This is the project I built for CS 188 at UC Berkeley. This project apply an array of AI techniques to playing Pac-Man game. This project involves techniques such as informed state-space search, probabilistic inference, and reinforcement learning.

This project involves multiple parts. The first part focuses on searching, the second part focuses on multi agents searching, the third part focuses on reinforce learning, and the forth part focuses on probabilistic inference.

Single Agent Searching


In this part, I implemented Depth First Search, Breadth First Search, Uniform Cost Search, and A* Search for my Pacman to find paths within a maze to reach a particular location while collecting food efficiently.

Multi-Agent Searching


In this part, I had to design searching for both my Pacman and ghosts. Techniques involved include Minmax and Alpha-Beta Pruning.

Reinforcement Learning


In this part, I inpletmented value iteration and Q-learning to play the Pacman game.

Probabilistic Inference


In this part, a new rule has been added which is my Pacman can eat the invisible ghosts. I hade to apply Probabilistic Inference technique to find all invisible ghosts and eat them.