Contains exercises from Game AI CS6150 at Northeastern Univ, Boston (Spring 2017)
1. Decision Making in Ms. Pacman
(a) implement, (b) test, and (c) compare and contrast results from two decision making techniques.
- Decision Tree
- Reactive Action Packages (or reactive planning)
2. Reinforcement Learning in Ms. PacMan
Trained Ms. Pac Man to play the game using Q-learning. Developed code allowing Ms. Pac Man to explore the game while updating Q-values.
- Please consult the Artificial Intelligence For Games Book: Chapter 5.1 through 5.4
- To implement Reactive Planning, follow the idea by Firby R. J. in "An Investigation into Reactive Planning in Complex Domains", implement a Reactive Action Packages structure. Create RAPPacMan.java
DTPacMan.java
RAPPacMan.java
QLPacMan.java