Physicl Raises $10M Seed Funding to Create Realistic Simulations for Robot Training

New Venture Focuses on Robot Simulation

Alex de Vigan, founder of nfinite, has launched Physicl, a new company that has raised $10 million in seed funding. Inovia Capital led the funding round. Physicl's objective is to assist world model designers and robot builders in solving training problems by providing more realistic simulations.

Leveraging Previous Technology for AI

De Vigan's previous company, nfinite, specialized in creating 3D models and product renderings for e-commerce. With the advent of generative AI, de Vigan identified a new application for his team's existing technology stack: creating faithful 3D environments for AI. He noted that while AI excels in generative imagery, it lacks understanding, interaction, and manipulation of space and objects.

Addressing Challenges in Robot Training

Researchers developing physical AI models for robots face difficulties in training them effectively, particularly for tasks involving diverse, small objects. Training on video is often insufficient, and gathering data through remote operations is expensive. Digital simulations are a promising alternative, but ensuring these simulated worlds accurately reflect real-world physics remains a challenge. Physicl aims to bridge this gap by transforming unstructured real-world information into high-fidelity physical representations with correct geometry, materials, and physics.

Product Offering and Market Position

Physicl is entering a competitive market of startups providing data and tools for foundation model builders and robot operators. The company plans to focus on post-training models for difficult edge cases. Its product, Studio, includes 10,000 3D environments and 100,000 assets, allowing users to create scenes and generate synthetic data for specific robot training problems.