Isaac Lab
NVIDIAIsaac Lab is NVIDIA's GPU-accelerated framework for robot learning, covering reinforcement learning, imitation learning and motion planning. It ships more than thirty ready-to-train environments wired to RL libraries such as RSL RL, SKRL and Stable Baselines, runs on Isaac Sim's PhysX physics and RTX rendering, and is adding the open Newton engine as a second physics backend.
Isaac Lab needs Isaac Sim to run, and Isaac Sim's runtime components are proprietary NVIDIA software. Isaac Sim is a separate candidate and is not part of this entry.
Openness
5 medium confidence- license
- BSD-3-Clause(framework
- source
- public(the framework builds from the repo
- core features withheld
- no — no feature of the framework is withheld
The framework itself is BSD-3-Clause, with nothing set aside for a paid tier. It cannot run on its own, though: it sits on Isaac Sim, whose Omniverse Kit runtime and 3D assets are proprietary NVIDIA software, and the optional mimic extension needs the proprietary cuRobo.
- https://raw.githubusercontent.com/isaac-sim/IsaacLab/main/docs/licenses/dependencies/isaacsim-license.txt recorded 2026-09-27
Vendored Isaac Sim license: Isaac Sim's code is Apache 2.0, but "building or using the software requires additional components licenced under other terms ... the Omniverse Kit SDK, as well as 3D models and textures".
- https://raw.githubusercontent.com/isaac-sim/IsaacLab/main/LICENSE recorded 2026-09-27
LICENSE: "Copyright (c) 2022-2025, The Isaac Lab Project Developers ... SPDX-License-Identifier: BSD-3-Clause" followed by the BSD 3-Clause text.
- https://raw.githubusercontent.com/isaac-sim/IsaacLab/main/README.md recorded 2026-09-27
README License section: the framework is BSD-3, the isaaclab_mimic extension Apache 2.0; "Isaac Lab requires Isaac Sim, which includes components under proprietary licensing terms", and isaaclab_mimic requires cuRobo, "which has proprietary licensing terms".
- https://ungh.cc/repos/isaac-sim/IsaacLab/files/main recorded 2026-09-27
Full listing of the main branch, 1,993 paths; no enterprise, ee or commercial path.
Adoption
1 low confidencePyPI installs of the isaaclab package. The README installs Isaac Lab from a source checkout rather than from PyPI, so most users never touch the package and the figure undercounts use.
- https://pypi.org/pypi/isaaclab/json recorded 2026-09-27
PyPI metadata for isaaclab: Documentation https://isaac-sim.github.io/IsaacLab/, Home page https://developer.nvidia.com/isaac/lab; no repository URL.
- https://pypistats.org/api/packages/isaaclab/recent recorded 2026-09-27
pypistats: last_month 1,726 downloads for isaaclab.
Capability
5 medium confidenceIsaac Lab combines everything the other simulators here offer in part: parallel environments on the GPU, rendered cameras, LIDAR and contact sensors, a large set of ready-to-train tasks for common robots, and a second physics engine for checking a policy before it goes to hardware.
- https://raw.githubusercontent.com/isaac-sim/IsaacLab/main/docs/source/experimental-features/newton-physics-integration/index.rst recorded 2026-09-27
Newton integration docs: available on the develop branch as part of Isaac Lab 3.0 Beta, focused on the MuJoCo-Warp solver, with a limited set of locomotion examples.
- https://raw.githubusercontent.com/isaac-sim/IsaacLab/main/README.md recorded 2026-09-27
README: "a GPU-accelerated, open-source framework designed to unify and simplify robotics research workflows, such as reinforcement learning, imitation learning, and motion planning"; "more than 30 environments, which can be trained with popular reinforcement learning frameworks such as RSL RL, SKRL, RL Games, or Stable Baselines"; sensors "cameras, LIDAR, or contact sensors".
- https://raw.githubusercontent.com/isaac-sim/IsaacSim/main/README.md recorded 2026-09-27
Isaac Sim README: "high-fidelity, GPU-accelerated physics engines to simulate accurate dynamics and support multi-sensor RTX rendering at scale".
Verified 2026-09-27