Research On MAS Game Based On Emergence

The electronic game is a complex system that composed of electronic gaming environment, human players and computer players from the electronic gaming system structure point of view. Gamers hope computer gamers can be intelligence and organization, rather than dull or dead automatic machine, so how to design computer games and their communities becomes the major problem which has plagued the game developers. The level of computer players is directly related to the performance of electronic games, and has also become the important factors that can influence the fate of the game studios and game companies. As computer hardware and software develop rapidly, building vivid and delicate object of the game in the graphics is relatively simple. Using the appropriate complex science theory and related artificial intelligence algorithms to design computer players and their communities becomes one of the necessary means.This paper takes the complex adaptive system theory and game artificial intelligence algorithms as a guide, and uses the corresponding Multi-Agent architecture to build the corresponding game model according to the specific game rules. By studying the characteristics of computer gamers and the interaction between them, dynamics model is established to observe the behavior of individual behavior and group behavior. Building the MAS game model can obtain the internal workings of MAS and the emergence of the phenomenon of deep understanding, which applying valuable reference for MAS game designers. The major work of this paper is as follows:Firstly, related technology in the MAS game is analyzed which becomes the theoretical foundation in designing game. Secondly, the distributed MAS structure and genetic algorithm is used for designing game mode. The Agent Multi-Agent Systems based on genetic algorithm has gotten the group behavior patterns and the MAS obtained a deep understanding of the phenomenon emerged. The results have proved that this method is feasible, robust and scalable. Finally, the neural networks, genetic algorithms and MAS are adopted for designing the hierarchical MAS system game model, which is an extension of the original model. Experimental results verify that this method is feasibility, the performance of agent is significantly improved and group behavior shows a phenomenon that from the chaotic to aggregate and concrete actions.