Forging the Tools

Phase 1: The Simulation Engine (Unity)

We use Unity 2022 LTS (Long Term Support) for stability.

Step 1.1: Install Unity Hub

Download and install Unity Hub from the official site. This manages your Unity editors and projects.

Note: When running the installer, you can accept all default installation settings (install location, etc.).

Once installed, open Unity Hub. You should see a Welcome screen:

If you are new to Unity, you will need to click Create account. This is free and required to use the software. Follow the on-screen prompts to verify your email and sign in.

Once signed in, you will see the main dashboard:

Step 1.2: Install Unity Editor 2022.3 LTS

1. Open Unity Hub.
2. Go to the Installs tab > Install Editor.
3. Select 2022.3.x LTS (The exact 'x' version doesn't matter, e.g. 2022.3.10f1).
4. IMPORTANT: Ensure "Microsoft Visual Studio Community 2022" (Windows) or "Visual Studio for Mac" is checked in the modules list. This is your code editor.
5. Click Install. This will take 15–30 minutes.

Phase 2: The Neural Brain (Python)

We need Python 3.10.12 specifically, as ML-Agents is sensitive to Python versions.

Step 2.1: Monitor & Python

Open your terminal (Command Prompt on Windows, Terminal on Mac/Linux).

python --version

(Option A: Recommended) Anaconda / Miniconda

1. Download Miniconda.
2. Install and open "Anaconda Prompt" (Windows) or Terminal.
3. Create a clean environment:

   conda create -n agents python=3.10.12
   conda activate agents

Phase 3: The Bridge (ML-Agents)

This software connects Unity (C#) to Python (PyTorch). We will install version 1.1.0 (Release 21).

Step 3.1: Install PyTorch

Visit PyTorch.org to get the command for your system. Standard CPU command:

pip3 install torch torchvision torchaudio

Step 3.2: Install Unity ML-Agents

Run these commands in your agents environment:

python -m pip install mlagents==1.1.0
python -m pip install graphics

Step 3.3: Verify Installation

Run the help command to ensure the CLI is ready:

mlagents-learn --help

✅ Success: You should see a cool ASCII art logo of the Unity cube and a list of commands.

❌ Failure: If you see "command not found," ensure your environment is activated.

Phase 4: Workshop Assets

Clone the starter repository which contains the base environment art.

git clone https://github.com/sounny/AIMLAgentsWorkshop.git