New world AI models, libraries, and other infrastructure for robotics developers were unveiled by Nvidia on Monday.
Cosmos Reason is the most notable, a 7-billion-parameter “reasoning” vision language model for physical AI applications and robots.
Cosmos Transfer-2, which can expedite the generation of synthetic data from 3D simulation scenes or spatial control inputs, and a distilled version of Cosmos Transfers that is more optimized for speed, are also joining the current collection of Cosmos world models.
Nvidia stated in its announcement at the SIGGRAPH conference on Monday that these models are intended to be employed to generate synthetic text, image, and video datasets to train AI agents and machines.
According to Nvidia, Cosmos Reason enables robots and AI agents to “reason” due to its memory and physics comprehension, which allows it to “serve as a planning model to reason what steps an embodied agent might take next.” According to the organization, it applies to video analytics, automation planning, and data curation.
The company also introduced new neural reconstruction libraries, designed to simulate the real world in 3D using sensor data.
This rendering capability is also being integrated into the open-source simulator CARLA, a favored developer platform. There is even an update to the Omniverse software development kit.
Additionally, there are novel servers that are designed for robotic workflows. The Nvidia DGX Cloud is a cloud-based management infrastructure, whereas the Nvidia RTX Pro Blackwell Server provides a unified architecture for robotic development workloads.
These pronouncements coincide with the semiconductor giant’s expansion into robotics, as it pursues the next significant application for its AI GPUs beyond AI data centers.
