AGV视觉导引控制问题研究

IIIIIIRESEARCHONVISION-GUIDEDCONTROLALGORITHMOFAGVABSTRACTAGV(AutomatedGuidedVehicle)isincreasinglywidelyusedasthekeyequipmentinautomaticlogisticssystemandflexiblemanufacturingsystem.GuidingtechnologyisthecrucialtechnologyforAGV.Vision-guidedtechnologyusingalanemarkisthekindoftechnologythathastheapplicablefeasibilityinthenearfutureandwideapplicableprospect,notonlybecauseithassuchmeritsashigherinformation,betterflexibilityandeasiertoinstallandchangepathandsoon,alsobecauseitissimpleandapplied,lowerrequirementtohardwareandhasthecapabilityofadaptingexistingstructuredroadenvironment.Manydomesticandoverseasresearchersarecommittedthemselvestoit.Therefore,itissignificanttoresearchonvision-guidedtechnologyusingalanemark.Thecoreproblemsinallsubjectsonvision-guidedtechnologyarehowtoimprovetheaccuracy,stabilityandreal-timecapabilityinpathidentificationandpathfollowing.IVThispaper,onthebasisofAGVSscientificresearchplatformconstructedin“985”project,aimsatdiscussingthemethodsofimprovingtheaccuracy,stabilityandreal-timecapabilityinpathidentificationandpathfollowing.Themaincontentsarelistedasfollows:(1)Setupthehardwaresystemofvision-guidedsystemofAGV.(2)Severalgeneralimagesegmentationmethodsarecomparedbysegmentingtheidenticalroadimageinthewayofofflineexperiments.Accordingtotheexperimentresults,theimagesegmentationmethodbasedonHSLcolorthresholdispresented.Comparedwithtraditionalimagesegmentationmethodbasedongraylevelthreshold,ithasmanyadvantagessuchasstronganti-noiseability,usingmoreinformationduringimageprocessing,bettersegmentationresultwithoutbeingaffectedbyilluminativecondition,betteradaptabilitytocomplicatedroadenvironment.(3)Accordingtothefeaturesofimageshotanderrorsrequiredincludinglateralerrorandorientationerror,theimageincludingnumberedlinesdatainthemiddleareextractedoutofsourceimageandthenisprocessedtoacquirelateralerrorandorientationerror.Thiswaynotonlyensurestheaccuracyoferrorsacquiredbutalsogreatlyreducescomputingtimeaswellasgreatlyimprovesthereal-timecapability.(4)Afuzzylogiccontrollerisdesignedtocontrolpathfollowing,inwhichlateralerrorandorientationerroraresetasinputsandsteeringangleVissetasoutput.Iteffectivelyovercomesthebadinfluencecausedbynon-linearityanduncertaintyinpathfollowingsystemofAGVandimprovestherobustnessofsystem.(5)DiscusshowtoprogramcontrolalgorithmadoptingLabVIEWgraphicsprogramminglanguage.Finally,manyexperimentsaredoneandtheresultsindicatethatthemethodsofpathidentificationandpathfollowingpresentedinthispaperareeffective.visionguided,pathidentification,pathfollowingimagesegmentation,fuzzylogiccontrolLabVIEW11.1AGVAGV(AutomatedGuidedVehicle)[1]AGVJISD6801AGV[2]AGVAGVAGV[3](1)(2)(3)(4)AGV(5)AGV(6)(7)AGVAGVS[4][5]AGVS2AGVAGVSAGVAGV1.2AGVAGV50AGVAGVAGVAGVAGVAGVAGVAGVAGVAGVAGVSFMS(FlexibleManufactureSystem)FAS(FlexibleAssembleSystem)1974VolvoKalmarSchiindler-DigitronAGVSAGVKalmarAGVSAGVS1984GeneralMotorsAGVS1986AGV1407GeneralMotorsAGVAGVS[6]AGVAGVAGV1116AGV34AGV1975AGV80AGVAGV90CIMSAGVCIMSAGVAGVAGVSAGVAGVAGVAGVSAGVAGVAGVSAGV1.3AGV[3][7][8]AGVAGV2080AGVAGVAGVAGV()()AGV1-151AGVAGVAGVAGVAGV213,4/[9]AGV/AGVAGV3/AGV/65AGVAGV621172LAB1C2ABAC⎩⎨⎧+==)cot/(tantanbaaLyyx7CCD71.480[10]AGV.NabLab-5UBMVormas-pJLUIV1.4.1[11][12]13D281.4.23[13]1.5AGVAGVWTOAGVAGV9985AGVSAGV985AGVAGVAGVSAGVAGVAGV1AGV23AGV4AGV5LabVIEWAGVAGVAGVAGVHSLAGVAGV10AGVAGVAGVAGV11AGV2.1AGVAGVAGVAGVAGVSAGVAGVAGV11CPU2CPUCPUCPU212VLSI(VeryLargeScaleIntegration)3TransputerDSPVaMoRsP60Transputer[14]AGV1m/sAGV12CPU2.2AGVAGVAGVAGVAGVAGV21985AGVAGVAGVAGV2.3AGV2.3.1AGVAGV221CCD23AGV132.3.2AGV1CCDCCDChargeCoupledDeviceCCDCCD1970AGV14CCDCCD410-sCCDAGVHowardNCK41CVCCDCCDPAL50/6404800.11x8mmCCDCCDAGVmm[15]CCD21AGVCCD250mm322mm239mm0.5/mm0.4/mm2CCDCCDA/DD/ANINIPCI-1411879978PALNTSCRS-170RGBHSLHSLAGV15PCICPU132MB/S16M3E6C2-AG5c10240.35101AGVIPC-644P4-80PCA-6104P44PCIPCI-6771FCPUDASPCICPUCeleronTM850128MB20GPCI-1712DASPCLD-8712PCI-1712DAS168A/D16/212D/A±5V±10VPWM25A8PWM[16]PWMPWMPWMPWM[17]PWM2AGV16102SYX7.524V0.34KW3.3Nm3MetalrotaMR240FR2T/2.4AGVAGV17AGVAGVAGV3.1AGV313.2[18][19]18[20]CCDAGV[21][22][23][24][25]=),(yxg⎩⎨⎧-otheryxfyxmidyxfyxmid);,(),(),();,(e(31)),(yxf),(yx),(yxg),(yx),(yxmid),(yxe32a32b32aIRGBI19RGB3.3AGVAGVAGV[26][26][27][28]203.3.1RobertsSobelPrewittRoberts222PrewittSobel23333Laplacian3433Laplacian93335d=2,42422Laplacian21450[10]123412033.3.2[25][26][27]22[10]S0nS1S2SnnT0T1Tn-1(T0T1Tn-1)=),(yxg⎪⎪⎪⎩⎪⎪⎪⎨⎧≤≤≤----112100110),(;),(;),(;),(nnnnnTyxfTyxfTTyxfTTyxfggggMM),(yxf),(yxgT=),(yxg⎩⎨⎧≤TyxfTyxf),(;),(01(34)231)(zhz⎪⎩⎪⎨⎧∂∂=∂∂0)(0)(22zzhzzhT2)(zp)(zp)()(2211zpPzpP+1P2P)(1zp)(2zp24⎪⎪⎩⎪⎪⎨⎧⎥⎥⎦⎤⎢⎢⎣⎡--=⎥⎥⎦⎤⎢⎢⎣⎡--=222222221212112)(exp21)(p2)(exp21(z)psmpssmpszzz1m2m1s2s)(zp⎥⎥⎦⎤⎢⎢⎣⎡--21212112)(exp2smpszP⎥⎥⎦⎤⎢⎢⎣⎡--22222222)(exp2smpszP1P2P1P2P1m2m1s2s)(zp)(zh[]∑=-=niiimszhzpne12)()(1)(zp361m2mTTT25⎪⎩⎪⎨⎧==∫∫∞∞-TTdzzpTEdzzpTE)()()()(1221)(TE)(12TEP)(21TEP0)(=∂∂TTE)()(2211TpPTpP=⎥⎥⎦⎤⎢⎢⎣⎡--21212112)(exp2smpsTP⎥⎥⎦⎤⎢⎢⎣⎡--22222222)(exp2smpsTP1s2s)12ln(212221PPTmmsmm-++=3tt1212(1)1∑==tjjnn01q315(2)2∑-+==112Gtjjnnq=1-1q316jnn26G(3)m)(10∑-=×Gjjjnnf317(4)