NVIDIA Omniverse Mega Blueprint

2025-01-06 · Publisher : NVIDIA · Domain : Industry 4.0 & 5.0

At CES 2025, NVIDIA unveils Mega, an Omniverse reference blueprint for testing robot fleets in a factory or warehouse digital twin before any real-world deployment.

### Background On 6 January 2025, at CES, NVIDIA announces four new Omniverse blueprints, including **Mega**, designed to develop and test **robot fleets at scale** inside a factory or warehouse digital twin. Robot "brains" are trained and validated in simulation before reaching the real site. ### Key facts - Mega relies on Omniverse Sensor RTX APIs and the OpenUSD format. - Simulation loop: robots perceive the results of their actions through simulated sensors, then plan the next action until policies are validated. - First Omniverse integrators named by NVIDIA: Accenture, Altair, Ansys, Cadence, Foretellix, Microsoft and Neural Concept; Siemens announces Teamcenter Digital Reality Viewer the same day. - On 31 March 2025, at Hannover Messe, Mega becomes available in **preview** on build.nvidia.com; Accenture and Schaeffler use it to simulate Agility Robotics' Digit humanoid. ### Why it matters For manufacturing engineering, tuning mobile robotics and shop floor layouts moves to simulation first. Costly on-site trials are replaced by virtual iterations, provided a faithful twin of buildings, flows and equipment is available. ### Takeaway The digital twin becomes a training ground for physical AI, not just a visualization tool.

Tags : Jumeau numérique, Simulation, Robotique, NVIDIA, IA physique