Employing Artificial Intelligence Algorithms to a Strategy Game

MATTHEW L. RODRIGUEZ, AUSTIN HUANG, and ARDIT PRANVOKU

A goal of any artificially intelligent agent is to perform tasks that equal or surpass human efforts. We conducted research on techniques and algorithms to emulate expert human behavior in a strategy game based off the prominent game franchise Fire Emblem. This game allows players to expend their credits to create a team of characters (or units) based on each character's cost. Each unit then has its own attributes much like chess pieces governing how they move and attack. For example, each unit has different ranges of motion, ranges of attack and modes of attack. To demonstrate the practical applications of this research, we designed a game utilizing a graphical user interface which includes game music, images, and animations. The game can be played in two mode: player versus player or player versus computer. The AI was implemented using the MINIMAX algorithm and further optimized using techniques such as alpha beta pruning. The game also employed Machine Learning capabilities which improve the performance of the AI's decisions the more experience with the game the AI is exposed to.