NVIDIA Brings Generative AI to OpenUSD for Robotics and Industrial Design
NVIDIA introduces new generative AI models and NIM microservices that make OpenUSD more accessible for robotics and industrial digital twins. The tools allow developers to generate 3D code and materials using simple text prompts.
NVIDIA announces major advancements to OpenUSD that expand the 3D framework beyond creative industries into robotics, industrial design, and engineering. These new tools, built on the NVIDIA Omniverse platform, help developers build highly accurate virtual worlds and digital twins for the next evolution of physical AI.
The company introduces the world's first generative AI models for OpenUSD as NVIDIA NIM microservices, including USD Code and USD Search. These microservices enable developers to generate OpenUSD Python code from text prompts, apply materials to 3D objects, and search massive 3D libraries using natural language, effectively acting as AI copilots for 3D workflows.
New USD connectors and developer tools also allow users to stream massive, RTX ray-traced datasets directly to Apple Vision Pro. NVIDIA states that these enhancements make physically based virtual worlds accessible to industries like manufacturing and automotive, marking the arrival of the generative AI boom for heavy industries.