Quickstart

Quickstart#

Isaac Lab is a GPU-accelerated framework for robot learning built on vectorized simulation. Environments run thousands of parallel copies on the GPU, and a modular manager design lets you swap robots, sensors, and controllers without rewriting task logic.

This page gets you installed and running a first training job in minutes. For deeper topics (configurations, project scaffolding, standalone apps, preset catalogs), see Quickstart Details.

Install#

Clone Isaac Lab, create a Python 3.12 environment, and install. Choose the path that matches your workflow:

Fastest start — Newton physics only, no Isaac Sim download required.

# Install uv (https://docs.astral.sh/uv/getting-started/installation/)
curl -LsSf https://astral.sh/uv/install.sh | sh

git clone https://github.com/isaac-sim/IsaacLab.git --branch develop
cd IsaacLab

uv venv --python 3.12 --seed env_isaaclab
source env_isaaclab/bin/activate
./isaaclab.sh -i

:: Install uv: https://docs.astral.sh/uv/getting-started/installation/

git clone https://github.com/isaac-sim/IsaacLab.git --branch develop
cd IsaacLab

uv venv --python 3.12 --seed env_isaaclab
env_isaaclab\Scripts\activate
isaaclab.bat -i

See Modularized Installation for install tokens and Kit-less Installation for feature availability without Isaac Sim.

Recommended for PhysX, RTX rendering, ROS, URDF/MJCF importers, and the Kit visualizer.

git clone https://github.com/isaac-sim/IsaacLab.git --branch develop
cd IsaacLab

uv venv --python 3.12 --seed env_isaaclab
source env_isaaclab/bin/activate
uv pip install --upgrade pip
uv pip install "isaacsim[all,extscache]==6.0.0.1" \
  --extra-index-url https://pypi.nvidia.com \
  --index-strategy unsafe-best-match --prerelease=allow
uv pip install -U torch==2.10.0 torchvision==0.25.0 \
  --index-url https://download.pytorch.org/whl/cu128
./isaaclab.sh -i

:: Install uv: https://docs.astral.sh/uv/getting-started/installation/

git clone https://github.com/isaac-sim/IsaacLab.git --branch develop
cd IsaacLab

uv venv --python 3.12 --seed env_isaaclab
env_isaaclab\Scripts\activate
uv pip install --upgrade pip
uv pip install "isaacsim[all,extscache]==6.0.0.1" ^
  --extra-index-url https://pypi.nvidia.com ^
  --index-strategy unsafe-best-match --prerelease=allow
uv pip install -U torch==2.10.0 torchvision==0.25.0 ^
  --index-url https://download.pytorch.org/whl/cu128
isaaclab.bat -i

On Linux aarch64 (DGX Spark), use cu130 for PyTorch and see Local Installation for additional setup notes.

For conda, binary installs, Docker, and troubleshooting, see Local Installation.

Run Training#

Use the reinforcement learning training command with a task name and physics=, renderer=, and presets= to select backends and task-specific options:

# Kit-less: Newton MJWarp physics + Newton visualizer
./isaaclab.sh train --rl_library rsl_rl \
  --task=Isaac-Cartpole-Direct \
  --num_envs=16 --max_iterations=10 \
  physics=newton_mjwarp --visualizer newton

# With Isaac Sim: PhysX physics (default renderer)
./isaaclab.sh train --rl_library rsl_rl \
  --task=Isaac-Cartpole-Direct \
  --num_envs=4096 \
  physics=physx

# Camera task: typed physics + renderer + domain preset
./isaaclab.sh train --rl_library rsl_rl \
  --task=Isaac-Cartpole-Camera-Direct \
  physics=newton_mjwarp renderer=newton_renderer presets=rgb

isaaclab.bat train --rl_library rsl_rl ^
  --task=Isaac-Cartpole-Direct ^
  --num_envs=16 --max_iterations=10 ^
  physics=newton_mjwarp --visualizer newton

isaaclab.bat train --rl_library rsl_rl ^
  --task=Isaac-Cartpole-Direct ^
  --num_envs=4096 ^
  physics=physx

For commands that do not configure a visualizer, omit --visualizer / --viz to run without a viewer. Use --help on any script to see task-specific physics=, renderer=, and presets= options.

See also

Next Steps#

  • List registered environments: python scripts/environments/list_envs.py

  • Scaffold a new project: ./isaaclab.sh --new (Linux) or isaaclab.bat --new (Windows)

  • Walk through tutorials: Tutorials

  • Browse all environments: Available Environments