NVIDIA details how developers are using frontier AI models with Omniverse libraries to build, test and refine simulation applications.

NVIDIA said developers are using frontier AI models together with Omniverse libraries to turn simulation ideas into working applications, a workflow aimed at building and refining digital twins, robot tests and scenario analysis tools. In a developer blog published Oct. 29, 2026, the company described a series of examples showing AI agents assembling assets, connecting physics and rendering, and checking whether scenes behave as intended.

The post said developers direct AI agents with natural-language instructions, review the outputs and guide changes, while Omniverse libraries supply GPU-accelerated physics, rendering and sensor simulation. NVIDIA framed the approach as a way to help explore scenarios, investigate failures and improve designs across robotics, industrial simulation and autonomous-driving workflows.

Among the examples, NVIDIA described David Wehr of its simulation technology team using Astra to reconstruct an editable building model from site scans. The workflow interpreted panoramic photos, created building geometry and aligned it with point-cloud measurements, with OpenUSD storing the model and the ovrtx library handling inspection renders. NVIDIA said the aim was a simulation-ready building where shelving and machinery can be added or repositioned before physical changes are made.

Pradeep Kaushik of the Omniverse developer relations team built a what-if application around a simulated gas turbine and generator. NVIDIA said the prototype lets users vary load, ambient temperature and fuel composition, then compare power, efficiency, fuel consumption and equipment response. The company said six test procedures included load changes, a stuck fuel valve and an exhaust-sensor error, and that the application used Cantera for thermochemistry and SciPy for equipment dynamics.

Source link