基于CMOS摄像头的智能车路径识别与方向控制毕业论文外文翻译.docx

基于CMOS摄像头的智能车路径识别与方向控制毕业论文外文翻译.docx

《基于CMOS摄像头的智能车路径识别与方向控制毕业论文外文翻译.docx》由会员分享，可在线阅读，更多相关《基于CMOS摄像头的智能车路径识别与方向控制毕业论文外文翻译.docx（10页珍藏版）》请在冰豆网上搜索。

基于CMOS摄像头的智能车路径识别与方向控制毕业论文外文翻译

英文原文

RouteIdentificationandDirectionControlofSmartCarBasedonCMOSImageSensor

Abstract

Thispaperisdesignedforthe2ndFreescaleCupNationalUndergraduateSmartCarCompetition.WithMC9S12DG128singlechipandsmartcarmodelsuppliedbythecommittee,aCMOSimagesensorisappliedtodetecttheblacktrackonwhiteraceway,whichextendsthedetectionrangeandishelpfultopredicttheforwardpath.Inthispaper,ten-linepixelsinanimageareanalyzedtolocatetheblacktrack,andthePDalgorithmbasedonPIDisemployedtocontrolthedirectionandangleofthesteeringgearrespectively.Byrepeatedtesting,thesmartcarcanrunstablyonthegivenracewayatahighspeed.

Keywords:

routeidentification,directioncontrol,smartcar,MC9S12DG128singlechip,imagesensor,PID

algorithm.

1.Introduction

Therulesof2ndFreescaleCupNationalUndergraduateSmartCarCompetition[1]maybesummarizedasfollows:

theracewayconsistsofalotofwhiteboardsonwhichablacktrackisattached;thesmartcardesignedbyparticipantsrunsalongtheblacktrack;everycarrunstwocirclesinthisgameandthebesttimesoftwocircleswillbethefinalscoreofthiscar,andapparentlytheteamwhosecartakesthebesttimeswillbearthepalm.Accordingtotherules,weshouldensurethatthecarcandistinguishtheblacktrackfromwhiteboardinordertomakethesmartcarrunstably.Therearetwocommonmethodsforrouteidentification:

oneisusinginfrareddiodeasthesensor,andanotherisusingCCD/CMOSimagesensor[2].ThispaperusingCMOSimagesensorasrouteidentificationsensor,thereasonsforwhichareasfollows:

（1）TherangewhichiscoveredbyainfrareddiodesensorismuchsmallerthanaCMOSimagesensorcovers,andonlywecandoistouseseveraldiodesensors,butthemaximumnumberofdiodesensorsusedinthesmartcaris16;

（2）TheworkingvoltageofaCMOSimagesensor（3.3V）islessthanaCCD（12V）or16infrareddiodes.Apparently,usingCMOSimagesensorcannotonlyreducethepowerconsumptionbutalsoextendthevisiblerangeofthesmartcar,andalsoenablethecartopredicttheforwardpath.Thispaperpresentsasystemicsolutionforidentifyingtheracewayandcontrollingthedirectionofsmartcar.

2.CMOScamera

ThereareseveralkindsofCMOSimagesensorsinthemarket.IncomparisonwithotherCMOSimagesensors,theOV6130CMOSimagesensor[3]madebyOmniVisionTechnologiesInc.isthebestchoiceforustodesignaCMOScameraforsmartcarwhetherfromtheviewpointofcostandperformanceorpowerconsumption.TheOV6130isablackandwhitesensorwhichhasa1/4inchCMOSimagingdevicecontainingapproximately101,376pixels（352×288）.Thissensorincludesa356×292resolutionimagearray,ananalogsignalprocessor,dual8-bitA/Dconverters,analogvideomultiplexer,digitaldataformatter,videoport,SCCBinterface,registers,anddigitalcontrolsthatincludetimingblock,exposurecontrol,blacklevelcontrol,andwhitebalance.ByassemblingtheexperimentalcircuitwetesttheOV6130outputportstiming（VSYNC,FODD,HERF）.Figure1showstheexperimentaltimingdiagrams.

（a）VSYNC-FODDtiming

（b）FODD-HERFtiming

Figure1OV6130experimentaltimingdiagrams

Figure2Structureoftheimagecapturinganddisplayingsystem

ReferringtotheOV6130datasheet,thesetimingdiagramsmatchwellwiththosegivenindatasheet,thusweproducetheOV6130CMOScamerabasedontheexperimentalcircuit.Inordertocheckwhethertheimagescapturedbycamerahavecleardefinitionandsharpcontrastornot,andalsotoconfirmthevisiblerangeofthecamera,wedesignaVBprogramforcapturingtheimagesanddisplayingthemoncomputerscreen.Thisprogrambasesonthreehardwaredevices:

CMOScamera,MCUorsinglechip,PC.Figure2presentshowthesethreedevicesworktogether.

Figure3comparestheoriginalimageofasnakelineofracewaywiththeimagecapturedbyCMOScameraandthendisplayedonscreenbyVBprogram.Itcanbeseenthatthecapturedimagehascleardefinitionandsharpcontrast,andthislaysafoundationforrouteidentificationtobediscussedlater.

（a）Smartcarreadytoscantheraceway

（b）CapturedimagebyCMOScamera

Figure3Comparisonbetweenoriginalimageandcapturedimage

3.Routeidentification

RouteidentificationaimsathelpingthesmartcartorecognizetheforwardtrackbyamethodwhichpicksuptheblacklinefromtheimagecapturedbyCMOScamera,andinfact,thismethodworkswellinthefollowingcases:

straightline,curvinglineandsnakeline.Byrepeatedtesting,wedecidetoanalyze10linesofawholeimagetopredicttheforwardconditionofsmartcar.Figure4illustrateshowweanalyzethe10-linepixelsofanimage.

Figure4Routeidentificationdiagram

Thedetailedalgorithmisintroducedasfollows:

Step1:

Calculatecoordinatesoftheblackpixelforeachlinereadytobeanalyzed.Asisillustratedinfigure4,thelines（L0,L1,…,L8,L9）aretobeanalyzed,andthewhitepoints（P0,P1,…,P8,P9）areblackpixelsforeachline.TheoriginOissuperposedbyP9,whichmeansthereisnoblackpixelinlineL9.AssumedthatP（x）andP（y）indicatex-coordinateandy-coordinateofpointP,respectively,herebothP9（x）andP9（y）equal0.Thekeyofthisstepistofindtheblackpixelofeachline.Here,bytakingthefollowingdatumwhichshowsthegrayvaluesofallpixelsinalineasexample,weintroduce

anewapproach:

19521020721520822823624323823423823523123323023523022219620720420820912916065171519187915117215317315014715914115314715413713915613614191411331384127110141112104

（1）AssumedthatPistheblackpixelofthisline,G（i）denotesapixelandiindicatesit’spositioninthisline,V（i）isthegrayvalueofG（i）,jisthepositionofthefirstblackpixelappearinginthelinefromlefttoright.HerebothP（x）andP（y）areinitiatedto0,andjis0fromthestart.

（2）Asforeachpixelinthislinefromlefttoright,comparingitsgrayvaluewiththeblackvalueB,herewesetBas30,G（i）isablackpixelonconditionthatV（i）islessthanB.Inthelineabove,thepixelsunderlinedinbold,whosegrayvaluesare17,15,16,18,arecomponentpointsoftheblacktrack,andthepixelsunderlinedbothinboldanditalic,whosegrayvaluesare9,4,areinvalidorinterferentialpixels.IfV（i）islessthanB,setjasi.ThencompareV（j+1）andV（j+2）withBrespectively,resultsgototwosides:

①IfbothV（j+1）andV（j+2）arelessthanB,checkthegrayvalueforeachpixelfromG（j+3）totheendpixelofthisline.IfthetotalnumberofwhitepixelswhosegrayvaluesaregreaterthanBsurpassesorequalsto3,thensetPasG（j+1）.Otherwise,goto（3）;②Ifnot,repeatthiswayfromthepointG（j+3）on.

（3）Ifthereisnoblackpixelinthisline,setbothP（x）andP（y）as0.

Step2:

Calculatetheaveragecoordinatesof10blackpixels.AsisshowedinFigure4,Mistheaveragepoint,M（x）andM（y）areexpressedasfollows:

Step3:

AccordingtothepositionofMintheimage,wecandecideinwhichdirection（ahead,left,orright）thesmartcarshouldturn.InFigure4,thesmartcarshouldturnrightobviously.

Step4:

Calculatehowmanyanglesthecarshouldturn.FurtherdescriptionsisillustratedinFigure5,whereαFisthecentralpointoftwofrontwheels,Mistheaveragepointmentionedpreviously（seeFigure4）,DandL1indicatethewidthandheightofthevisiblerangeofCMOScamerarespectively,L2isthedistancebetweenvisiblerangeandfrontwheels,L3isahalfoffrontwheel’sdiameter,L2+L3meansthedistancebetweenvisiblerangeandtheaxisoffrontwheels.ReferringtoFigure5,itisveryeasytocalculatetheangleα.

Figure5Calculatingtheangleforturning

4.Directioncontrol

Thecentralunitfordirectioncontrolofsmartcarissteeringgear,itsinputsignalisPWM（Pulse-WidthModulation）pulse,anditoutputscorrespondingangleinradian.ThispaperusesthePWMoutputportofMC9S12DG128singlechip[4]astheinputsignalofsteeringgear.Byinputtingthegivendiscretewidthofpulse,wetesttherelationshipbetweentheinputandtheoutput.Theformulabelowexpressesit:

whereXistangentialvalueoftheoutputangle,Yiswidthoftheinputpulse,y0isthecorrespondingvaluewhentheoutputangelequals0andkisslope.Usingthislinearrelationshipwecanoperatethesmartcarsimplyjustbyinputtingthetargetangel,thusthewidthofpulsewhichistheinputsignalofsteeringgearcanbecomputedeasily,thismethod,however,doesn’tworkwellinthefollowingcases:

curvinglineandsnaketrack.ThereforeweapplythePID[5]（Proportional,Integral,andDifferential）controllerwhichisverypopularinfieldsofautomationandcontroltechnology.ThekerneltheoryofPIDistodoproportional,integralanddifferentialoperationsontheinputdifferencerespectively,thenjointhethreeresultsasthefinaloutputvalue.Inpractice,itisveryflexibleforustouseaccordingtofeaturesanddemandsoftheobjecttobecontrolled.WemaychooseoneortwoorallofPIDmodules,forexample,wecanuseproportionalandintegralmodulestomakeupofPIcontroller.Asforthesmartcar,itisnonecessarytoconsiderthetrackswhichhavebeengoneacross,soweonlyuseproportionalanddifferentialmodulesasPDadjuster,andPadjusterhasbeendescribedpreviously（seeFormula

（2））,andthefollowingformulashowstheDadjuster:

whereenew,eolddefinethedifferencesofthistimeandlasttimewhentheangleiscomputed（heredifferencemeanstheangelbywhichthesteeringgearshouldturn）,tisscanningperiodofCMOScamera,kdisdifferential