AI Pac Man




This project was implemented as part of the AI module at my MSc course in Hull. The concept was to create an AI controller for a self playing Pac-Man game. The controller had to always win the game with no ghosts present and win as much as possible against four ghosts. To achieve this I used the trusty A* pathfinding algorithm, various search techniques, movement patterns for evasion and a combination of a rule based system and a finite state machine for behaviour selection. Other AI techniques (like Neural Nets and Fuzzy Logic) were also investigated.


table tennis spectator view table tennis single player

table tennis active cam table tennis menu single

Skills Practiced / Knowledge Acquired

Search - Pattern Movement - Pathfinding - A* - Fuzzy Logic - Genetic Algorithms - Neural Networks - Finite State Machines - Scripting - Flocking - Rule Based Systems - Reactive Behaviours - Learning.


AI PacMan Demo