Nvidia Debuts Cosmos Reason and Advanced AI Infrastructure for Next-Gen Robotics Development
In a significant move, NVIDIA unveiled an array of innovative AI models, libraries, and infrastructure tailored for robotics developers on Monday. The highlight of the announcement is Cosmos Reason, a sophisticated 7-billion-parameter “reasoning” vision language model designed for physical AI applications and robots.
Joining the existing Cosmos world models are Cosmos Transfer-2, capable of expediting synthetic data generation from 3D simulation scenes or spatial control inputs, and an optimized, speedier version of Cosmos Transfers. During its presentation at the SIGGRAPH conference, NVIDIA emphasized that these models are intended for generating synthetic text, image, and video datasets to train robots and AI agents.
Cosmos Reason enables robots and AI agents to make decisions based on its memory and physics understanding, acting as a planning model to predict the next step an embodied agent might take. The company suggests it can be used for data curation, robot planning, and video analytics.
NVIDIA also introduced new neural reconstruction libraries, including one for a rendering technique that allows developers to simulate the real world in 3D using sensor data. This capability is being integrated into open-source simulator CARLA, a popular developer platform. There’s an update to the Omniverse software development kit as well.
The announcements also include new servers for robotics workflows. The NVIDIA RTX Pro Blackwell Server offers a unified architecture for robotic development workloads, while NVIDIA DGX Cloud is a cloud-based management platform.
With these advancements, NVIDIA continues to expand its horizons in the realm of robotics as it explores new applications for its AI GPUs beyond AI data centers. Your feedback and insights are always welcome to help us evolve further. Please take our survey to share your perspective with us.
