NVIDIA Enhances Physical AI Simulations for Robotics Developers

NVIDIA has launched new libraries and models to accelerate industrial AI and robotics simulation applications. The technology includes RTX ray-traced 3D Gaussian splatting, integrated into the CARLA simulator, and used by over 150,000 developers. Autonomous vehicle leader Foretellix is also incorporating NuRec for physically accurate synthetic data generation. NVIDIA DGX Cloud is now available on Microsoft Azure Marketplace for streaming OpenUSD and RTX-based applications at scale. Industry leaders like Amazon, Boston Dynamics, and Uber are rapidly adopting these technologies for real-world physical AI applications.

NVIDIA has recently announced a suite of new libraries and models aimed at accelerating industrial AI and robotics simulation applications. These innovations include RTX ray-traced 3D Gaussian splatting, integrated into the CARLA simulator, and used by over 150,000 developers. The technology allows for the capture, reconstruction, and simulation of the real world in 3D using sensor data, significantly enhancing the capabilities of physical AI models.

Additionally, NVIDIA DGX Cloud is now available on Microsoft Azure Marketplace, enabling developers to stream OpenUSD and RTX-based applications at scale. This platform simplifies the deployment of industrial and physical AI applications in the cloud, reducing setup time from weeks to days. Industry leaders such as Amazon, Boston Dynamics, and Uber are rapidly adopting these technologies for real-world physical AI applications, demonstrating the practical utility and potential market impact of NVIDIA's advancements.

The new NVIDIA Omniverse libraries and NVIDIA Cosmos world foundation models (WFMs) announced at SIGGRAPH are designed to enable developers to build physically accurate digital twins, generate synthetic data for training physical AI models, and build AI agents that understand the physical world. These tools are expected to transform industries by facilitating the development of tomorrow's robots and autonomous vehicles.

NVIDIA's initiatives are part of a broader trend towards the convergence of computer graphics and AI, which is fundamentally transforming robotics. By combining AI reasoning with scalable, physically accurate simulation, NVIDIA is empowering developers to create more advanced and efficient AI systems.

References:

[1] NVIDIA News. (2023). NVIDIA Opens Portals to the World of Robotics with New Omniverse Libraries, Cosmos Physical AI Models, and AI Computing Infrastructure. Retrieved from https://nvidianews.nvidia.com/news/nvidia-opens-portals-to-world-of-robotics-with-new-omniverse-libraries-cosmos-physical-ai-models-and-ai-computing-infrastructure
[2] NVIDIA. (2023). NVIDIA Omniverse on DGX Cloud. Retrieved from https://www.nvidia.com/en-us/data-center/omniverse-dgx-cloud/