You are here

Artificial Intelligence of Modern Board Games: Battle Line

Daniel Bladow*, Gonzaga University (Undergraduate Student)
Talk Abstract: 
Many turn based board games played by computers use game state trees to determine what moves to make. Traversing through game state trees can be extremely time consuming, especially if the game involves random factors such as a deck of cards. Battle Line is a game where two players are competing over nine different three card poker hands using a 60 card deck with six suits and number values one through ten. Each hand is called a flag and the type of hand (such as straight flush or a three of a kind) is called a formation. Together, the deck and flags create an intractably large number of game states. We created an artificial intelligence where the computer uses a probabilistic approach to approximate future game states. We estimate future game states by determining the top formations we could make by playing a card from our hand on each flag. Then we calculate the odds of completing each formation, of each formation beating whatever the opponent can make, and finally of any of the top formations resulting in a win on that flag. By using probabilities the computer avoids traversing through numerous game states, and the computer’s strategy is based on mathematical principles. There are two advantages to this approach to artificial intelligence. First of all, it is less time consuming and can be played on weaker computers, possibly cell phones. Secondly, the computer’s thinking avoids ad-hoc strategic knowledge. It is given a mathematical concept and the rules of the game and then makes its own decisions about quality moves. Therefore, the computer is both more time efficient and capable of devising unique strategies based on mathematical principles.
Talk Subject: 
Mathematical Aspects of Computer Science
Talk Type: 
Poster Presentation
Wednesday, March 4, 2015 - 14:15