Repolode
Submit
newton-physics

newton-physics/newton

A GPU-accelerated physics simulation engine built on NVIDIA Warp, aimed at roboticists and simulation researchers.

View on GitHubWebsite
Stars
5,900
Forks
759
Open issues
429
Last commit
23 hours ago

What is it?

Newton is a physics simulation engine that runs on the GPU, built on NVIDIA Warp. It extends the deprecated warp.sim module and uses MuJoCo Warp as its primary backend. The project focuses on GPU computation, OpenUSD support, differentiability, and user-defined extensions for robotics simulation.

The project is a Linux Foundation project initiated by Disney Research, Google DeepMind, and NVIDIA. It can be installed with pip as the newton package, and a built-in examples module provides a large set of runnable scenes covering robots, cables, cloth, tendons, MPM materials, sensors, inverse kinematics, and differentiable simulation.

Supported platforms are Linux (x86-64 and aarch64), Windows (x86-64), and macOS on CPU only. CUDA acceleration requires an NVIDIA GPU with a compatible driver and Warp version.

Key features

  • 01Runs physics computation on the GPU via NVIDIA Warp.
  • 02Uses MuJoCo Warp as its primary simulation backend.
  • 03Supports OpenUSD for scene description.
  • 04Provides differentiable simulation and user-defined extensions.
  • 05Includes a runnable example gallery covering robots, cloth, cables, and MPM materials.

Use cases

  • Simulating robot locomotion and manipulation policies before deploying to hardware.
  • Running large batches of parallel physics simulations on GPUs for research.
  • Prototyping differentiable physics for gradient-based control or design optimization.

Quick start

  1. 1

    Install Newton with the examples extra

    pip install "newton[examples]"
  2. 2

    Run the bundled examples

    python -m newton.examples

Things to know

  • Requires Python 3.10+ and, for GPU acceleration, an NVIDIA Turing or newer GPU with driver R580+ (older drivers need an older Warp version).
  • macOS runs on CPU only.

In collections

Run and train AI on GPUsRobotics & embodied AI projects

Topics

#physics-simulation#robotics#gpu#nvidia-warp#simulation#differentiable-simulation#openusd#newton-physics

FAQ

What is newton?
A GPU-accelerated physics simulation engine built on NVIDIA Warp, aimed at roboticists and simulation researchers. Newton is a physics simulation engine that runs on the GPU, built on NVIDIA Warp. It extends the deprecated warp.sim module and uses MuJoCo Warp as its primary backend. The project focuses on GPU computation, OpenUSD support, differentiability, and user-defined extensions for robotics simulation.
Is newton free to use? What license does it have?
newton is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is newton still maintained?
Actively maintained. The last commit to newton was on Nov 6, 2026.
What can I use newton for?
Simulating robot locomotion and manipulation policies before deploying to hardware. Running large batches of parallel physics simulations on GPUs for research. Prototyping differentiable physics for gradient-based control or design optimization.
How do I install newton?
Install Newton with the examples extra: pip install "newton[examples]" Run the bundled examples: python -m newton.examples
Where can I get newton?
The source code and installation instructions are on GitHub: https://github.com/newton-physics/newton

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects