Repolode
Submit
Genesis-Embodied-AI

Genesis-Embodied-AI/genesis-world

A simulation platform for robotics and embodied AI combining a multi-physics engine, a photo-realistic renderer, and a cross-platform compiler behind a Python interface.

Open-source alternative to:NVIDIA Isaac SimMuJoCo

View on GitHubWebsite
Stars
30,094
Forks
2,892
Open issues
159
Last commit
2 days ago

What is it?

Genesis World is a simulation platform for physical AI development. It pairs a unified multi-physics engine with a photo-realistic renderer (Nyx) and a cross-platform compiler (Quadrants), exposed through a Python simulation interface. The project was previously named Genesis and is now officially supported by Genesis AI.

The physics layer handles rigid bodies, FEM, MPM, particle methods (PBD and SPH), and an IPC backend, all sharing one scene and state, with couplers between solvers. Assets can be loaded from formats such as URDF, MJCF, OBJ, GLB and USD, and the interface provides controllers, sensors (camera, lidar, IMU, contact force, tactile), parallel and heterogeneous environments, and a built-in GUI.

The compiler layer lowers Python kernel code to CUDA, AMD ROCm, Apple Metal, Vulkan, x86 and ARM64, so the same simulation code can target laptops or datacenter GPUs.

Key features

  • 01Unified rigid, FEM, MPM, and particle solvers sharing one scene.
  • 02Loads URDF, MJCF, OBJ, GLB, and USD assets.
  • 03Camera, lidar, IMU, contact force, and tactile sensors.
  • 04Parallel and heterogeneous environments for batched simulation.
  • 05Compiles to CUDA, ROCm, Metal, Vulkan, x86, and ARM64.

Use cases

  • Training robot manipulation or locomotion policies in simulation.
  • Generating synthetic data with rendered camera and sensor output.
  • Prototyping soft-body, cloth, or fluid interactions with robots.
  • Embedding a physics simulator inside research code.

Quick start

  1. 1

    Clone the repository from GitHub

  2. 2

    Install the package in editable mode with dev extras from the project folder

    pip install -e ".[dev]"
  3. 3

    Run an example script, such as the Franka cube demo

    python examples/rigid/franka_cube.py

Things to know

  • The README's install section was not included in the provided text, so the setup steps above come from the catalogue section.
  • Demos for the IPC and Nyx renderer need optional extras that must be installed separately.

Topics

#robotics#simulation#embodied-ai#physics-engine#rendering#python#gpu

FAQ

What is genesis-world?
A simulation platform for robotics and embodied AI combining a multi-physics engine, a photo-realistic renderer, and a cross-platform compiler behind a Python interface. Genesis World is a simulation platform for physical AI development. It pairs a unified multi-physics engine with a photo-realistic renderer (Nyx) and a cross-platform compiler (Quadrants), exposed through a Python simulation interface. The project was previously named Genesis and is now officially supported by Genesis AI.
Is genesis-world free to use? What license does it have?
genesis-world is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is genesis-world still maintained?
Actively maintained. The last commit to genesis-world was on Oct 19, 2026.
What can I use genesis-world for?
Training robot manipulation or locomotion policies in simulation. Generating synthetic data with rendered camera and sensor output. Prototyping soft-body, cloth, or fluid interactions with robots. Embedding a physics simulator inside research code.
How do I install genesis-world?
Clone the repository from GitHub Install the package in editable mode with dev extras from the project folder: pip install -e ".[dev]" Run an example script, such as the Franka cube demo: python examples/rigid/franka_cube.py
Where can I get genesis-world?
The source code and installation instructions are on GitHub: https://github.com/Genesis-Embodied-AI/genesis-world

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

Related projects