Bridging Digital and Physical Worlds
Alex de Vigan is repurposing technology originally developed for e-commerce product renderings to address challenges in robotics. His new venture, Physicl, aims to help robot builders and world model designers improve training through more realistic simulations.
Physicl has raised $10 million in seed funding led by Inovia Capital. The company focuses on solving training bottlenecks by providing high-fidelity digital environments that mirror real-world physics.
De Vigan previously founded nfinite, a company centered on 3D modeling for retail. As generative AI shifted the landscape, he identified a gap in the technology's ability to comprehend and manipulate spatial environments, leading to the creation of Physicl.
Current methods for training physical AI, such as utilizing video or remote operation data, are often limited by high costs and data quality issues. Physicl aims to solve this by transforming unstructured, heterogeneous real-world information into accurate 3D representations with correct geometry and physics.
Synthetic Data for Edge Cases
The startup is launching a platform called Studio, which provides 10,000 3D environments and 100,000 assets. This allows developers to generate synthetic data for specific edge cases, such as training a robot to identify objects in difficult lighting or cluttered spaces.
Early collaborator SpAItial, a world model builder, utilized Physicl to enhance the fidelity of its Echo-3 tool. By leveraging high-fidelity synthetic data, SpAItial can more effectively generate 3D simulations from 2D imagery.
Physicl intends to function as an agentic workflow for synthetic data generation. De Vigan envisions a system where customers can process vast amounts of lidar or image data to create billions of training permutations, potentially allowing models to auto-generate the specific data they need during the training process.
Matthias Niessner, founder of SpAItial and a professor of robotics at the Technical University of Munich, noted that the ability to generate data on the fly based on a model's current training state represents a significant advancement in the field.
SpAItial, a world model builder, collaborated with Physicl as an early customer to improve the fidelity of its products, the latest of which, Echo-3, was released last week. SpAItial’s tool generates 3D simulations from 2D imagery, and to do that effectively, its CEO and founder Matthias Niessner said they needed to train on the kind of high-fidelity synthetic data Physicl offers.
However, Physicl wants to be more than a data provider; it sees itself as the tooling for ongoing synthetic data generation. “What we created 10 years ago, without knowing it, was basically an agentic workflow to transform unstructured information into high-fidelity 3D,” de Vigan said. He imagines customers bringing trillions of images or millions of hours of lidar capture data to feed into his platform and turn them into high-quality 3D representations that can be used to generate billions of permutations to train their models.
“The real challenge is, you have to curate the data by hand, and this is a very long cycle,” Niessner, also a professor of robotics at the Technical University of Munich, told TechCrunch. “To us what seems very attractive is you can just do this on the fly […] You’re running Claude, and it checks, ‘Oh, I need more data of this in my training,’ and then it auto generates that data, depending on the training state. That, to me, is super exciting, and this is exactly what Physicl is offering.”
