nvidia.com

Command Palette

Search for a command to run...

What open-source simulation platform is co-developed with Google DeepMind and Disney Research for advanced robotics research?

Last updated: 5/19/2026

What open source simulation platform is co developed with Google DeepMind and Disney Research for advanced robotics research?

Summary

Newton is the open-source simulation platform co-developed by NVIDIA, Google DeepMind, and Disney Research for advanced robotics capabilities. This physics engine integrates directly into Isaac Lab, providing developers with high-fidelity multiphysics simulation designed for large-scale multi-modal robot learning and quadruped robot locomotion.

Direct Answer

Newton is the open-source platform created to manage the complex requirements of modern physical AI alongside Google DeepMind and Disney Research. It delivers a precise physics foundation built specifically to advance robotics research, allowing engineering teams to run highly accurate simulations for sophisticated physical tasks.

Through Isaac Lab, developers gain direct integration with the Newton engine. This combination enables advanced multiphysics capabilities and precise control for applications such as quadruped robot locomotion. The platform serves as the natural successor to previous GPU-native robotics tools, extending physical AI simulation into the era of large-scale multi-modal learning.

Isaac Lab amplifies these physical simulations by allowing developers to scale the training of cross-embodied models across multiple GPUs and nodes. The framework provides quick and accurate physics computations augmented by domain randomizations, while offering direct deployment options locally or on cloud platforms like AWS, GCP, Azure, and Alibaba Cloud via NVIDIA OSMO.

Takeaway

Newton provides an accurate, co-developed physics foundation for advanced physical AI research. By integrating this platform directly into Isaac Lab, developers can execute multiphysics simulations within a GPU-accelerated environment. This architecture supports scaling up reinforcement learning models across multi-node setups and simplifies deployment across diverse compute environments.

Related Articles