区域交通环境下的智能车全自主导航方法分析

《区域交通环境下的智能车全自主导航方法分析》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、上海交通大学硕士学位论文ABSTRACTforward-lookinglaserradarinstalledonthefront-topofthevehicle,lanedetectionviacamera,obstaclesdetectionbytheforward-lookinglaserradarinstalledonthefrontofthevehicle,globallocalizationbyGPS,deadreckoningbyencoderandodometer.Intheroadboundarydetection

2、section,aroadboundarydetectionalgorithmbasedonlaserradarisproposedtosolvetheproblemthatpresentmethodscannotbeappliedinvariousroadconditions.Thealgorithmcanbeefficaciouslyappliedtodifferenttypesofroadconditions.Somelinesegmentsareextractedthroughself-adaptiveparameterb

3、asedonenvironmentcharacteristicsafteraseriesofpreprocessingonrangedataoflaserradar.Theextractedlinesegmentsarecombinedandfurtherselected.ThenunitetheselectedsegmentstomatchtheroadmodelsandgeneratetheroadboundarypointswhicharetrackedbyNearestNeighborFilter.Extensiveexp

4、erimentsconductedinurbanandsemi-urbanenvironmentdemonstratetherobustness,effectivenessandviabilityoftheproposedmethod.Inthepartofnavigationinroadsection,anavigationmethodbasedonpassableregionisputforwardtobeappliedincomplexorabominableroadsection.ThemethodemploystheVe

5、ctorFieldHistogram+(VFH+)toaccomplishvehicle’slocalpathplanning.ThepassableregionsareIV万方数据上海交通大学硕士学位论文ABSTRACTdeterminedbyfusinglane,roadboundaryandobstacleinformation.Acostfunctionisusedtoselectthefinaldirectionoftheintelligentvehicle.Thismethodeffectivelyfiguresout

6、theproblemofnavigationfailureduetosingleroaddetectiontechniqueincomplexenvironment.Experimentsdemonstratethatthismethodisstable,reliableandrobust.Finally,analgorithmofobstacleavoidanceandovertakingbasingondouble-arcpathplanningisproposed.Inthesectionofautonomousnaviga

7、tioninintersection,anavigationmethodbasedonpassableregionispresentedtosolvetheproblemthatthereisnotsufficientroadsurfaceinformation,whichcanbeusedtolocalizationandnavigation.Globallocalizationisusedtodeterminewhetherthevehicleisnearcertainintersection.Inordertoachieve

8、intersectionlocalization,alocalcoordinatesystemisbuiltbythestoplineorzebracrossingwhichisdetectedbycomputervision.Thendeadre