This repository provides a demonstration of Deep Q-Learning using TensorFlow for reinforcement learning. It's targeted at researchers and developers interested in understanding and implementing Q-learning algorithms, offering a flexible framework for creating custom controllers and simulations.

How It Works

The project utilizes a modular design, separating controllers and simulators. Controllers define how an agent acts based on observations and store transition data for learning. The simulate function orchestrates the interaction between controllers and simulators, allowing for custom agent behaviors and game environments.

Quick Start & Requirements

• Install using pip install tensorflow-deepq.
• Requires future==0.15.2, euclid==0.1, and inkscape for GIF creation.
• A local Redis server is needed for the HumanController.

Highlighted Details

• Demonstrates a learned strategy for a game with accelerating actions.
• Includes a HumanController for interactive gameplay.
• Provides a framework for defining custom controllers and simulations.
• Supports GIF generation from simulation runs.

Maintenance & Community

The project is marked as obsolete, with a recommendation to use openai/baselines instead. No community links or active maintenance signals are present.

Licensing & Compatibility

The license is not explicitly stated in the README.

Limitations & Caveats

This repository is explicitly marked as obsolete and superseded by a newer implementation. It may not be actively maintained or compatible with current TensorFlow versions.