A Coordination Model Using Fuzzy Reinforcement Learning for
Multi-Agent System · 2007
Itisimportantformulti agentsystemthatthefun ctionally� independentagentsapplynegotiation,� coordinationand� cooperationtoperformsomesetoftasksortosati sfysome� setofgoals.� Inthispaper,� weproposeatwo layer � architecturecoordinationmodelbasedonfuzzy� reinforcementlearningformulti agentsystem.� Agen ts� makeuseoffuzzyinferencesystemtochoosetheop timal� behaviorlocallyandconfertheintentionsandacti onsof� othersaccordingtotheirstateinformation.� Then� coordinationlayerharmonizessub goalsamongagent s� andassignsrationaltasktoeachagentwhilelearn ingthe� strategiesofagentsusingfuzzyreinforcementlear ning.�As� aresult,� agentschooseandexecuteproperactiont o� accomplishthedesiredtasktogetherinactionlaye r.�The� simulationresultsshowedthattheperformanceof� attackingisobviouslyimprovedintheRoboCupsocc er� simulationgame.� � � � �