DocumentCode :
1873094
Title :
Hierarchical controller learning in a First-Person Shooter
Author :
van Hoorn, N. ; Togelius, Julian ; Schmidhuber, Jurgen
Author_Institution :
IDSIA, Lugano, Switzerland
fYear :
2009
fDate :
7-10 Sept. 2009
Firstpage :
294
Lastpage :
301
Abstract :
We describe the architecture of a hierarchical learning-based controller for bots in the First-Person Shooter (FPS) game Unreal Tournament 2004. The controller is inspired by the subsumption architecture commonly used in behaviour-based robotics. A behaviour selector decides which of three sub-controllers gets to control the bot at each time step. Each controller is implemented as a recurrent neural network, and trained with artificial evolution to perform respectively combat, exploration and path following. The behaviour selector is trained with a multiobjective evolutionary algorithm to achieve an effective balancing of the lower-level behaviours. We argue that FPS games provide good environments for studying the learning of complex behaviours, and that the methods proposed here can help developing interesting opponents for games.
Keywords :
computer games; control engineering computing; evolutionary computation; learning (artificial intelligence); recurrent neural nets; robots; Unreal Tournament 2004; behaviour selector; behaviour-based robotics; first-person shooter game; hierarchical controller learning; multiobjective evolutionary algorithm; path following; recurrent neural network; Artificial intelligence; Automata; Computational and artificial intelligence; Computer architecture; Evolutionary computation; Humans; Navigation; Neural networks; Robots; Weapons; FPS; First-person shooters; action selection; behaviour-based robotics; evolutionary algorithms; neural networks; subsumption architecture;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computational Intelligence and Games, 2009. CIG 2009. IEEE Symposium on
Conference_Location :
Milano
Print_ISBN :
978-1-4244-4814-2
Electronic_ISBN :
978-1-4244-4815-9
Type :
conf
DOI :
10.1109/CIG.2009.5286463
Filename :
5286463
Link To Document :
بازگشت