/sokoban-qlearning

sokoban game solver through Multiple Search Algorithms and Reinforcement Learning (Q-Learning)

Primary LanguageJava

Sokoban AI Solver (HKUST COMP3211 Group 6)

Group members

Name Student ID
GAO, Huaxuan 20328408
HUANG, Xuhua 20329347
LIU, Shuyue 20329256
WANG, Guanzhi 20329604
ZHONG, Zixuan 20328800

Set up

1) clone this project

$ git clone https://github.com/XUHUAKing/sokoban-qlearning.git

2) launch the GUI

$ cd AI_Sokoban/bin
$ java GUI 

Search Algorithms

1) Load the map

click on the Load Map button and choose /levels/5.txt.

Before the result is computed, you can play the game. (↑ ↓ ← → L)

2) Compute the result

Choose a search algorithm and click on the button Compute Result

3) Move the agent

Keep pressing on the enter(return) button to make the agent move according to the result.

Q-Learning

1) Load the map

click on the Load Map button and choose /levels/QLearning/1.txt.

Before the result is computed, you can play the game. (↑ ↓ ← → L)

2) Compute the result

Choose 8) Q-Learning and click on the button Compute Result

3) Move the agent

Keep pressing on the enter(return) button to make the agent move according to the result.

Java doc

default package

reinforcement_learning package

Credit

wfalkwallace/AI/tree/master/hw2