@MISC{Wulffert11künstlicheintelligenz, author = {Dimitri Wulffert}, title = {Künstliche Intelligenz für ein Massively Multiplayer Online Game}, year = {2011} }
Share
OpenURL
Abstract
Applying the peer-to-peer architecture to massively multiplayer online games (MMOGs) is a concept that until now didn’t reach any commercial game. The idea of creating a game that allows players to play MMOG over a peer-to-peer system is a great step forward for the future: it will allow small companies to create MMOG without having to sustain the cost of a server infrastructure or to request a monthly subscription to the players. On the other hand gamers will be more attracted to such games since they will not have to pay monthly fees to play a peer-to-peer MMOG. Aim of this work is to create a new AI for a peer-to-peer MMOG called Planet π4. First I analyzed the current state of art of the AI for computer games. Then I considered two different approaches to create the AI for Planet π4. The first AI implements a simple behavior while the second AI creates a more complex one. This is done in order to test if the complex AI plays in fact better than the simple AI. This thesis includes several simulations to verify both which AI is the best and to see how well each AI adapts to the game. Creating such AI for a peer-to-peer MMOG is necessary to simulate Planet π4 with thousands of peers. Each of these peers will be controlled by an AI and such AI should simulate human behavior. Finally this thesis makes some recommendations on how to improve the game and the current AI for future work.