Playing FlappyBird with AI

Program Members

Xagen Mclaughlin, 6th grade student at KST, 3rd place winner in Louisiana State Science & Engineering Fair

Dr. Hao Wang, Advisor at LSU

Mr. Elkhan Akhundov, Advisor at KST

Objectives

We develop smart AIs to play the classic game FlappyBird. Through this project, the participating student is expected to learn and experience:

• How AI understands the real world using computer vision techniques
• How AI makes decisions to achieve the highest score in games
• How to understand AI’s behaviors and performance
• What knowledge we need to develop better AIs

Roadmap

1. Setup the game environment
2. Learn about basic reinforcement learning (RL) strategies
3. Deploy a simple RL-based AI to play the game
4. Analyze the AI’s performance and explain its behaviors
5. Develop a new AI to improve its performance
6. Compare and analyze different AIs’ performance
7. Project demo and competition between AI and human

FlappyBird played by Reinforcement Learning.