IntelligentHQ
IntelligentHQ
Skip to content

Digital Twins & Simulation

Tools for building physically accurate virtual copies of factories, warehouses, data centers and products: Omniverse libraries, the OpenUSD open standard, DSX simulation for AI factories, Isaac Sim for robots and Cosmos models for synthetic data.

Technology profiles for this category are in research.

Overview

A digital twin is a simulation tied to a real facility or product, used to test layouts, processes or robots before changing the real thing. This category covers NVIDIA libraries for industrial twins and the open standard they build on.

Omniverse is now offered as libraries, including ovrtx for RTX rendering, ovphysx for physics and SimReady for checking assets, and it is part of NVIDIA Agent Toolkit; NVIDIA labels some of these libraries pre-release and not enterprise-supported. Its licensing page says Omniverse is free for development and production, with enterprise support through AI Enterprise; the Omniverse Launcher is deprecated. Scenes are described in OpenUSD, which Pixar created and which is now developed through the Alliance for OpenUSD, founded by Pixar, Adobe, Apple, Autodesk and NVIDIA. For AI data centers, DSX Sim and the Omniverse DSX Blueprint model the facility before construction.

A twin pays off when changes are expensive or risky in the real world. If a floor plan or a spreadsheet model answers the question, a full 3D twin may cost more than it saves.1234

Problems it addresses

  • Testing layout changes safely1

    Moving lines or racks in a running site is costly. Industrial facility digital twins are a listed Omniverse use.

  • 3D data spread across tools3

    CAD, BIM and simulation tools use different formats. OpenUSD gives them a shared scene description.

  • Too little training data15

    Perception models need many labeled examples. Synthetic data generation is an Omniverse use, and Cosmos can generate synthetic video.

  • Designing AI data centers4

    Power, cooling and network choices interact. The Omniverse DSX Blueprint provides building blocks for physically accurate AI factory twins.

A typical workflow

  1. Define the decision

    Agree which decision the twin should support, such as throughput, layout or energy use.

  2. Bring assets into OpenUSD1

    Convert CAD and facility data to OpenUSD and validate assets with SimReady.

  3. Add physics and rendering1

    Simulate behavior and sensor output with ovphysx and ovrtx.

  4. Simulate scenarios46

    Test robots in Isaac Sim or AI factory designs in DSX Sim.

  5. Generate data variations5

    Add generated variations to synthetic data with Cosmos models.

Digital Twin Opportunity Lab

Find the first digital twin worth building for your site.

Open the lab

Next steps

  1. Choose one decision with a known cost, such as a line rebalance, as the twin's first purpose.

  2. List which of your CAD and 3D sources can export to USD.

  3. Use the Digital Twin Opportunity Lab to estimate whether a twin pays back for your site.

  4. Check the license of each Omniverse library you plan to ship, since some GitHub libraries are pre-release.

Sources

  1. NVIDIA Omniverse (product page) (opens in a new tab)NVIDIA · Vendor-reported
  2. Omniverse Licensing (opens in a new tab)NVIDIA · Vendor-reported
  3. Alliance for OpenUSD (opens in a new tab)Alliance for OpenUSD · Third-party reporting
  4. NVIDIA DSX AI Factory Platform (opens in a new tab)NVIDIA · Vendor-reported
  5. NVIDIA Cosmos (product page and FAQ) (opens in a new tab)NVIDIA · Vendor-reported
  6. NVIDIA Isaac Sim (developer page and FAQ) (opens in a new tab)NVIDIA · Vendor-reported

Fill out the form below to request your copy.

Name(Required)