Flappy-AI with Reinforcement Learning

This is a project that uses AI to train an agent to play a game. The project is structured as follows:

• ai/: Contains the code for the AI agent, including the Q network, replay buffer, and training logic.
• assets/: Contains any assets used by the project.
• configs/: Contains configuration files.
• environment.yml: Contains the list of dependencies for the project.
• experiments/: Contains any experimental code or data.
• game/: Contains the game logic and rendering code.
• model/: Contains the trained model.
• runs/: Contains the output from training runs.
• utils/: Contains utility code used across the project.
• train.py: The main script to start training the agent.
• test_game.py: The script to test the game without the agent.

Installation

1. Clone the repository:

git clone https://github.com/amc-madalin/flappy-ai.git

2. Navigate to the project directory:

cd flappy-ai

3. Install the dependencies:

conda env create -f environment.yml

Usage

Activate environment:

conda activate flappy-ai

To train the agent, run:

python train.py

To test the game, run:

python test_game.py

License

This project is licensed under the MIT License.