Physicl raises $10M to make the digital world real enough for robots

Physicl, a new startup led by founder and CEO Alex de Vigan, has raised $10 million in seed funding led by Inovia Capital, TechCrunch reports. The company aims to give world model developers and robot makers more realistic simulations for training. De Vigan previously spent a decade building nfinite, which helped brands and retailers produce 3D models of products for online display, a business he says generative AI was set to overtake. He is now repurposing that team's technology for converting messy real-world information into 3D representations with accurate geometry, materials and physics.

According to TechCrunch, researchers working on physical AI have been disappointed by the results of training on video and put off by the cost of collecting data through remote operation. Simulation is a promising alternative, but matching real-world physics has been hard. Physicl is focusing on post-training models for difficult edge cases and is releasing Studio, a product with 10,000 3D environments and 100,000 object assets. Users can build scenes that generate synthetic data for specific problems, such as a robot struggling to spot a cup on a cluttered counter or in shadow.

SpAItial, which turns 2D imagery into 3D simulations and released its Echo-3 product last week, worked with Physicl as an early customer. Its CEO, Matthias Niessner, said hand-curating data is slow and that generating it automatically according to training needs is the appeal. De Vigan concedes the crowded market for robotics data and tooling is still "very blurry," and he wants Physicl to be ongoing infrastructure rather than simply a data vendor.