Wumpus World – Dynamic Logic Agent (AI2002 Assignment 6)

A Knowledge-Based Agent that navigates a Wumpus World grid using Propositional Logic and Resolution Refutation.

Stack

• Backend: Python 3 + Flask (inference engine, KB, CNF conversion)
• Frontend: Vanilla HTML/CSS/JS (no framework needed)

Project Structure

wumpus/
├── backend/
│   ├── app.py            # Flask API + KB + Resolution engine
│   └── requirements.txt
└── frontend/
    ├── index.html
    ├── style.css
    └── app.js

How to Run

Backend

cd backend
pip install -r requirements.txt
python app.py

Server runs at http://localhost:5000

Frontend

Open frontend/index.html in your browser, or serve it:

cd frontend
python -m http.server 8080

Then open http://localhost:8080

Logic Engine

Knowledge Base (KB)

When the agent visits a cell and receives percepts:

• No Breeze → all neighbors are safe from pits → ¬P(r,c) for each neighbor
• Breeze → at least one neighbor has a pit → P(n1) ∨ P(n2) ∨ ...
• Same logic applies for Stench and Wumpus

CNF Conversion

All KB facts are stored directly in Conjunctive Normal Form:

• safe_pit(r,c) → [¬Pit(r,c)] (unit clause)
• or_pit(neighbors) → [Pit(n1) ∨ Pit(n2) ∨ ...]

Resolution Refutation

To prove a cell (r,c) is safe:

1. Negate the query: add [Pit(r,c)] to KB
2. Repeatedly resolve pairs of clauses
3. If the empty clause [] is derived → contradiction → cell IS safe
4. If no new clauses can be derived → cannot prove safety

API Endpoints

Method	Endpoint	Description
POST	/api/new_game	Start game with {rows, cols}
POST	/api/move	Move agent to {row, col}
POST	/api/auto_move	Agent picks best safe move
GET	/api/state	Get current game state

Deployment

• Backend: Deploy to Railway or Render
• Frontend: Deploy to Vercel or Netlify

Update the API constant in frontend/app.js to point to your deployed backend URL.

🌐 Live Demo

Frontend: https://wumpus-agent-614.vercel.app
Backend: https://wumpus-agent-614.onrender.com