The huddle: Combining AI techniques to coordinate a player's game characters

Abstract

We present the huddle, a concept for extending games in which the player is responsible for a group of game characters. The huddle combines several AI methods to allow the player to create a cooperative strategy for his characters to solve a scenario of the game and it takes away from the player the need to frantically jump around in controlling his characters to employ the strategy idea he has. The huddle is entered from a saved game state and allows the player to provide his characters with strategy ideas in form of situations and the actions he wants the characters to take (SAPs). A learner then uses these ideas and adds to it additional SAPs to create a complete strategy. The learner uses a simulation of the real game that uses models for the nonplayer characters based on the experiences the player had with the game, so far, to evaluate strategy candidates. We evaluated the huddle idea with a fantasy-themed role playing game and show that the huddle indeed allows a player to concentrate on his strategy while still requiring him to come up with the solution ideas for scenarios.