Back to Gap Map Model components / Robotics & embodied AI

Isaac Sim

NVIDIA
open source / Overall score: 2.8

Isaac Sim is NVIDIA's robotics simulation platform, built on Omniverse, for developing, testing, training and deploying robots in simulated environments. It imports robots from URDF, MJCF and CAD, simulates them with GPU-accelerated physics, renders RTX cameras, lidar and radar alongside physics-based sensors, and provides synthetic data generation, a ROS bridge and the simulation that Isaac Lab trains on.

Building or running Isaac Sim needs the Omniverse Kit SDK and NVIDIA's 3D assets, which are proprietary and come under a separate NVIDIA license. The pip packages install prebuilt binaries under that license too.

Openness

5 medium confidence
5.0
license
Apache-2.0 (OSI, the code in the repository)
source
public (builds from the repo, but the build downloads the proprietary Omniverse Kit SDK and NVIDIA 3D assets it needs to run)
core features withheld
no — the proprietary components are free to use, not a paid tier

NVIDIA publishes Isaac Sim's own code under Apache-2.0 and it builds from the repository. Building or running it still needs the Omniverse Kit SDK, 3D models and textures, which NVIDIA licenses separately for use on NVIDIA GPUs only and forbids modifying or redistributing; they cost nothing, so nothing is held back for a paid tier.

Adoption

1 low confidence
1.0

PyPI downloads of the isaacsim metapackage, which a pip install of Isaac Sim pulls before fetching the rest from NVIDIA's own package index. Workstation downloads, containers and source builds do not pass through it and are not counted.

Capability

4 medium confidence
4.0

Isaac Sim provides the GPU physics and RTX-rendered sensors that the strongest simulators here are built from, and it can run its scenes on the Newton engine as well as PhysX. The library of ready-to-train robot tasks lives in Isaac Lab, which is built on it, so Isaac Sim sits a step below.

  • https://docs.isaacsim.omniverse.nvidia.com/latest/sensors/index.html recorded 2026-11-12

    Documentation navigation lists "Newton Physics Backend" under physics simulation, "RTX Lidar Sensor", "RTX Radar Sensor", "RTX Acoustic Sensor", "Camera Sensors", "Depth Sensors", and physics-based "Contact Sensor", "IMU Sensor", "Joint State Sensor".

  • https://raw.githubusercontent.com/isaac-sim/IsaacSim/release/6.3/README.md recorded 2026-11-12

    README: "The simulator leverages high-fidelity, GPU-accelerated physics engines to simulate accurate dynamics and support multi-sensor RTX rendering at scale. It comes equipped with end-to-end workflows including synthetic data generation, reinforcement learning, ROS integration"; Isaac Lab listed under "Key Applications".

Verified 2026-11-12