教育资源为主的文档平台

当前位置: 查字典文档网> 所有文档分类> 工程科技> 电子/电路> Study+on+welded+seam+recognition+using+circular+laser+vision+sensor

Study+on+welded+seam+recognition+using+circular+laser+vision+sensor

上传者:林棻
|
上传时间:2015-05-04
|
次下载

Study+on+welded+seam+recognition+using+circular+laser+vision+sensor

328

CHINESEOPTICS

LETTERS/V01.5,No.6/June

10,2007

Study

on

weldedseamrecognitionusing

circularlaserVlSlonSenSor

Peiquan

Xu(徐培全)1,XinhuaTang(唐新华)2,RiNa(娜日)3,and

Shun

Yao(姚舜)2

201620

CollegeofMaterialsEngineering,ShanghaiUniversityofEngineering

Science,Shanghai

WeldingEngineeringInstitute.Shanghai

Jja0Tong

University,Shanghai200030

BuMness

3SchoolofComputer&Electronic

Engineering,ShanghaiSch001.Shanghai200235

ReceivedNovember1,2006

arcweldingsystembasedoncircularlaservisionsensorisdeveloped.Afterimage

de—noising,imagesegmentation,andimagethinning,therelationofdepthvalueofworkpieceandoff-axisangle7,three-dimensional(3D)calculation,andseamtrackingexperimentsarecarriedout.Finally,theerrorforseamtrackingsystemisanalyzed.Theresultsshowthat113Dinformationcallbeobtainedusing

Anovelvisualrobotic

proposedvisualroboticarcweldingsystemandthereal-timeseatlltrackingisrealized;2)theseamtrackingerrorissmallenoughforgastungstenarcwelding(GTAw)process,andthissystemcanbeusedforseamlocationandseamtrackingorsealnfinder.

the

DC昭codes:100.0100.080.0080.

rapiddevelopmentincomputertechnologyandrobottechnologyhasgreatlypromotedtheautomatie-ityofweldingprocess.Thereweremanyfactorsthatinfluencedtheweldingquality,whichnotonlyincludethicknessofthematerialstobewelded,thechemicalele—

computesthelocationoftheweldingseamwithrespecttothevisionsensor.Inordertocompensateforvari-ationsinpartorjointgeometry,thisprocessingresultwasthenusedintherobotcontrolmoduletocorrectthe

weldingpath.

Forhardware.theoverallsystemwascomposedoftwomodules:arobotcontrolmoduleandavisionprocessingmodule.Therobotcontrolsystemconsistsoftwounits:

mentcompositionoftheweldedmaterial,surfacestatusoftheworkpiece,butalsoincludesuchparametersas

weldingvoltage,weldingspeed,weldingpath,andthree-dimensional(3D)ofworkpiece,whetherornotwerede-tectedinrealtime.Theautomationandartificialintell[-gencecontroloftheweldingprocessrequirethedevelop—

mentofboththejointshapeandtheseamtrackingcon-tr01.sothattherobotmayselectcorrectpathautomat[一

callywhenthedeviationbetweenweldingpathandactualseamhasbeenfound.Inordertomeettherequirementof

mainprocessingunitandjointservounit.ThemainprocessingunitfINTELPentiumIIICPUoperatingat780MHzloperatesmajorprocessingsucha8coordinatetransformationfortraiectorycontrol。sensoryfeedbackcontrolforseamlocationandseamtracking,andsoon.

Thevisionprocessingmodulegrabsandanalyzestheim-agedataobtainedfromthevisionsensortorecognizethetypeofweldjointandcomputethepositionoftheweld

automaticweldingprocess,allkindsofsensorshavebeeninvestigated[1,引.Amongthem.“spotlaser’’wasthesire-plestmeasurementmethodamongactivevisionsensors

andothervisionsensors.Accordingtolasertriangularrules,differentpatternlaservisionsensorswereputfor—wardtoapplystructuredlightandscanninglightusingsinglelaserstripe,twolaserstripes,fivelaserstripes,andcircularlaserstripeaslightsource.However.‘‘spot’’or“line”typelasersensorshavesuchshortcomingsaslessinformation,explanationambiguity,andonlybeusedtosingleortwodirectionseamtracking[。一….Xueta1.setupawholeofcircularlaservisionsensorsystem[7J.Onthebasis,anovelvisionsensorwasdeveloped.Basedon“radius-depthrelation”oflaserconeandimaginglawof

joint

center.

Theprincipleofvisionsensorwasthat:circularlasertrajectorywasusedasaactivelighttoprojectontotheworkpiecetobewelded,andwhenthetasksofseamloca,tionandseamtrackingwerecarriedoutbyweldingtorch,the3Dofweldmentcanbemeasuredbyvisionsensorus—ingcalibrationmodelanddepthcalculationalgorithm,andtheerrorofweldingtorchandweldmentcanbefur.therdetectedandsentbacktorobotg].Thevision

sensorsensor

(CCD)image

consistsofchargecoupledeviceandlightscanningsystembasedon

rotatedlensthatgeneratescircularlaserbeam,a18showninFig.1.Thecamerawasfittedwithanarrow—bandop—ticalinterferencefilterhavinganarrowbandpassof0.99amcenteredat611.49am.Thenarrow—bandinterfer—fliterefficientlyremovesmuchoftheambientand

weldingarclight矗omtheimage.Thecameracircuitry

ence

camera,3Dmodel

arc

forfeaturepointsofweldseams

undergastungstenwelding(GTAw)processwas

es—

tablished.Subsequently,seamlocationandtrackinginroboticweldingprocesscanberealized.

proposedsystem,anarticulatedrobotwithsix

degreesoffreedomwascarryingaweldingtorchandavi-sionsensorsystem.Thevisionsensorviewstheweldingjointaheadoftheweldingtorchwhiletherobottravels

our

In

generatesanNTSCstandardvideosignalthatwasstoredinthevisiongrabberasa768×576arrayof8一bitpixel.Onepixelintheimageplanecorrespondsapproximately

alongtheweldingseam.Theimagesfromthevisionsen-sorwereanalyzedbyanimagegrabbercardandusedtoreconstructthedimensionsoftheweldingseam,which

8.3越monafocusedobject。Thesensorfieldofviewandresolutionwerereconfigurabledependingontheap-

to

plicationrequirements.Asalightsource,the

10一mW

laserdiodeemittingat

万方数据 

1671—7694/2007/060328-04⑥2007ChineseOptics

Letters

June

10,2007/V01.5,No.6/CHINESE

OPTICSLETTERS

329

Fig.1.Configurationofthevision

sensor

byscanningtechnique.

635nmwasused.Themainadvantageofilluminationbythelightsourceinthisbandwasitsvisibilityandintell-sitv.whichconsiderablyfacilitatestheeahbrationofthesensorinthestageofdevelopmentofthevisionsensor.Simultaneously,itstillfacilitatesthesegmentationofthelaserlightfromtheweldingseamimagebecausemostsignificantspectralcomponentsofscatteredlightfromthearcexistintheultravioletfUV)andvisiblebands.Also.itguaranteesgoodtransmissionofthelaserlightthroughthesmokeandfumegeneratedduringwelding.Thesensoroperatesonaprincipleofproposed3Dcal-culationmodelbasedon“radius—depthrelationoflaserconeandcameraamplificationcoefficient”.

Theexperimentwascarriedoutincomputerizednu-mericalcontrol(CNC)platform,andIRB2400weldingrobotwasusedtoholdoiltheproposedvisionsensorbasedoncircularlaser.GTAWmethodwasoperated.andtheparametersusedinthissensorysystemwereshowninTable1.

DuringthecourseofimageacquisitionIimperfectcir-cularlasertrajectoryalwaysappearedbecauseofthelimitationofscanningmotorandchargecoupleddevice

based

on

1/3

valueratherthan

mid-whie.Incontrast

withlinearfiltermethod,order-statisticfiltercouldbet—tersolvecertaintypesofimageprocessingproblems.Thisfiitermethodmakesthecircularlasertrajectoryperfect.Inaddition.itsabilitytoreducerandomDOisewithoutblurringedgeswasasmuchasacomparablelinearlowpassfilter.Thelaser—strengtheningandnoise-reducingeffectsthatorder-statisticfilterhasonallimagedependontworelatirethings:thespatialextentofthe

maskandthenumberofPixelsinvolvedinthemedian

computation.Here,theywereinthefive-by-fivemask,and9pointswereselectedforthemask.

example.theoriginalwelded

imageduringG墨&粕processusingcircularlaservisionsensorasweldingseamdetectorwasillustratedinFig.2(a).Becauseofthehigh-intensityarclightinterference,

process

as

蹦ngGTAW

novelbarrierlayerwasadoptedtoremovemostofthearclight.Therefore.theweldingseamwasseparatedintotwoparts:onewasforcircularlaserzone;theother

one

was

arc

lightzone.De—noisemethodbased

on

3一level

(CCD)camera.Here,weaddressorder-statisticfilter

Table1.SpecificationofSensorySystemUsedfor

SeamTrackingExperiment

Item

Distance

and

Specification

betweenTorchTip

Coordinate8ofLaserCone

Center

W=84

mm;H=5

ITIITI

DistancebetweenLens

and

CoordinatesofLaserCone

178mm

FocalLengthofLens

(afterCalibration)

Heightof

6.21mm

ImagePlane

4.8mm6。4mm

12.88。39.250300

Width

ofImagePlane

ot

ConeAngle

SeparationAngle口SeamTrackingSpeed

Detect

mm/min

millmmA

Fig.2.Results

RangeY

士20士2552928

DetectRange

WjldCurrent

ArgonFlowRateSamplingRate

for“imagepre-processing”.(a)Originalim-

L/rain

frame/s

age;(b)3Dimage;(c)histogram;(d)imagesegumntation;(e)5×5imagethinning;(f)histogramafterimage

内容需要下载文档才能查看

equalizing.

万方数据 

330CHINESEOPTICS

LETTERS|V01.5,No.6

June1012007

Fig.3.De-noiseparametersandresultforcircularlaserseamimageof3-levelwavelet.(a)Waveletgraph;(b)originalim—

age(1eft)(Ijoint,butt,GTAwelding,I=52A,v=5mm/s,100kgf/cm2)andde—noiseimage(right).

wavelettechniquewasputfcIrwardtode—noisetheweld-ingseamimage,whichisillustratedinFig.3.Figure3(a)illustratesthewaveletgraphforde—noiseprocess,thecontrasteffectimageoforiginalimageandde—noised

imageisshowninFig.3(b),theweldingparameters

are:

,=52A.v=5mm/s.100kgf/cm2.

In

addition,“two—stepsegmentation(TTS)”andim-

agethinningtechniquewereusedforfeatureextractionofweldimagesunderGTAWcondition.asshowninFigs.

2(d)and(e).Theresultsshowthatthistechnique

can

be

tosegmentweldingseamimagefrombackground

accurately.

Afterimagesegmentationandimagethinning,the3DexperimentcanberealizedbytheproposedvisionsensorandmathematicalmodelIsJ,includ—buttweldingseamwithgap,buttweldingseamwith

weldingseam,filletweldingseam,andrampseametc..Figure4illustratestheprocessof

trackingforbothI—typebuttjointandsinewelded

Here.takingI—typebuttweldedseamasanexample.lasertrajectorywasproiectedtoI-typeweldmentactivelight;the3Dcalculationcanberealized.Onewasgrabbedusingcircularlaservisionsensorandpositionsweremeasuredatthesametime.The3D

ofseamtrackingbased

on

circularlaser

sensor.(a)Ion

string

welded

joint;(b)sine

welded

万 

方数据119

舍7邑5

毋3

一l

吕u

壬如

Fig.5.ExperimentalresultsforI-typebuttweldedjointus-ingcircularlaservisionsensorasseamguider.(a)RelationofdepthvalueZLand7angle;(b)3Dcalculation;(c)seaJn

tracking.

calculationresultwasshownin

Fig.5(a).The

relationof

depthinformationofworkpiecetobedetectedandoff-axisangle吖wasillustratedinFig.5(b).

Guidedbycircularlaservisionsensor,seamtrackingexperimentforbuttweldjointwascarriedoutinABBl40robotplatform.Seamtrackingcanberealizedbytwo

types

seamtrackingmodel.Onewasoff-lineprogram-

mingseamtracking,thatistosay,seamtrackingprocess

wasdividedintotwo

stages.Stage

1:3Dcalculation

ofweldediointcharacteristicpointfl‘om2Dinsensitiv-

ity

planethroughlinageacquisition,imageprocessing,and

3Dcalculationalgorithm;stage2:seamtracking

accordingtotheobtained3Dpositiondataoftheweld.ingseam.Theotherwasseamtrackinginrealtime,thatistosay,seamtrackingwassimultaneouswiththe3Dpositiondatacalculationoftheweldingseam.When

seamtrackingwasinprocess,twothreadswereoccurring:computerthreadusedforimageacquisition,imagepro—cessingand3Dcalculationfrom2Dinsensitivityplane,robotthreadusedforcommunicationwithcomputerand

guidanceformanipulatoralongtherealweldingseam.TheseamtrackingresultswereshowninFig.5(c1.Developedseamtrackingsystembasedoncircularlaservisionsensorwasamulti—hardwareconsistingofimagegrabber,visionsensor,androbot;amulti-taskdatapro—cessingwithimageprocessing,weldseam3Dcalcula-tion,andcommunicationbetweenrobotandcomputer;evenamulti—nodeinformationtransfercomplexsystem.Theprecisionofthesystemwasaffectedbykindsoffac—tors,includinglightscanningsystem,cameracalibrationandrobotcalibration,imageprocessingandevenweld—ingseam3Dalgorithm.Themainerroranalysiswas

asfollows.1)Cahbrationerror.Intheseamtrackingsystem,calibrationconsistsoftwoparts:onewascall—brationforABBrobot;theotherwascameracalibration.

pointstoolcenterpointfTCP)”calibration

methodwasusedtocalibratetheABBrobot.Conse—quently,weldingtorchcoordinatescanbedeterminedbyrobotcontroller.Thecalibrationerror,meanerror

内容需要下载文档才能查看 内容需要下载文档才能查看

ofO.24

usedimagecalculationcircularinggroove,lapweldingseamjoint.

circularimage563as

Fig.4.Illustratevision

Wherein.“4

joint.

June

10,2007/V01.5,No.6/CHINESE

OPTICSLETTERS331

andmaximum

error

of0.67wasachieved.21Scanning

systemerror.Scanningsystemerrorresultsfromscan-ninglens,includingsphericalaberration(errorcausedby

differentdistancesofseverallightswhentheyillustratethescanninglens.thethicknessofwhichcannotbeomit—ted).The

error

analysismanifestthatwhilethedistance

betweentheIightandcenteroflenswaswithin40mm.themaximumerrorwas0.707mm.

ThisworkWassupportedbytheYouthFoundationofShanghaiUniversityofEngineeringScience(No.06—7),SpecializedResearchFundforExcellentYoungTeachersofShanghaifNo.06xpyql71.AndmanythankswillbegiventoDr.FengguiLu,Prof.H.L.Yu,S.B.Chen.andDr.C.J.FanfromShanghaiJiaoTongUniversityand

W.Zhang

fromShanghaiLaserManufacturingInstitute.

P.XU’Se-mailaddressisxupeiquan7810@yahoo.corn.an.

万 

方数据R息fefences

1.Z.Li,Z.Jing,andZ.Sun,Chin.Opt.Lett.2,578

(2004).

2.J.-Y.JuandS.-J.Na,Mechatronics

7,599(1997).

3.H.Zhang,Z.Xiong,J.Jia,andJ.Pan,Chin.J.Mechan icalEngineering(inChinese)38,113(2002).

4.G.M.Graham,“Georgia

Institute

ofTechnology”Ph.D.

thesis(School

of

MechanicalEngineeringGeorgiaInsti-tuteofTechnology,1995).

5.A.DeandD.Parle.ScienceandTechnologyofWbldingandJoining8,340(2003).

6.H.LUOandX.Chen.InternationalJournalofAdvancedManufacturingTechnology26,1012(2005).

7.P.Xu,X.Tang,F.Lu,S.Yao,andH.Yu,Journalof

ShanghaiJiaotongUniversity(inChinese)6,902(2006).8.P.Xu,S.Yao,F.Lu,X.Tang,andW.Zhang,ChinOpt.Lett.3,640(2005).

内容需要下载文档才能查看 内容需要下载文档才能查看 内容需要下载文档才能查看

Study on welded seam recognition using circular laservision sensor

作者:作者单位:

Peiquan Xu, Xinhua Tang, Ri Na, Shun Yao

Peiquan Xu(College of Materials Engineering, Shanghai University of EngineeringScience, Shanghai 201620), Xinhua Tang,Shun Yao(Welding Engineering Institute,Shanghai Jiao Tong University, Shanghai 200030), Ri Na(School of Computer &Electronic Engineering, Shanghai Business School, Shanghai 200235)中国光学快报(英文版)CHINESE OPTICS OF LETTERS2007,5(6)

刊名:英文刊名:年,卷(期):

参考文献(8条)

1.Z Li;Z Jing;Z Sun Multi-sensor image fusion using discrete wavelet frame transform[期刊论文]-Chinese Optics of Letters 2004(02)2.J -Y Ju;S -J Na 查看详情 1997(07)3.H Zhang;Z Xiong;J Jia;J.Pan 查看详情 20024.G M Graham Georgia Institute of Technology 19955.A De;D Parle 查看详情 20036.H Luo;X Chen 查看详情 2005

7.P Xu;X Tang;F Lu;S.Yao,H.Yu 查看详情 2006(06)8.P Xu;S Yao;F Lu;X.Tang,W.Zhang 查看详情 2005(03)

引用本文格式:Peiquan Xu.Xinhua Tang.Ri Na.Shun Yao Study on welded seam recognition using circularlaser vision sensor[期刊论文]-中国光学快报(英文版) 2007(6)

版权声明:此文档由查字典文档网用户提供,如用于商业用途请与作者联系,查字典文档网保持最终解释权!

下载文档

热门试卷

2016年四川省内江市中考化学试卷
广西钦州市高新区2017届高三11月月考政治试卷
浙江省湖州市2016-2017学年高一上学期期中考试政治试卷
浙江省湖州市2016-2017学年高二上学期期中考试政治试卷
辽宁省铁岭市协作体2017届高三上学期第三次联考政治试卷
广西钦州市钦州港区2016-2017学年高二11月月考政治试卷
广西钦州市钦州港区2017届高三11月月考政治试卷
广西钦州市钦州港区2016-2017学年高一11月月考政治试卷
广西钦州市高新区2016-2017学年高二11月月考政治试卷
广西钦州市高新区2016-2017学年高一11月月考政治试卷
山东省滨州市三校2017届第一学期阶段测试初三英语试题
四川省成都七中2017届高三一诊模拟考试文科综合试卷
2017届普通高等学校招生全国统一考试模拟试题(附答案)
重庆市永川中学高2017级上期12月月考语文试题
江西宜春三中2017届高三第一学期第二次月考文科综合试题
内蒙古赤峰二中2017届高三上学期第三次月考英语试题
2017年六年级(上)数学期末考试卷
2017人教版小学英语三年级上期末笔试题
江苏省常州西藏民族中学2016-2017学年九年级思想品德第一学期第二次阶段测试试卷
重庆市九龙坡区七校2016-2017学年上期八年级素质测查(二)语文学科试题卷
江苏省无锡市钱桥中学2016年12月八年级语文阶段性测试卷
江苏省无锡市钱桥中学2016-2017学年七年级英语12月阶段检测试卷
山东省邹城市第八中学2016-2017学年八年级12月物理第4章试题(无答案)
【人教版】河北省2015-2016学年度九年级上期末语文试题卷(附答案)
四川省简阳市阳安中学2016年12月高二月考英语试卷
四川省成都龙泉中学高三上学期2016年12月月考试题文科综合能力测试
安徽省滁州中学2016—2017学年度第一学期12月月考​高三英语试卷
山东省武城县第二中学2016.12高一年级上学期第二次月考历史试题(必修一第四、五单元)
福建省四地六校联考2016-2017学年上学期第三次月考高三化学试卷
甘肃省武威第二十三中学2016—2017学年度八年级第一学期12月月考生物试卷

网友关注视频

【部编】人教版语文七年级下册《逢入京使》优质课教学视频+PPT课件+教案,安徽省
七年级英语下册 上海牛津版 Unit9
苏科版数学八年级下册9.2《中心对称和中心对称图形》
二年级下册数学第一课
19 爱护鸟类_第一课时(二等奖)(桂美版二年级下册)_T3763925
冀教版小学英语四年级下册Lesson2授课视频
精品·同步课程 历史 八年级 上册 第15集 近代科学技术与思想文化
第19课 我喜欢的鸟_第一课时(二等奖)(人美杨永善版二年级下册)_T644386
8.对剪花样_第一课时(二等奖)(冀美版二年级上册)_T515402
冀教版英语四年级下册第二课
冀教版小学数学二年级下册第二单元《有余数除法的简单应用》
外研版英语七年级下册module3 unit2第一课时
【部编】人教版语文七年级下册《泊秦淮》优质课教学视频+PPT课件+教案,天津市
沪教版牛津小学英语(深圳用) 四年级下册 Unit 4
三年级英语单词记忆下册(沪教版)第一二单元复习
沪教版八年级下次数学练习册21.4(2)无理方程P19
沪教版牛津小学英语(深圳用) 五年级下册 Unit 10
二次函数求实际问题中的最值_第一课时(特等奖)(冀教版九年级下册)_T144339
北师大版数学 四年级下册 第三单元 第二节 小数点搬家
冀教版英语五年级下册第二课课程解读
3月2日小学二年级数学下册(数一数)
【部编】人教版语文七年级下册《老山界》优质课教学视频+PPT课件+教案,安徽省
冀教版小学数学二年级下册第二单元《余数和除数的关系》
沪教版牛津小学英语(深圳用) 四年级下册 Unit 2
【获奖】科粤版初三九年级化学下册第七章7.3浓稀的表示
第五单元 民族艺术的瑰宝_15. 多姿多彩的民族服饰_第二课时(市一等奖)(岭南版六年级上册)_T129830
外研版英语三起6年级下册(14版)Module3 Unit2
北师大版数学四年级下册3.4包装
《小学数学二年级下册》第二单元测试题讲解
沪教版八年级下册数学练习册21.4(1)无理方程P18