Building robotics simulations and high-fidelity AI factory digital twins requires more than just visual 3D models. Developers often struggle with manual work to add properties like collision geometry, mass, and friction. SimReady Central, a new library on Hugging Face, addresses this by providing over 2,000 OpenUSD assets pre-configured for simulation workflows.
Beyond Simple Visualization
SimReady assets differ from standard CAD models because they come prepared with the physical properties required for interaction. For a robot manipulation task, this includes mass and friction data. In AI factory workflows, assets may include electrical or thermal metadata necessary for specialized applications.
The library is built on OpenUSD, an open framework for describing 3D scenes. It is governed by the SimReady Foundation, which provides specifications, profiles, and validation tools to ensure assets meet the requirements of specific simulation use cases.
A Growing OEM Ecosystem
Leading Original Equipment Manufacturers (OEMs) are already contributing to the library. Robotics companies such as FANUC, Universal Robots, and Comau are offering discoverable robot arms and grippers. Meanwhile, infrastructure providers like Vertiv offer assets like cooling equipment and server racks for AI factory digital twins. Even consumer electronics giants like LG Electronics are participating, providing models of appliances to help train robots for home environments.
AI-Assisted Scene Authoring
The library integrates advanced natural-language search tools, enabling developers and AI agents to find assets efficiently. The platform supports AI-assisted authoring where frontier models, such as OpenAI’s GPT-6 Astra, can be used to search for assets and write Python code to assemble complex simulation scenes automatically. This allows developers to quickly compare different factory layouts or evaluate robot behaviors under varying conditions.
