NVIDIA unveiled major updates to its Isaac robotics platform, including the open-source Newton Physics Engine, the new Isaac GR00T N1.6 foundation model, and expanded Cosmos World Foundation Models.
Newton, codeveloped with Google DeepMind and Disney Research, is a GPU-accelerated physics engine designed for realistic humanoid simulation.
Now available in the Isaac Lab under Linux Foundation management, Newton enables developers to train robots for complex, real-world tasks, such as walking on uneven terrain or manipulating delicate objects.
Global research institutions, from ETH Zurich to Peking University, are already adopting it for advanced robotics work.
The Isaac GR00T N1.6 model introduces Cosmos Reason, a customizable reasoning vision-language model that gives robots humanlike planning and problem-solving skills.
It helps humanoids translate vague instructions into step-by-step actions, integrating prior knowledge and physics awareness.
Cosmos WFMs also expand with updates like Cosmos Predict 2.5 and Cosmos Transfer 2.5, which generate diverse datasets and photorealistic simulations at scale for training physical AI.
Alongside these tools, NVIDIA announced new AI infrastructure including the Jetson Thor platform, RTX PRO Servers, and the GB200 NVL72 rack-scale system, all aimed at powering multi-AI workflows for robotics.
With support from leading developers like Boston Dynamics, Agility Robotics, and Figure AI, NVIDIA is positioning Isaac and Omniverse as the central ecosystem for next-generation humanoid and physical AI development.
You may also want to check out some of our other recent updates.
Wanna know what’s trending online every day? Subscribe to Vavoza Insider to access the latest business and marketing insights, news, and trends daily with unmatched speed and conciseness! 🗞️
