Physical AI at NeurIPS 2025 | Annoucements

Omniverse Announcements
1

Physical AI at NeurIPS 2025

Here are two major NeurIPS 2025 updates that highlight how NVIDIA is advancing the Physical AI ecosystem, from industry-scale AV models to accessible tools for generating high-quality synthetic data:

Alpamayo-R1

NVIDIA DRIVE Alpamayo-R1, the world’s first open industry-scale reasoning vision language action model for mobility.

See how Cosmos can accelerate AV development:

  • LidarGen, the first world model that can generate lidar data for AV simulation.
  • Omniverse NuRec Fixer, a model for AV and robotics simulation that taps into NVIDIA Cosmos Predict to near-instantly address artifacts in neurally reconstructed data, such as blurs and holes from novel views or noisy data.
  • Cosmos Policy, a framework for turning large pretrained video models into robust robot policies — a set of rules that dictate a robot’s behavior.
  • ProtoMotions3, an open-source, GPU-accelerated framework built on NVIDIA Newton and Isaac Lab for training physically simulated digital humans and humanoid robots with realistic scenes generated by Cosmos world foundation models (WFMs).

Read the blog.

Cosmos Cookbook

Building powerful physical AI requires diverse, high-fidelity, controllable data, but real-world collection is costly and often risky. In our new blog, we cover how to generate synthetic data using NVIDIA Cosmos Transfer, our world-to-world transfer model, with detailed recipes from the Cosmos Cookbook.

Featured recipes from the cookbook:

  • Augmenting Video Data: Effortlessly change backgrounds, lighting, object properties, and maintain consistency using control modalities (depth, segmentation, edge, vis).
  • Autonomous Driving Data: Transform videos for domain adaptation in AV research, featuring dynamic scene augmentation across weather, lighting, and more.
  • Sim2real Data Augmentation for Robots: Improve Sim2Real performance for mobile robots by generating photorealistic, domain-adapted data from simulation.
  • Synthetic Data for Urban Traffic Scenarios: The cookbook provides a complete workflow to create photorealistic smart city data, accelerating training for vision-language models.

Access the Cookbook.

Stay up-to-date with the latest Omniverse announcements by following us on Discord, Instagram, LinkedIn, Threads, X, and YouTube.

1 Like