Using an Inference Engine for AI in the Office Tactics Video Game

RM 215

FSM (or its improvements) is the most common method employed when implementing AI on video games. Its major advantages are its simplicity and the ease of implementation but its greatest disadvantage is the predictability of the next state, which could lead to the player predicting the next step the game will take. Although there are some improvements done in FSM to alleviate this predictability, the inference engine allows a reasoning process and could come up with a strategy or move that the player might not be expecting. The inference engine consists of three parts: the knowledge base, the agenda, and the working memory. This paper developed an inference engine using the scripting language of UDK and applied to a video game, OfficeTactics. The resulting AI is very diverse and provides a lot of options that makes the game more exciting and enjoyable to play.