SSRP Abstract

Board 12: Artificial Intelligence in Modern Board Games

Student Scientist: Aadarsha Gopala Reddy ’23
Research Mentor: Sean McCulloch (OWU Department of Mathematics & Computer Science)

Automated agents have been developed for card and board games for decades. My research similarly involves creating a digital version of the card game Lost Cities and implementing an automated agent to play against. My goal is to improve the agent to a level where it can do better than a human at the game.

Lost Cities is a two-player card game which takes about 10 minutes to play. It involves the players placing or discarding wager cards and expedition cards in some of the five columns/expeditions. Each expedition where a card has been placed has an initial cost. At the end of the game, when there are no cards available to draw from the undealt cards, the total points minus the cost of starting the expedition times the wagers are added up from each expedition. Whichever player has the highest total is the winner. I developed a program using Java that implemented these rules and perfectly works for human vs. human games. I also developed an intelligent agent for a human to play the game against. The methodology behind the agent is to estimate the final score for both players. The estimate uses the probabilities of drawing cards from both the undealt cards and the discarded cards to form an expected score for each expedition. This estimate is calculated for each action available to the agent. The action that creates the best estimated score over the opponent’s would be played. We also added calculations to determine the best available card drawing action. We plan to test this methodology against human players and expect to improve upon this design in future summers.