欢迎来到天天文库
浏览记录
ID:39402565
大小:1.79 MB
页数:87页
时间:2019-07-02
《基于模糊理论和强化学习的自主式水下机器人运动规划技术》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、分类号UDC密级:编号:工学硕士学位论文Y780712基于模糊理论和强化学习的自主式水下机器人运动规划技术硕士研究生:指导教师:学位级别:学科、专业:所在单位:论文提交日期:论文答辩日期:学位授予单位:姜沛然张铭钧教授工学硕士机械电子工程机电工程学院2005年3月2005年5月哈尔滨工程大学哈尔滨工程大学硕士学位论文AbstractFirst,inmypaper,tothequestionaboutunderwaterrobotlocalpathplanning,thereal—timepathplanningsystembasedODfuzzycontrolisdes
2、igned.ThiSsystemsolvesthepathplanningprobleminthecomplicatedenvironment.Thesimulationresultsverifiedthewaypresentedinthispaperiseffective.Second,Considerthatthefuzzyplanningsystemisdependingontheexperienceandknowledgeofoperator.Itisinsufficientthatthefuzzyrulemayexist,andconfirmthatthepe
3、rfectfuzzyregularworkloadisverylarge.So,considerimprovingthemethodofplanningReinforcementlearningisasimplekindofstudymechanism,anditonlyneedonestrengthensignaltojudgethequalityofmovements,doesnotneedmanypeople’Sinterventionverymuch.However,practiseandrealizethelocalpathplanningsystemsimp
4、lywithreinforcementlearning,thetimeforstudyiscertainlylong,andstudybadpossibility.TheplusesandminusespractisedonthebasisofthefuzzytheoryandReinforcementlearning,SOconbinethetwotogether,maketheelementaryfuzzyrulefirst,thentherobotutilizesreinforcementlearningtobepractisedandstudyindepende
5、ntlyontheelementaryandfuzzyandregularbasis,andadjustthecorrespondingfuzzyrule.In哈尔滨工程大学硕士学位论文thisway,canreducealargeamountofartificialwork.Meanwhile,theenvironmentthattherobotisinisnotunalterable,manyunknownfactorscan’tforeseeforpeople,anditcanbestudiedaccordingtotheconcreteenvironmentwh
6、ileutilizingthiswayunderwaterrobot.Torevisetheimproperfunction,ithasstrengthenedtheadaptivecapacityoftherobot.Becausehasalreadyconfirmedtheelementaryfuzzyruleinadvance。thetimeforstudyofrobotwillbeshortenedgreatly.ThesimulationresultsverifiedthewaypresentedinthiSpaperiseffective.Keywords:
7、AutonomousUnderwaterVehicle;Real—timePathPlanning;FuzzyControl;ReinforcementLearning哈尔滨工程大学学位论文原创性声明本人郑重声明:本论文的所有工作,是在导师的指导下,由作者本人独立完成的。有关观点、方法、数据和文献的引用已在文中指出,并与参考文献相对应。除文中已注明引用的内容外,本论文不包含任何其他个人或集体已经公开发表的作品成果。对本文的研究做出重要贡献的个人和集体,均己在文中以明确方式标明。本人完全意识到本声明的法律结果由本人承担。作者(签字):塞壁垦
此文档下载收益归作者所有