Physicl secures $10m to bridge the gap between digital simulations and real-world robotics

Physicl, a new venture founded by Alex de Vigan, has secured $10 million in seed funding, led by Inovia Capital. The company aims to address significant hurdles in training physical AI models by creating highly realistic digital simulations for robot builders and world model designers. De Vigan, who previously built nfinite, a platform for e-commerce 3D modeling, believes that while generative AI excels at imagery, it lacks the capacity to understand and manipulate physical space. Physicl’s technology is designed to transform unstructured, real-world data into high-fidelity physical representations, ensuring simulations accurately reflect real-world physics. The startup is launching Studio, which provides 10,000 3D environments and 100,000 assets, allowing users to generate synthetic data for training robots on specific edge cases, such as object recognition in shadow. Early customer SpAItial, a world model builder, utilized Physicl to enhance its Echo-3 product. Matthias Niessner, CEO of SpAItial, noted the appeal of Physicl's capability to auto-generate necessary training data on the fly, overcoming the manual curation challenges. De Vigan envisions the platform evolving beyond a mere data provider, allowing customers to feed massive datasets, like trillions of images or millions of hours of LIDAR capture, into the system to generate billions of training permutations.