Advertisement
Published in:
2013 | OriginalPaper | Chapter
Evolutionary FSM-Based Agents for Playing Super Mario Game
Authors : R. M. Hidalgo-Bermúdez, M. S. Rodríguez-Domingo, A. M. Mora, P. García-Sánchez, Juan Julian Merelo, Antonio J. Fernández-Leiva
Published in: Learning and Intelligent Optimization
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Abstract
Most of game development along the years has been focused on the technical part (graphics and sound), leaving the artificial intelligence aside. However computational intelligence is becoming more significant, leading to much research on how to provide non-playing characters with adapted and unpredictable behaviour so as to afford users a better gaming experience. This work applies strategies based on Genetic Algorithms mixed with behavioural models, to obtain an agent (or bot) capable of completing autonomously different scenarios on a simulator of Super Mario Bros. game. Specifically, the agent follows the rules of the Gameplay track of Mario AI Championship. Different approaches have been analysed, combining Genetic Algorithms with Finite State Machines, yielding agents which can complete levels of different difficulties playing much better than an expert human player.