欢迎来到天天文库
浏览记录
ID:32206088
大小:9.01 MB
页数:67页
时间:2019-02-01
《基于有限理性模型的游戏智能行为系统的分析与实现》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、基于有限理性模型的游戏智能行为系统的研究与实现========!==!!!==!=!========================!========!!!=====!==!!!!!!=詈!!=!==!!!!!===========!!!!!!=!========!!!!!!!!!=======AbstractInthefaceofthegraduallymaturedanddiversifiedgamemarket,thegamedevelopersexpectthatthemeliorationoftheintelligencedesignwillenh
2、ancethegame’Splayability.Apartfromtheexternalperformanceofthebehaviorandtheoptimizationofthepath—findingalgorithm,itisthemainconcernofgamedevelopersthattheintelligentagentisreasonablycontrolledintheintelligentdesign.Thedevelopersnarrowthegapbetweenvirtualandrealtoimprovethegamechal
3、lengingthroughmakingtheintelligentagentsimulateplayers’behaviorandstrategies.Thispaperchoosesthemodel—basedintelligentbehaviorsystemastheoriginalsystem,andreferencestheSOARcognitivesciencemodelandtheboundedrationalitymodeltomaketheintelligentagentfittheplayers’behaviorpatternfrom3a
4、spects:theperceptron,thedecision—makingandthebehaviorcontrolunit.Themainworkandachievementshavebeendiscussedasfollows:(1)Atfirst,thispaperfirstreconstructstheintelligentagent’Sperceptiondomainthroughintroducinginthereceptivefieldofthebiologicalvisualperceptionmodel,inordertosolweth
5、eproblemwhichexitsinusingcone-horizonbooleantosettheperceptiondomain.Atthesametime,thevisualsensitivitymodelisexploitedtocalculatethevalueoftheuncertainperceptiondomain.Therefore,theinputinformationwillbedeterminedbycomprehensivelyanalyzingthepreviouslydata.Theconfrontationalexperi
6、mentsshowthatthenewperceptroncanprovideamoreadequateandusefulinformationforthedecision—makingwithoutaffectingthesystemefficiency.(2)Inthispaper,therecursivepredictionmethodcombiningtheforward—lookingtreeisproposedtoimprovethedecision—makingability.Ineverytimethesystemupdate,thismet
7、hodautomaticallybuildsaforward—lookingtreebasedontheintelligentagent’Sattributes.Meanwhile,thesystemcalculatestheutilityvalueofeveryexecutorinthistreeaccordingtothepreferencerelationandtheutilityfunction,andthenselec,tsthemostsatisfactorybehaviorasoutputthroughsystematicassessingth
8、ecomprehensiveutilitiesoft
此文档下载收益归作者所有