Master Thesis: Rivalry AI

This master thesis describes how to create a socially interactive character, which engages into a rivalry with another game agent, if a conflict developed between the two of them. The thesis covers a summary about the four key emotions without story in games and introduces the psychological theories of aggressive behaviour and competitive situations. These theories have been used as foundation to set up goals, which the Rivalry Artificial Intelligence (AI) should accomplish.

Further different techniques to implement socially interactive characters into games are discussed. To provide a socially interactive character for the concept of the Rivalry AI, the μ-SIC System is used and adapted. This system is also used to model the player’s emotional state and relationships. The concept of the Rivalry AI will be implemented in an existing game. The game is called BareFoot and is a fun race. To be able to understand further implementations of the Rivalry AI into the game, a common method to create a racing AI is introduced. Additionally to the race track representation for an AI, the thesis briefly covers an approach to automatically parse a race track for a racing AI. Afterwards the architecture of the Rivalry AI is presented and further describes how it was implemented into the game. To evaluate if the Rivalry AI accomplished the prior set goals, user tests were conducted and the results con-firmed or at least partially confirmed the set hypotheses of the Rivalry AI.

Master Thesis.pdf