breaktoprotect / qlearn-plays-pitfalls

A simple q-learning agent (using Q-table) via reinforcement training learn to play a simple self-made game called Pitfalls.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Q-learning Agent plays Pitfalls

Description

A simple q-learning agent trains via reinforcement learning on a Q-table to learn how to play the game Pitfalls.

What is the game Pitfalls?

Player starts at top-left-hand corner and must learn to find his/her way down to the goal at bottom-right-hand corner. For example:

S F F F

F F F H

F H F H

H F F G

F = normal floor

S = starting / normal floor

H = hole (game ends)

G = Goal (wins! and game ends)

AI Gym Environment

The Pitfalls game is a special AI environment that behaves similarly to Open AI Gym environments.

About

A simple q-learning agent (using Q-table) via reinforcement training learn to play a simple self-made game called Pitfalls.


Languages

Language:Python 100.0%