The Industrial Metaverse is Thriving: Exciting Advancements in OpenUSD and Generative AI
The industrial metaverse is far from dead; it has evolved through projects focused on OpenUSD, the universal 3D data interchange framework essential for digital twins.
Nvidia has recently unveiled significant enhancements to Universal Scene Description (OpenUSD), aimed at broadening its application across robotics, industrial design, and engineering. These advancements will empower developers to create highly accurate virtual environments for the next wave of AI.
Generative AI Meets OpenUSD
During the Siggraph computer graphics conference in Denver, Nvidia revealed its integration of generative AI into OpenUSD. "We have built the world's first generative AI models that understand OpenUSD-based language, geometry, materials, physics, and spaces," stated Rev Lebaredian, Nvidia's vice president of Omniverse and simulation technology.
With these new generative AI capabilities and Nvidia-accelerated frameworks on the Omniverse platform, various industries can now develop applications for visualizing industrial design and engineering projects, as well as simulating environments for advancing physical AI and robotics.
Innovative Microservices for AI Development
The new Nvidia NIM microservices enhance OpenUSD by allowing the generation of OpenUSD language, automating Python code creation, and applying materials to 3D objects—all aimed at expediting digital twin development. Additional USD connectors facilitate seamless integration with robotics and industrial simulation data formats, enabling users to stream extensive, fully RTX ray-traced datasets to devices like Apple Vision Pro.
"The generative AI boom for heavy industries is here," Lebaredian noted. Historically, digital environments have primarily served creative industries. However, with recent advancements in Nvidia NIM microservices, any sector can now create physically based virtual environments and digital twins, fostering innovation and preparing for advancements in robotics.
Key Microservices Available Now
Nvidia has launched several preview microservices:
- USD Code NIM: Generates OpenUSD-Python code based on natural language prompts, facilitating input into OpenUSD viewing applications.
- USD Search NIM: Allows developers to search vast libraries of OpenUSD, 3D, and image data using natural language or images.
- USD Validate NIM: Ensures uploaded files' compatibility with OpenUSD release versions and generates fully RTX-rendered images via Omniverse Cloud APIs.
Anticipated upcoming microservices include:
- USD Layout NIM: Helps users build OpenUSD scenes using spatial intelligence and text prompts.
- USD SmartMaterial NIM: Predicts and applies realistic materials to computer-aided design objects.
Real-World Applications and Collaborations
Foxconn, a global manufacturing leader, is leveraging Nvidia's computing platform to create a digital twin of a factory in development. "Digital twins will help us accelerate the next wave of industrial manufacturing and autonomous machines," stated Zhe Shi, chief digital officer at Foxconn.
WPP, a top marketing and communication services company, is an early adopter of the USD Search and USD Code NIM microservices, utilizing them for generative AI-enhanced content creation for clients like The Coca-Cola Company. "The compatibility with our work processes and the leveraging of open standards have allowed us to launch innovative production tools rapidly," said Stephan Pretourius, WPP's CTO.
Expanding OpenUSD's Reach
Nvidia is extending its collaboration with Siemens to facilitate industrial workloads using OpenUSD, integrating it with Siemens’ Simcenter portfolio for enhanced decision-making and collaboration. This partnership will enable high-fidelity, real-time visualizations of complex data.
The newly announced OpenUSD Exchange software development kit allows developers to create robust OpenUSD data connectors, advancing the ecosystem further.
Join the OpenUSD Revolution
Nvidia's latest offerings, including the USD Search, USD Code, and USD Validate microservices, are now available for preview. Developers can now explore integrating generative AI into OpenUSD workflows with new tools and a reference workflow for building generative AI-enabled synthetic data pipelines.
To stay updated on generative AI and accelerated computing innovations, tune into Nvidia CEO Jensen Huang's discussions at Siggraph, the leading computer graphics conference.
Ranking
