Block-Reign — Player vs AI Grid Battle 🕹️🤖

Overview

Block-Reign is a lightweight Player vs AI grid battle game where the AI learns from every match you play. The project includes a simple Q-learning AI (used by the game server) and a more advanced DQN trainer for experiments.

---

🔑 Key Features

• 🎮 10×10 grid combat with movement and shooting
• 🤖 Self-learning AI that adapts from actual matches you play
• 💾 Persistent training (saved automatically under training/models)
• 🧠 Advanced DQN trainer available in ai_trainer.py for experimentation
• 🛠️ Easy reset — delete training/ to get a fresh (untrained) AI

---

Current AI Progress ✅

• Win rate: 59%
• Latest match log: 🏁 Match 78 complete - Winner: ai
• Recent learning result: ✅ AI learned from match. Wins: 46, Losses: 36

These numbers reflect the Q-learning AI’s current record (persisted in training/models/simple_ai.pkl).

---

Quick Start — Play Locally (Windows / macOS / Linux)

1. Install Python 3.8+ and pip.
2. (Optional) Create and activate a virtual environment:

python3 -m venv .venv
source .venv/bin/activate  # macOS/Linux
.venv\Scripts\activate     # Windows

1. Install dependencies:

pip install -r requirements.txt

1. Run the game server:

python3 game_server.py

1. Open your browser at: http://localhost:5000

• Controls: Arrow Keys = Move, Space = Shoot

---

How Learning Works 💡

• The running server uses a simple Q-learning AI (SimpleAI in game_server.py) that stores a Q-table and tracks wins/losses.
• After each completed match, the client POSTs match data to /learn_from_match and the AI updates its policy on-disk.
• The project includes ai_trainer.py (a DQN-based trainer) for more advanced experiments and batch training.

---

Resetting to an Untrained AI 🔄

If you want to start over and play against a brand-new AI:

• Delete the entire training/ folder (destructive):

rm -rf training/

• Or remove only the saved model:

rm training/models/simple_ai.pkl

After removal, start the server and play matches — the AI will begin learning from scratch as you play.

⚠️ Deleting training/ will remove all saved models, replay buffers and saved checkpoints. Make backups if needed.

---

Development & Troubleshooting 🔧

• Logs and training data are saved under training/models.
• To inspect AI stats during runtime, look for prints from SimpleAI at server startup and printed match summaries.
• If models fail to load, ensure training/models exists and is writable by your user.

---

Contributing

Contributions welcome! Fork, make changes, and open a PR. If you’re adding features, please include tests and update this README.

---

License

This project is licensed under the project LICENSE in the repository.

---

Enjoy playing — and see how quickly the AI learns from your style! 🎯