Nvidia, Google DeepMind, and Disney Research have announced the creation of the Newton physics engine, set to enhance performance in robotics simulations. The collaboration was highlighted at GTC in Silicon Valley and aims to improve robotic training.
Announcement of the Collaboration
On March 18, 2025, Nvidia CEO Jensen Huang announced the partnership with Google DeepMind and Disney Research to develop an advanced physics engine. This collaboration is focused on achieving significant advancements in robot training and strengthens ties between leading tech companies and the robotics industry.
Applications of the Newton Physics Engine
The Newton physics engine, co-developed by the partners, aims to revolutionize robotics simulation. It utilizes Nvidia's Warp framework and incorporates Google's MuJoCo-Warp to significantly enhance performance. The engine is expected to accelerate humanoid simulations by over 70 times and in-hand manipulation tasks by 100 times. Plans are underway to integrate Newton into Nvidia's ISAAC simulator, with demonstrations of Disney's robots using the technology planned for 2026.
Future Prospects and Expectations
The reaction from industry participants has been optimistic. Kyle Laughlin of Disney noted that Newton will help create more expressive robots. Jensen Huang remarked that the engine is designed to help developers simulate how robots interact with the natural world, which can be a challenge for robotics developers. It’s anticipated that Newton will speed up robotics development cycles and broaden applications in sectors such as entertainment, manufacturing, and research. The release is scheduled for late 2025.
The collaboration between Nvidia, Google DeepMind, and Disney Research signifies a major step towards improving robot interaction with the environment, opening new horizons in robotics for the coming years.
