Autonomous Navigation Agent

Pathfinding algorithm comparison with ML-based selector. Author: Ali Alperen Civil

---

Project Structure

autonomous_navigation/
├── environment.py    # 2D grid environment
├── algorithms.py     # BFS, DFS, A* implementations
├── agent.py          # Autonomous agent class
├── visualizer.py     # Matplotlib animations and charts
├── ml_selector.py    # ML-based algorithm selector
├── experiments.py    # Multi-map experiments and stress test
└── main.py           # Entry point

---

Requirements

Python 3.8+ and the following packages:

pip install -r requirements.txt

---

How to Run

# Single map demo with visualizations
python main.py single

# Multi-map experiment
python main.py all

# 20-trial stress test
python main.py stress

# ML selector experiment
python main.py ml

# Agent demo
python main.py agent

---

GIF Animations

Pre-generated search animations on the Maze map are included:

• BFS_maze.gif — BFS search (329 frames)
• DFS_maze.gif — DFS search (175 frames)
• Astar_maze.gif — A* search (322 frames)

To regenerate:

from environment import Environment
from algorithms import run_all
from visualizer import save_animation_gif

env = Environment.make_maze()
results = run_all(env)
for algo, r in results.items():
    save_animation_gif(env, r, f"{algo.replace('*','star')}_maze.gif", fps=30)