❄️ Frozen Lake Search Agents

Welcome to the Frozen Lake Search Agents repository!
Here you'll find intelligent agents that solve the classic Frozen Lake maze problem using search algorithms.

🧊 The Problem

Imagine a frozen lake represented as a grid, where each cell can be:

• S: Start position
• G: Goal position
• F: Frozen (safe to walk)
• H: Hole (danger! falls in)

Your agent must find the safest path from S to G without falling into any holes.
At each step, the agent can move:

• ⬅️ Left (0)
• ⬇️ Down (1)
• ➡️ Right (2)
• ⬆️ Up (3)

🤖 The Agents

This repo includes two search-based agents:

• UCSAgent: Uniform Cost Search agent (explores all paths with equal cost)
• AStarAgent: A* agent (uses smart heuristics to find the shortest path faster)
  • Manhattan distance heuristic 🏙️
  • Euclidean distance heuristic 📏

🏁 How It Works

• The environment is generated randomly for different grid sizes (4x4, 8x8, 16x16, 20x20).
• Agents use the environment's model to plan their moves.
• Performance (steps, reward, time) is compared for each agent and environment.

🚀 Getting Started

1. Install dependencies with Poetry
   If you don't have Poetry, install it first:

   curl -sSL https://install.python-poetry.org | python3 -
   

   Then install all dependencies:

   poetry install
   

2. Explore interactively
   Open frozen_lake.ipynb for step-by-step exploration and visualization.

📊 Results

After running, you'll see a comparison of UCS and A* agents on different maze sizes, showing:

• Reward earned 🏆
• Steps taken 👣
• Time spent ⏱️

📝 Files

• main.py — Runs the agent comparison
• ucs_agent.py — Uniform Cost Search agent
• a_star_agent.py — A* agent with heuristics
• model.py — Maze model and graph builder
• priority_queue.py — Priority queue for search
• frozen_lake.ipynb — Interactive notebook

---

Enjoy solving the Frozen Lake challenge! ❄️🧠🚶‍♂️