unity-ml-drl-data is a GitHub repository for experienting with Deep Reinforcement Learning (DRL) using Unity and ML-Agents. This project uses simulated 3D enviroments to study and train agents with DRL algorithms, while logging performance and behavioral data using TensorBoard for analysis using Machine Learning Techniques.

unity-ml-drl-data/
│
├── unity/                 # Unity project files (scenes, agents, environment scripts)
├── training/              # Python training scripts, configs, and utilities
├── data/                  # Collected data and schema definitions
├── docs/                  # Documentation, research notes, and reports
├── README.md              # This file
├── CONTRIBUTING.md        # Guidelines for making contributions
└── SETUP.md               # In-depth installation steps

See SETUP.md for in-depth installation steps.

See training/README.md for training instructions.

See data/ for results.

Unity side: ML-Agents 2.0.1 (installed automatically via Unity Package Manager)

Python side: Dennis Soemers’ ML-Agents fork (see training/requirements.*.txt)

This project includes the official Unity ML-Agents Examples and corrosponding training configuration files, sourced from the Unity ML-Agents GitHub Repository.

All rights to these examples belong to Unity Technologies. We claim no ownership over them.

The Unity ML-Agents Examples included here remain under their original Apache License, Version 2.0, as provided by Unity Technologies. All other code and assets created for this repository are licensed under the terms specified in this project’s LICENSE file.