Physicl raises US$10 million for physical AI training data
Inovia Capital led the seed round, with U.S. Venture Partners participating. The Paris startup plans to expand its team and further develop its platform.
Paris-based Physicl has secured US$10 million in seed financing for its work creating training data for artificial intelligence operating in physical environments. The round was led by Inovia Capital and included U.S. Venture Partners.
The startup targets unusual conditions that can cause errors in robotics, autonomous driving and 3D vision models. These situations, known as edge cases, can involve unexpected obstacles or unfamiliar lighting.
Its initial product, Studio, allows researchers to reconstruct problematic scenarios as 3D scenes. Users can select a setting or describe one, then adjust illumination, camera viewpoints and object placement. The system generates training data intended to address the model’s weakness, allowing researchers to test changes and assess the results.
Physicl says its platform offers 10,000 environments in 3D and more than 100,000 objects, enabling billions of scene combinations. It already works with organizations including Meta, World Labs, Epic Games, Amazon Web Services, Nvidia and the Technical University of Munich.
Founder and chief executive Alex de Vigan previously established Nfinite, which produces 3D product imagery for online retailers. Physicl plans to put the funding toward hiring and further platform development.
