varied small objects, while collecting data through remote operations can be expensive. Simulations offer another route, but making digital environments faithfully reflect real-world physics remains difficult, de Vigan said.

Physicl’s new product, Studio, includes 10,000 3D environments and 100,000 assets representing objects. Users can build scenes and generate synthetic training data for specific challenges, such as recognizing a cup on a cluttered counter or in shadow.

SpAItial worked with Physicl as an early customer to improve the fidelity of its products. Its latest world model, Echo-3, was released last week. SpAItial founder and CEO Matthias Niessner said the company needed high-fidelity synthetic data for its tool, which creates 3D simulations from 2D imagery. Physicl says it aims to support ongoing, on-demand synthetic data generation as models are trained.