A hands-on project combining classic arcade gameplay with the power of Reinforcement Learning (RL).
This setup demonstrates how an AI agent can learn to master a Pong game through trial-and-error, rewards, and intelligent adaptation.
Built using PhaserJS, this lightweight browser-based game simulates a simplified Pong environment, ideal for testing RL algorithms and observing their performance in real time.
The backend is developed using Python 3.12.2 and Flask 3.1.0, with custom support for WebSockets to allow real-time communication between the AI agent and the game environment.
| Component | Technology |
|---|---|
| Game Engine | PhaserJS |
| Frontend Build | Vite + Node.js |
| Backend Server | Flask 3.1.0 (with WebSockets) |
| AI Framework | PyTorch |
| Language | Python (AI/backend) + JavaScript (frontend) |
- ✅ Live RL agent playing Pong via WebSocket communication
- 🔄 Training feedback dashboard (real-time graphs or logs)
- 🧠 Model saving/loading with PyTorch
- 📈 Performance tracking over episodes
Contributions, ideas, and feedback are always welcome!
Feel free to fork the repo, open issues, or create pull requests.
This project is licensed under the MIT License.