isaac-sim/IsaacSim

NVIDIA Isaac Sim™ is an open-source application on NVIDIA Omniverse for developing, simulating, and testing AI-driven robots in realistic virtual environments.

What is Isaac Sim?

Isaac Sim is NVIDIA’s open‑source simulation platform built on the Omniverse framework. It lets you create realistic virtual worlds for AI‑powered robots – from industrial arms to mobile platforms – and run them on a GPU‑accelerated physics engine. The simulator can import robot models (URDF, MJCF, CAD), render photorealistic sensor data with RTX, and expose the whole stack to Python, ROS 2, or custom reinforcement‑learning policies.

Core capabilities

Capability What it means for a robotics developer
Asset import & export Bring existing robot descriptions (URDF/MJCF/CAD) into the USD‑based Omniverse scene and export back if needed.
Robot tuning Adjust physics parameters, collision shapes, and visual fidelity to trade off accuracy vs. speed.
Simulation & control Built‑in controllers, motion‑generation tools, kinematics solvers, and a hook for plugging in learned policies.
Sensors Simulated cameras, LiDAR, depth, IMU, etc., rendered with RTX for realistic noise and lighting.
Synthetic data generation Record sensor streams automatically for training perception models.
Reinforcement‑learning framework (Isaac Lab) GPU‑scaled RL, imitation learning, and motion‑planning pipelines that run directly on the simulated robot.
ROS 2 bridge Seamlessly connect the simulation to ROS 2 nodes, enabling the same code to run on real hardware.

Who would use it?

  • Robotics researchers building and testing new control algorithms or perception pipelines.
  • AI engineers needing massive synthetic datasets for training vision or navigation models.
  • Product teams prototyping robot software before hardware is available.
  • Educators who want a hands‑on, visual environment for teaching robot dynamics and AI.

Getting started (high‑level)

  1. Set up a supported machine – Windows 11 or Ubuntu 22.04/24.04 with an RTX‑class GPU (minimum RTX 4080/​A40). Install the required driver, Git LFS, and a C++ toolchain (VS 2022 on Windows, GCC 11 on Linux).
  2. Clone and build the repo using the provided build.sh/build.bat. The first launch downloads the Omniverse Kit SDK and extensions.
  3. Run the isaac‑sim.sh (Linux) or isaac‑sim.bat (Windows) executable. The UI opens in Omniverse Kit where you can load example scenes or import your own.
  4. Explore tutorials – quick‑start guides, asset libraries, and the Isaac Lab notebooks walk you through RL training, ROS integration, and synthetic data pipelines.

Packaging options

  • Standalone binary – a self‑contained archive for distribution.
  • Python wheels – install the simulator into a Python 3.12 virtual environment (pip install *.whl).
  • Docker image – containerised build for cloud or CI environments.

Community & support

  • Use GitHub Discussions for questions and ideas.
  • File issues for bugs or concrete feature requests.
  • The repository is Apache‑2.0 licensed; contributions are limited but accepted via a contribution guide.

All details are taken directly from the repository’s README; no additional features are inferred.

Related

  • Project
    isaac-sim/IsaacLab

    Isaac Lab is a GPU-accelerated, open-source robotics simulation framework built on NVIDIA Isaac Sim, designed to simplify and scale robot learning tasks like reinforcement learning and motion planning.

  • Project
    umfieldrobotics/OceanSim

    A GPU-accelerated underwater robot perception simulation framework that uses physics-based rendering to model visual and acoustic sensors and reduce the simulation-to-real gap.

  • Project
    isaac-sim/IsaacLab-Arena

    Isaac Lab‑Arena is an open‑source, alpha‑stage framework that sits on top of NVIDIA Isaac Lab/Sim. It lets you compose robot simulation environments from reusable scene, robot (embodiment), and task blocks, then run massive parallel policy evaluations on a GPU or across nodes, collecting fine‑grained metrics and sensitivity analyses. Features include relational object placement, sequential task chaining, experimental natural‑language‑driven environment generation, and a client‑server policy API. Installation uses the `uv` Python manager or provided Docker scripts, and the library is Python‑centric (≥ 3.12, Linux).

  • Project
    AgibotTech/genie_sim

    Genie Sim is a comprehensive, agent-friendly simulation platform for embodied AI that provides tools for scene reconstruction, large-scale synthetic data generation, and standardized robot performance evaluation.

  • Project
    abizovnuralem/go2_omniverse

    A high-fidelity digital twin project for Unitree Go2 and G1 robots in Nvidia Isaac Sim, enabling real-time synchronization between physical robots and their virtual counterparts.