Q-Drive - Self Play Car Game Using Deep Q Learning

Overview

bandicam.2024-07-06.21-07-57-872.mp4

Reinforcement Learning (RL) stands at the forefront of artificial intelligence, enabling the creation of intelligent agents capable of learning through interaction with their environment. Among the various RL techniques, Deep Q-Learning (DQN) has gained significant traction. In DQN, a neural network is trained to make decisions by optimizing a Q-function, which is iteratively improved based on the states and rewards received from the environment.

Deep Q-Learning has been successfully applied to various applications, including mastering popular Atari games. Inspired by these advancements, this project aims to experiment with a car game, exploring whether a DQN agent can autonomously learn to play the game.

Project Goals

• Develop a DQN agent capable of understanding and playing a car game.
• Experiment with different neural network architectures and hyperparameters.
• Analyze the learning process and performance of the agent.
• Let the agent figure out which one to collect and which one to avoid without giving an instructions.

Features

• Interactive environment for training and testing the DQN agent.
• Visualization of agent performance and learning progress.
• Configurable parameters for experimenting with various DQN setups.

Getting Started

Prerequisites

• Python 3.x
• PyTorch
• OpenCV-Python
• NumPy
• Pygame

Installation

Clone the repository:

git clone https://github.com/sidharth72/Q-Drive.git
cd Q-Drive

Install the Requirements

pip install -r requirements.txt

Run the agent

python agent.py

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

Contributions

Contributions are welcome! Please feel free to submit a Pull Request or open an issue to discuss improvements and new features.