NVIDIA shows frontier AI agents building Omniverse simulations
Reported by NVIDIA and Blockchain News2 sources ↓
Turning a simulation idea into a working application means assembling assets, connecting physics and rendering, and checking that the scene behaves as intended.
Sources
Details
NVIDIA described projects in which developers used AI agents, including GPT-6 Astra, with Omniverse libraries to assemble and inspect simulation applications. The workflows involved developers giving natural-language instructions, reviewing results and directing changes; NVIDIA says its libraries provide GPU-accelerated physics, rendering and sensor simulation.
In a warehouse example, NVIDIA engineer Bhavik Sanganee used Astra to expand a simulation and investigate slowdowns. After splitting physics calculations, image rendering and video preparation into concurrent work, the simulation ran at 30 frames per second at full HD for two minutes with 80 conveyors, 40 robot arms and 800 boxes; the arms completed 640 box placements. Doubling the scene again reduced performance, and testing traced the slowdown to the physics and robot-control pipeline.
Other demonstrations included using panoramic photos and point-cloud measurements to create an editable building model, and an interactive gas-turbine application with six test procedures covering cases such as load changes, a stuck fuel valve and an exhaust-sensor error. NVIDIA also showed a kitchen scene assembled from 180 placements of 63 SimReady library assets, plus 18 custom geometry pieces; the company said the library, announced at GTC Berlin, contains more than 2,000 simulation-ready USD assets.
Written from coverage as of
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh