IsaacSim-IsaacLab-installation-for-Windows-Easy-Tutorial

• Video Tutorials showing the IsaacSim installation: steps 0-6:
  • Video 1: https://youtu.be/6NlJqvWBhg4
  • Video 2: https://youtu.be/E6tgh798AYU
• Video Tutorial showing the IsaacLab installation: steps 7-8
  • Video 3: https://youtu.be/8XecEL5LvPI

Ground rules

• Do not use WSL (Windows Subsystem for Linux) for Isaac Sim/Isaac Lab. Run everything in Windows Terminal or PowerShell.
• Use Python 3.11 for Isaac Sim 5.x. Other versions will fail.
• This tutorial uses IsaacSim 5.1, which works with Nvidia GPU Driver version 580/581. Anything much lower or above this (eg: 595 - 610...) will make IsaacSim crash due to compatibility issues.

-> This tutorial is inspired by the official tutorial (Isaac Sim: https://isaac-sim.github.io/IsaacLab/main/source/setup/installation/pip_installation.html) but it explains all the steps in a more beginner-friendly way.

IsaacSIM

Step 0 — Verify your GPU and driver

• Why: Isaac Sim 5.1 was tested on Windows driver 580/581. Different driver versions can fail or disable RTX features.

• How: on Windows PowerShell type nvidia-smi

• Check the Driver Version in the output. If it’s different from 580/581, update or downgrade via GeForce Experience or NVIDIA’s driver page https://www.nvidia.com/en-us/drivers. Then reboot.

  • Go to https://www.nvidia.com/en-us/drivers > select Manual Driver Search > input your computer specs > search for drivers 580/581 > download it > install it >
    • Choose NVIDIA Graphics Driver only
    • Choose Custom (Advanced)
    • You'll see a list of components with checkboxes. Leave them all checked, and at the bottom tick the box that says "Perform a clean installation."
    • reboot computer
  • After reboot, open Windows Powershell as admin (right-click on its icon and select "admin"), run command nvidia-smi. This should output the driver's specs and version should read 580/581

• Optional: on laptops with Intel + NVIDIA: force high‑performance GPU for Python and Isaac Sim in Settings → System → Display → Graphics by adding the IsaacSim app, clicking on "Add desktop app" and choosing "High performance" on settings. (Microsoft Learn)

Step 1 — Enable long file paths (Windows one‑time setup)

• Why: Long paths can break installs.

• How (on PowerShell as Admin): click on Windows search option, search for powershell, don't open it. Instead, click on the app with the right button and click on open as admin. Type:

• New-ItemProperty -Path "HKLM:\SYSTEM\CurrentControlSet\Control\FileSystem"

• -Name "LongPathsEnabled" -Value 1 -PropertyType DWORD -Force

Reboot if this is your first time enabling it.

Step 2 — Install the prerequisites

2.1 Git (for cloning Isaac Lab):

• Download and install Git for Windows or use winget: winget install --id Git.Git -e
• update pip install python -m pip install --upgrade pip

2.2 A Python env manager:

• Download “Miniconda3 Windows x86_64” and install it: https://www.anaconda.com/docs/getting-started/miniconda/install > https://www.anaconda.com/download. Accept defaults.

• to confirm installation, go to your Windows search bar, type in Anaconda Prompt, open it to spawn a fresh Anaconda CLI. In your (base) environment, type conda list. You should see the list of resources.

• official instructions. (Isaac Sim)

Step 3 - Create a clean Python 3.11 environment

• on a conda cli: click on the Windows search option, type “anaconda prompt”, click on it to open the cli

conda create -n env_isaaclab python=3.11 -y 
conda activate env_isaaclab 

Step 4 — Install Isaac Sim 5.1

• In the conda activated env:

pip install "isaacsim[all,extscache]==5.1.0" --extra-index-url https://pypi.nvidia.com

Source from the docs. (Isaac Sim)

Step 5 — Install PyTorch with CUDA for Windows

• In the same env:

pip install -U torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128

• Quick test:

python -c "import torch; print(torch.__version__, torch.cuda.is_available())"

If everything went as expected up to here:

• PyTorch installed correctly (import torch worked).
• CUDA runtime is active (otherwise it wouldn’t detect the GPU).
• Windows driver + PyTorch are communicating properly with your GPU. -> you don’t need to install the standalone CUDA Toolkit — the cu128 wheel already included the required runtime.

Step 6 — Validate Isaac Sim before running Isaac Lab

6.1 Run the Compatibility Checker app :

isaacsim isaacsim.exp.compatibility_check

Expected output (depends on your machine configuration)

• Driver Version (✅ green) your Windows NVIDIA driver is new enough and recognized.

• GPU: It will show your GPU version (✅ green) Isaac Sim can see and talk to your GPU correctly through Vulkan + CUDA.

• VRAM (light green) GPU memory detected and usable; “light green” just shows a normal range for mobile 4090s.

• RAM GB (🟧 orange) Informational warning — the check recommends ≥ 64 GB for very heavy RTX simulations, but 32–34 GB is still fine for learning, tutorials, and medium workloads. It’s not an error.

• CPU Governor “not available” Windows doesn’t expose Linux-style CPU governors; this field is only relevant on Linux. Safe to ignore.

If all these are checked, your environment (GPU, driver, Vulkan, CUDA, Isaac Sim core) are working correctly.

Close it after the checks.

6.2 Launch the full app once to pull extensions and accept the EULA:

Isaacsim

• On first launch, reply Yes to the Omniverse EULA prompt (one time)

• The first run can take several minutes while extensions download and cache. This is expected.

IsaacLab

-> Before proceeding with the next steps, make sure you activate the previously created conda environment and that you are inside your user folder.

Step 7 — Install Isaac Lab from source

7.1 Clone the repo:

. Clone the IsaacLab GitHub repository into your user folder. . What it does is simply copy NVIDIA's original IsaacLab project (with all its folders and Python scripts) to your local folder inside your environment. . Make sure you are inside your user path (where you had previously installed IsaacSim).

cd %userprofile%
git clone https://github.com/isaac-sim/IsaacLab.git 
cd IsaacLab

Source: (Isaac Sim)

7.2 Install Isaac Lab and libraries:

. Inside [YourUser]\IsaacLab:

isaaclab.bat --install

. isaaclab.bat --install registers the cloned IsaacLab folders and files as a Python library, so that Python can find them and you can import them into your local code with import isaaclab.

• Install auxiliary libraries in their correct versions - compatible with IsaacSim 5.1:

pip install h5py==3.15.1 tensordict==0.11.0 pandas==3.0.1 click==8.1.7

Step 8 — Smoke tests for Isaac Lab

8.1 Minimal sim window:

isaaclab.bat -p scripts\tutorials\00_sim\create_empty.py

• A black viewport window should appear. Close with Ctrl+C or stop in the terminal.

Source: (Isaac Sim)

Expected outputs

• it opens the isaacsim application window showing just a black screen on the view port.
• anaconda prompt logs :

[22.704s] app ready
[27.287s] Simulation App Startup Complete
[27.386s] [ext: omni.physx.fabric-107.3.26] startup
[INFO]: Setup complete...

8.2 Run Quick RL example

Navigate to C:\Users\[YOUR USER]\IsaacLab -> This will run template tasks that already come with the IsaacLab project folder (Internal Projects)

Run it headless: It will run the RL script but not open the IsaacSim application

isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task=Isaac-Ant-v0 --headless

It will run the RL training (1000 iterations). You can stop it with Ctrl+c

Run it full: It will run the RL script AND open the IsaacSim application showind the progress there. Takes longer

isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task=Isaac-Ant-v0