AI-Agent-Game

How to Use AI-Agent-Framework

Overview

The AI-Agent-Framework provides a flexible framework for training and playing games using AI agents. It follows a simple architecture where the game, agent, and model components interact with each other to train and play games.

Prerequisites

Before using the AI-Agent-Framework, ensure you have Python 3.6 or higher installed. You'll also need to install the required packages listed in the requirements.txt file using pip install -r requirements.txt.

Implementing the AIAgentGame Interface

1. Import the AIAgentGame class: Start by importing the AIAgentGame class from ai_agent.game.

2. Create a Game Class: Create a subclass of AIAgentGame for your game. This class will implement the game logic and interactions with the AI agent.

3. Implement Required Methods:

   • get_state(): Return the current state of the game as a list or numpy array.
   • play_step(action): Take an action (input) and return a tuple (reward, game_over, score).
   • reset(): Reset the game to its initial state.

Example implementation for any game:

from ai_agent.game import AIAgentGame

class MyGame(AIAgentGame):
    def __init__(self):
        super().__init__(input_size=YOUR_STATE_SIZE, output_size=YOUR_ACTION_SIZE)
        # Your game initialization code here

    def get_state(self):
        # Your get_state implementation here

    def play_step(self, action):
        # Your play_step implementation here

    def reset(self):
        # Your reset implementation here

Training the AI Agent

1. Import the Agent class: Import the Agent class from ai_agent.

2. Create an Agent Instance: Create an instance of the Agent class, passing your game instance to its constructor.

3. Train the Agent: Call the train method of the Agent instance to start training the AI agent.

Example training code:

from ai_agent import Agent

def main():
    game = MyGame()
    agent = Agent(game)
    agent.train(model_path="model.pth")

if __name__ == "__main__":
    main()

Playing the Game with the AI Agent

1. Create an Agent Instance: Create an instance of the Agent class, passing your game instance to its constructor.

2. Play the Game: Call the play method of the Agent instance to play the game with the trained AI agent.

Example playing code:

from ai_agent import Agent

def main():
    game = MyGame()
    agent = Agent(game)
    agent.play(model_path="model.pth")

if __name__ == "__main__":
    main()

Follow these steps to integrate the AI-Agent-Framework into your game and train/play with your AI agent. Remember, the specific game implementation details don't matter as long as you implement the required functions correctly. The input_size and output_size in the super().__init__ should match the size of the state returned by get_state and the actions received by play_step, respectively.

GUI for the AI Agent with 3 examples

Images

Main GUI Window

Snake Game Window

Flappy Game Window

Dino Game Window