The development
What happened
NVIDIA’s SimReady Central Library supplies simulation-ready OpenUSD assets for training and evaluating physical AI systems and building AI factory digital twins. Its current Hugging Face post describes more than 2,000 assets, prepared with properties such as mass, collision geometry and friction, and workflows for composing them into robot workcells and industrial environments.
Key facts
SimReady assets are OpenUSD models prepared with properties required for simulation workflows, beyond the detailed 3D geometry of a model such as CAD.
NVIDIA describes more than 2,000 assets on Hugging Face, with free, ungated access; this is the publisher’s current count, not an independent inventory.
The post describes OEMs converting CAD equipment models into SimReady assets and says the roadmap includes more OEM partners and a growing library.
Examples in the post include building a warehouse pick-and-place scenario in Isaac Sim and comparing AI factory layouts with racks, cooling equipment and supporting infrastructure.
The context
Why it matters
Prepared physical properties can reduce manual setup of environments for robot training and evaluation. Developers still need assets with profiles suited to their simulation runtime and task. NVIDIA’s post describes an asset library and example workflows, not independently validated performance or a guarantee that every asset supports every use.