• Aucun résultat trouvé

Development of a Real-Time Laser Scanning System for Object Recognition, Inspection, and Robot Control

N/A
N/A
Protected

Academic year: 2021

Partager "Development of a Real-Time Laser Scanning System for Object Recognition, Inspection, and Robot Control"

Copied!
9
0
0

Texte intégral

(1)

READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.

https://nrc-publications.canada.ca/eng/copyright

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la

première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez

pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.

Questions? Contact the NRC Publications Archive team at

PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the

first page of the publication for their contact information.

NRC Publications Archive

Archives des publications du CNRC

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. /

La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version

acceptée du manuscrit ou la version de l’éditeur.

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Development of a Real-Time Laser Scanning System for Object

Recognition, Inspection, and Robot Control

Beraldin, Jean-Angelo; Rioux, Marc; Livingstone, F.R.; King, L.

https://publications-cnrc.canada.ca/fra/droits

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site

LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

NRC Publications Record / Notice d'Archives des publications de CNRC:

https://nrc-publications.canada.ca/eng/view/object/?id=1a1b62c2-5457-4d27-b6e0-3eaec766698e

https://publications-cnrc.canada.ca/fra/voir/objet/?id=1a1b62c2-5457-4d27-b6e0-3eaec766698e

(2)

2

1

o o

1

2

NRC35063{SPIE{Vol.2057TelemanipulatorTechnologyandSpaceTelerob otics(1993),pp.454{461.

Recognition, Inspection, and Robot Control

Abstract

1 Introduction

1

F.R.Livingstone,L.King J.-A.Beraldin,M.Rioux

HymarcLtd. Institute forInformationTechnology

38AurigaDrive NationalResearch CouncilCanada

Ottawa,Ontario,CanadaK2E 8A5 Ottawa,Ontario,CanadaK1A 0R6

A real-timelaser scanner for object recognition, insp ection, and rob otcontrol is presented inthis pap er. Range

and intensity imagesare generated inp erfect registration at arate of 10 Mega-samples p er second. Imageshave a

resolutionof 483lineseach having512pixels. Owing to itscompatibilitywiththevideostandard RS-170, therange

cameracanb edirectlyinterfacedtoanimagepro cessorthroughavideoframegrabb erwitheitherananalogordigital

input. Theangular eldofviewis30 30 . Thestand-o distanceis0.5mandtheop erationaldepthof eldis1m.

Furthermore, co op erative targets canb e measureduptoseveral metersawayfromthecamera. Therangeresolution

is12bits. Undernormalop eratingconditions,arangeprecision of0.2mmand2mmare achievedat0.5mand1m

resp ectively. Thecamera weighs only2.75kg. Thesource inthe present systemis aNd-YAGlaser emittingalaser

b eamwith awavelength of 1064 nmand a maximump ower onthe scene of 2W. Futuredevelopmentswillinclude

theuseofasource at 1540nmtomakethesystemeye-safe. Thetestingoftheprototyp eiswellunderwayand so on,

imagesofavarietyofrepresentative objectswillb e available.

Canada is participating inthe Space Station Freedom Program by providinga MobileServicing System (MSS)

that will b e used in assembling and maintaining the station as well as servicing and manipulating payloads and

loading/unloadingtheshuttle. TheMSSconsistsof twoelementsonthestation, theMobileServicing Centre (MSC)

andtheMSSMaintenanceDep ot(MMD).ThroughtheStrategicTechnologiesinAutomationandRob otics(STEAR)

Program,the Canadian Space Agency (CSA) is fostering additional technological development b eyond the activity

b eingundertakenbytheMSSindustrialteam(SPAR,IMP,CAE,CAL,SED,MDA).Thetechnicalobjectiveofoneof

theprojectsistodevelopvisionsystemtechnologyforapplicationtothecontrolandop erationofrob oticsystems. The

purp oseof theproject is nottobuild actual ighthardwareor softwareforthe MSSbut to develop technology that

willmeettheneedsofthelongertermgrowthandupgradingoftheMSS.Concepts mayb edevelop edinanapplication

areawhere theContractoralreadyhasexp ertise,aslongasthelevelofcomplexityordicultyiscomparableto that

envisagedontheMSS.

This pap er describ es areal-timelaser scanner builtby HymarcLtd. under contract to theCSA.The systemis

based on recent advances made at the National Research Council on such a system . The laser scanner generates

registered rangeandintensityimagesaccordingto theRS-170standard. Compatibilityto thisvideostandard allows

for a direct link to the Arti cial Vision Function (AVF) that is a key element of the MSS. Integration of the laser

scanner to the AVF will allowthree dimensionalsurveying of the rob otic work-site, identi cation of known objects

fromregisteredrange andintensityimages,andobject trackingbased onrangedata. Theinvestigationofalgorithms

formo del-basedtrackingarerep ortedinaseparatearticle .

Thegeneral requirementsof theMSS arereviewedinSection 2. Section 3discusses theneedfor visioninspace

rob otics. An in-depthdescription of the scanner and its calibrationare presented inSection 4. Finally,conclusions

(3)

MSS Command

Processing

MSS

AVF* Tracking

Control Mode

AVF* Command

Processing

Camera/PTU

Control

AVF*

Text/Graphics

Display

MSS Command

Processing

Image

Processing

Video

Switch

AVS

Software

AVF* Video

Display

EE

PTU

AVF*

AVU**

Camera

Camera

Camera

Camera

Camera

Object

Laser

Scanner

AVF*

AVU**

- Artificial Vision Function

- Advanced Vision Unit

3     3.1 General

Impactofgrowthon basicAVF.

2 MSS Vision Requirements

3 Vision for Space Robotics

Figure1:

ThecurrentAVFisbasedontheSpaceVisionSystem(SVS)thatwasdemonstratedonCANEX-2. TheSVS

analy-sesvideosignalsfromtheclosedcircuittelevisionsystemofthespaceshuttleandprovidesreal-timep osition/orientation

and velo city information of an object with aco op erative targetarray . The baseline requirements for the Arti cial

VisionFunction(AVF)areasfollows:

Toidentifysuitablyilluminatedobjects(targets) fromtheirvideoimages.

Toestimatethep osition,attitude,translationalrate,androtational rateoftheobject.

Toprovideappropriatecameracontroltotracktheobject.

Tob eabletotrackobjects b eforecapturebyMSSmanipulatorsandb erthobjects/payloadshandledby

manip-ulators.

TheCSAintendsto enhance thecapabilities oftheAVFby developingadditionalvision systemssuch as a

laser-based range nding camera, as shown in Figure 1. The enhanced AVF willb e capable of p erforminglab elled and

unlab elledobjectidenti cation,shap edeterminationofunknownobjects,automatictargetacquisition,up dateofworld

mo deldatabase,andmanipulatorendp ointandlinkde ectionsensing.Thesenewfeatureswillb eusedinthefollowing

areas: enhancedtargetrecognitioninAVFtracking/b erthing,worldmapveri cationforcollisiondetection/avoidance,

andobjectrecognitionand veri cationduringautomaticop erations,e.g.,truss assemblytask.

Computervisionwillb eintegraltothesuccessoftheSpaceStationProject. Thespaceenvironmentisanop enone,and

eventso ccurunexp ectedly andwithdeviationsfromanypre-sp eci ed plan.Thus,sensory-basedfeedbackisnecess ary,

andvisionisthesensethroughwhichsomeofthemostvaluableinformationab outtheenvironmentisavailable. Space

op erations are alsocharacterized by anoverriding requirement forop erational safety and reliability. Given thevast

amountsofuncertaintythat characterize most complexrob oticop erations andgiven theirhighlyunstablenature, it

followsagainthatlargeamountsofsensingdataarerequiredforreliability. Todate,mostsensinghasb eeninteractive.

Ahumanop eratorhasdirectvisualfeedbackeitherthroughtheco ckpitwindowsorviaavideoimagingsystemmounted

(4)

y

4

5

4 Real-Time Laser Scanner

3.2 Active versus Passive Vision

3.3 Synchronized Scanning Principle

4.1 Vision System Electronicsand Data Processing

StationProjectisthatatthepresentlevelofplanning,theamountofextravehicular activity(EVA)requiredapp ears

high. Thus, any technology likelyto decrease this need willreduce op erational costs and increase the safety of the

astronauts.

Vision involves the analysis of the prop erties of the luminous ux re ected or radiated by objects. To recover the

geometricalstructure s of theseobjects, either torecognize or to trackthem through time,twobasicvision strategies

are available. One strategy, passive vision, attempts to analyzethe structure of the scene under ambient light. In

contrast,active visionisanattempttoreducetheambiguityofscene analysisbystructuringthewayinwhichimages

areformed. By keeping theprop ertiesofinvarianceinrotationandtranslationof3-Dobjects,sensors thatcapitalize

onactive visioncan resolve mostoftheambiguitiesfoundwith2-Dimagingsystems. Moreover,with structuredlight

approaches, the 3-D information b ecomes insensitive to background illumination and to surface color and texture.

Thus,thetaskofpro cessingthe3-Ddataisgreatlysimpli ed. Triangulation-basedlaserrangecamerasareexamples

ofvisionsystemsbuiltonsuchastrategy.

For close-range spaceactivities,arangecamerabased ontriangulationwith asingle-sp ot scanningmechanismis

recommended. Single-sp otscanningisemphasizedhereb ecauseatypicalsceneinspacewillb ehighincontrastandwill

generate spuriousre ections. Theseoriginatefromtheintense naturalradiation eldcausedmostlybythesun. Also,

thethermalblanketsusedtoprotect somepiecesof equipmentarehighlysp ecularat theop eratinglaserwavelength.

Further,whenthespacestationisintheshadowoftheEarth,thevisionsystemwillhavetoop erateinquasi-darkness.

Thissituationcallsforacamerathatcanop erate overawidedynamicrangeofintensities. Throughthecontrolofthe

cameraparameters onapixel topixelbasis,increased dynamicrangeoftherangecameraisachievable. Hence,range

camerasbased onsingle-sp otscanningcan b eusedinalongertimewindowduringoneorbitofthespacestation .

Rioux intro duced an innovative approach to triangulation based range imaging by using a synchronized scanning

scheme which allowsvery large elds of viewwithsmalltriangulationangles withoutcompromisingprecision. With

smallertriangulationangles,areductionofshadowe ectsisinherentlyachieved. Implementationofthistriangulation

techniquebyanauto-synchronizedscanner approachgivesaconsiderablereductionintheopticalheadsizecompared

to standard triangulation metho ds. Range measurement andlateral p osition measurementare essentially decoupled

andthereisanimprovedambientlightimmunitydue tothesmallinstantaneous eld ofview.

Many issues a ect the requirements for the signal pro cessing hardware. The basic and key requirement is that the

electronicsprovide rangeandintensitydataatnominally10MHz. However,a nal designmustaddressotherissues,

includingdesignandmanufacturingcosts,systemsize,andp owerrequirements. Theunderlyingconsideration,forany

spaceorterrestrialpro duct,isthat thesystemhasfourbasic functionalelements:

1. Controlofthep olygonalmirrorand -axisscannerandgeneration ofp osition andintensitysignalsat10MHz

2. Dataacquisition,p ointpro cessing(e.g.,calibration)and intermediatestorage -allinrealtime(10MHz)

3. Imagepro cessing

4. Man/machineinterface

Eachofthesefourmainareaswillrequireuniqueelectronicstosatisfythetechnicalrequirements. However,many

ofthese requirements areachievable,fromatechnical p ointof view,with technology readilyavailableto day. This is

nottosaythatanimplementationwithto day'stechnologieswouldb eappropriateforaspaceenvironment,butrather

(5)

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

AAAAAA

Analog Unit

VME bus

Local bus

MVME-147

MaxGraph

MegaStore-8

MegaStore-8

MaxScan

Daughter Card

Compute Unit

Timing Unit

MAXbus

Console

Terminal

RS-170

Video

Monitor

Polygon

Controller

Laser

340M

Disk

Tape

Drive

Real-time

Scanner

x y o o TM TM TM y = N =N =N Systemarchitecture .

4.2 Video Rate Scanner

Figure2:

and -axis scanner, scan synchronization, signal generation, analog-to-digital conversion, real-timecalibration, and

short-termhigh-sp eedstorage areareaswheresolutionsexistortechnologiesarereadilyavailabletocreatesolutions.

The functionalelements 2 and 4 are implementedon a MAXVideo familyof imagepro cessing b oards from

Datacub e. Figure2showsadiagramofthesystemarchitecture. Thearchitectureofthereal-timelaserscannerisbuilt

aroundtheindustry standardVMEbus,aMotorola68040generalpurp ose CPUrunningunder areal-timeop erating

systemOS9 fromMicroware, someMaxVideo b oardsand4customsignalpro cessing b oards. Atthe moment,

theimagepro cessingisdoneonadi erentplatform. Sp ecializedhardwareb oardscanb eaddedinalaterphasewhen

allthealgorithmsrequired forthecontrolandop erationofrob oticsystemswillhave b eenfullyinvestigated.

Figure 3a shows a schematic diagram that emphasizes the di erent elements of the laser scanner. The raster typ e

imagesareobtainedinthefollowingway. Thelaserb eamisrelayedtothep olygonscannerthroughaseriesoffo cusing

lenses. Therotationofthep olygonimpartsascanningmotiontothelaserb eamalongthe axis. Thescanalongthe

axisis achievedwith a galvanometer (ormotor). The re ected lightfromthe scene is collected back through the

mirrormountedonthegalvanometer(ormotor),afacetofthep olygonscanneropp ositetotheprojectionofthelaser

b eam,thecollectinglens,and nallyitimpingesonap ositiondetectorwheretheinformationondepthandintensity

isderived. Theauto-synchronizatione ect isachievedbytwofacetsofthep olygonalmirror.

IntheRS-170standard, afullimageiscomp osedof 525horizontallines(483 are active) whichare displayed on

amonitor in1 30second. Repro duction of a fullimageon amonitor is achieved by the sequential transmissionof

twointerleaved elds at 60 Hz. The equivalent line rate is therefore 15750lines/second. At a samplingrate of 10

MHz,512samplescan b e acquiredp er line. Due tothe size ofthecollectingmirror,it isnotp ossible toachieve the

required15750scans/sec ondusingagalvanometerdrivenmirrororresonantscanner. Thetechnologythatcanachieve

thissp eedisap olygonalmirrordrivenathighsp eedbyamotor. Ap olygonalmirrorwith facets rotatestheb eam

through 720 . A 30 scanangle willrequire a24-facet p olygon,which isone ofthestandards intheindustry. To

achievevideolinerate,itmustrotateat15750 revolutionsp ersecond,i.e.,656.25rps. Suchscannersusuallyrotate

continuouslyinonedirectionata xedsp eedtoproviderep etitiveunidirectionalscans. Themirrorismounteddirectly

onthe electric motorshaft. The combinedinertia of the p olygonand motorrotor contributeto rotational stability.

Thishigh inertia rendersthem impracticalfor applicationsrequiringrapid changesinscanvelo city. For true RS-170

compatibility, thegalvanometer mustb e drivenat 60 Hz. Withthis prototyp e, it isdriven at 30Hz (alsoknown as

(6)

y

x

z

Galvanometer

Scanner

Position

Detector

Polygon

Scanner

Source

a)

Source

Collimating

Lens

Position

Sensor

Imaging

Lens

Field

of View

β

fo

d

γ

a

b)

r ' o o o 6 5 4 y ! a   a ! l a ! l 4.3 Optical Design

Schematic diagramof(a)completeoptical arrangement,(b) mainoptical pathunfolded. Figure3:

Withanylaser-basedrangecamera,depthof eld(DOF)andaccuracyarerelated. Increasedaccuracycanb eachieved

byreducing thedepth of eld. For rob otic applications,adepth of eld of 1000mmisreasonable, according to the

technicalliteratureandfeedbackfromusersofrob ots. Figure3bshowsthegeometricalarrangementofthetriangulation

approachusedherewiththe -axisscannerremovedforclarity. Thelaserb eamiscollimatedandfo cusedinthevolume

ofviewinsuchawaythatthelasersp otsizeisalmostconstantwithinit. Togetmaximumresolutionfromtheimaging

device,thedesired depth of eldwasmatchedtotheRayleighcriterion ofafo cused Gaussianlaserb eam. Thisleads

tothefollowingrelationship,

= 2

(1)

where isthe wavelengthof thesource, is thedepth of eld,and is theradiusof thefo cused lasersp ot at the

center of theDOF. Using =1000mm, = 1000mmand a source emitting at 1064nm,one gets 0.41mm.

Witha eld of viewof ab out 400mmat = 1000 mm, then with asp ot radius of 0.41 mm,more than 975pixels

canb esampledalongthescanaxis. Inpractice, 512pixelsp erlinearesampledinordertomatchRS-170-compatible

imagepro cessingsystems. Thesp eci cationsoftherob oticvisionsystemaresummarizedinTable1.

Theestimationoftherequired laserp ower isanimp ortantstep inthedesignpro cess. Bytakingintoaccountthe

normalop eratingrangeofthep osition detector,a2W laserisrequired tomeasuresurfaces withvaryingre ectanc e.

A dio de laseris preferred b ecause ofits size; thesingle-mo de typ e gives b etter b eamquality thanmulti-mo deones.

Atthepresent time,themaximumlaser p ower available with thistyp e of laseris 1W.Because ofthe level oflaser

p owerrequiredbythisrangecamera,eye safetyb ecameimp ortantfromthestartoftheproject. Twop ossibilitieshad

to b e considered, i.e., either lo cateor design ap osition detector with a current gainmechanism (e.g.,avalanche or

micro-channelplate-basedp ositionsensitivedetectors )ormovetoasaferop eratingwavelength. Thelattersolutionwas

favoredb ecauseoftechnicallimitationsofthepreviousdevices. AnInGaAsp ositiondetectorsensitiveat1540nmwas

selected . Thiscontinuousresp onse detectorpro duces twocurrentsrelatedto thep ositionofthecentroidofthelight

distributionimpingingon itssurface. Thebandwidthof this InGaAsdetector is 7.2MHz, which meets the required

videobandwidthof4MHz. Inthe visiblesp ectrum, afraction ofamilliwattis sucienttocause eye damage,since

thep owerofcoherentlightenteringthepupilismagni ed,attheretina,through thelens,byafactorofroughly10 .

Atlongerwavelengths,however,strongabsorptionbywaterpresent intheeye serves toreduceopticalp owerreaching

thelens,thusincreasingthedamagethreshold. At1540nmthethreshold p ower foreye damageisroughly10 times

greaterthanwhatitisinthevisiblesp ectrum. Also,thenoisee ectivep oweroftheInGaAsp ositiondetectorisab out

(7)

A

B

C

D

E

F

y y 7 7

4.4 Range Data Calibration

Sp eci cations ofreal-timelaserscanner.

Conceptforthe scanner(not toscale): A=130mm,B=70mm,C=40mm,D=70 mm,E=40mm,F=190mm.

Stand-o 500mm

X-axisscan (atmid-DOF) 400mm

Y-axisscan (atmid-DOF) 400mm

Rangeprecision min-DOF 0.2mm

mid-DOF 2.0mm

max-DOF 12mm

Samplingrate 10MHz

Videorate: RS-170 Samplesp erline 512

Linesp er images 525

Imagerate 30

Table1:

Figure4:

surfaces aresimilarat 1540 nmand inthevisible region . Thus, there is p otentialfor signalimprovementof 40dB

(optical)if1540nmischoseninsteadofvisiblewavelengthsformakingeye-safemeasurements. Forthis rstprototyp e,

aNd:YAGlaser(1064nm)wasusedinstead oftheerbium-dop ed b er laser b ecauseofthelaserp ower required. At

thepresent rateofdevelopment,however,itisreasonabletoexp ect higherp owersinthenearfuture.

The mechanical layout of a scanner for rob otic vision is mainly determined by the space requirements of the

p olygonalmirror,drive motor,and the -axismotorandmirror. Aconcept forthescanner isillustratedinFigure 4.

Themirror/motorassemblyismountedonanopticalb edplatewhich alsosupp ortsalloftheopticalcomp onentsand

thusensuresastableopticalalignment. Thelaserinputisfroma breoptic connectionandtheb eamisdirectedonto

thep olygonalmirrorby asteering mirror. Thep ortion ofthecasingthat containsthe -axismotorand scan mirror

hasb eenmademo dularsothatitcanb eremovedandusedasasingle-axisscanner. Themo dulecouldalsob ereplaced

bysp eciallensassemblyforahighaccuracyversionof thescanner. Thedimensionsoftherangecameraare givenin

Figure4. Thep olygonmotor,which istheheaviest comp onent,weighs0.6 kg. Therange cameraweighs ab out2.75

kg.

Therange camerayields threequantities p er samplinginterval: twofor theangular p osition ofthe mirrorsand one

(8)

2 8 x;y 5 Conclusion 6 Acknowledgements References

pro cessingalgorithms. Are-mappingofthesevariablestoamorecommonco ordinatesystemlikearectangularsystem

istherefore required.

Thecalibrationisp erformedwitha atplatecomp osedofanarrayoftargetsthatisp ositionedatknownlo cations

infrontof therangecamerawith aprecise linearstage. This plate is madeupof evenlyspaced targets onagridof

11 13. Oncetheregisteredrangeandintensityimageshaveb eenacquiredforallthetargetarrayp ositions,theedges

ofthosetargetsareextractedtosub-pixelaccuracyusingamomentpreservingalgorithm. Thecenterofeachtarget(in

termsoftherawcameraco ordinates)isthendeterminedby ttinganellipsetotheedgep oints. Aniterativenonlinear

simultaneous least-squares adjustmentmetho d extracts theinternaland externalparameters of the cameramo del .

There-mappingof theraw datato arectangularco ordinate systemis doneo -linethrough software. Unfortunately,

resultscouldnotb e shownduetounforseenproblemswiththelasersource. Thetestingoftheprototyp eiswellunder

wayandso on,imagesofavarietyofrepresentative objects willb e available.

Thispap er presentedareal-timelaserrange cameraforobject recognition,insp ectionand rob ot control. Thevision

requirementsfortheMobileServicingSystemthat willb edevelop edbyCanadathrough theCanadianSpaceAgency

anditspartners(industrial,universityandresearch lab oratories)werereviewed. Theneedforvisioninspacerob otics

wasclearlyidenti ed. Particularly,arangecamerabasedup onsinglelasersp otscanningwasidenti edasaverygo o d

strategythatcan meetthosevisionrequirements.

Theprop osed systemmeasuresregisteredrangeandintensityimagesat10Mega-samplesp ersecondaccordingto

theRS-170videostandard,i.e.,483lineswith 512pixels p erline. Withastand-o distanceof 0.5m,anop erational

depth of eld of 1 m,and arange resolution of 12 bits, the rangecamera willpro duce imagesadequate for various

rob otic tasks, e.g. collisiondetection and avoidance. Here, arate of 30 images p er second should provide enough

dataab out theenvironmentsurroundingthe rob ot. Thecontrol ofthe camera,thedata storage and low-levelp oint

pro cessingareimplementedonacommercialplatform. Futuredevelopmentswillincludededicatedb oardstop erform

thereal-timecorrectionandresamplingoftheimagesonaregular grid. Also,softwarewillb erequiredtointerpret

thedata and communicateits meaningto thesystem user. The particulartasks considered areobject identi cation

from surface structure , work cell mo delling,and p ose determination. Gathering the range data is only part of the

solutioninprovidingavisionsystemcapability.

TheauthorswishtothanktheCanadianSpaceAgencyforthe nancialandtechnicalsupp ortalongthecourseofthis

project. TheauthorswishtoexpresstheirgratitudetoG.Thompsonforhistechnicalassistanceprovidedinthecourse

ofthe systemtesting and L.R. Schmidtfromthe STEAR program for hisvaluable comments. Finally,theauthors

wishtothankE.Kidd forherhelp inpreparingthetextandP.Amiraultforthetechnicalillustrations.

[1] J.-A. Beraldin, M. Rioux, F. Blais, J. Domey and L. Cournoyer, \Registered Range and Intensity Imagingat

10-Mega Samplesp erSecond,"Opt.Eng.31(1),pp.88{94(1992).

[2] W.Cheung,P.LettsandG.Roth,\Mo del-basedTrackingusingaReal-timeLaserRangeFinder,"InPress,SPIE

Vol.2063(1993).

[3] S.G.Maclean,M.Rioux,F.Blais,J.Gro dski,P.Milgram,H.F.L.Pinkneyand B.A.Aikenhead,\VisionSystem

DevelopmentinaSpace Lab oratory,"SPIE, Vol.1395,8{15(1990).

[4] F.Blais,M.Rioux,andS.G.Mclean,\Intelligent,VariableResolutionLaserScannerfortheSpaceVisionSystem,"

(9)

[6] P. Maigne, J.-A. Beraldin, T.M. Vanderwel, M. Buchanan and D. Landheer,\An InGaAs/InP PositionSensing

Photo detector", IEEEJ.ofQuant.Electr.,26(5),820{823,(1990).

[7] M.Rioux,J.-A.Beraldin,M.S.O'SullivanandL.Cournoyer,\Eye-SafeLaserScannerforRangeImaging,"Appl.

Opt., 30(16),2219{2223(1991).

[8] J.-A.Beraldin,S.F.El-HakimandL.Cournoyer,\PracticalRangeCameraCalibration,"InPress,SPIEVol.2067,

Références

Documents relatifs

However the use of computing resources should also be linked to the dynamical behavior of the plant(s) to be controlled. Indeed while controlling different subsystems in a

Scan path planning for on-machine inspection in a 5-axis machine tool is still a challenge to measure part geometry in a minimum amount of time with a given

[r]

The attention mechanisms of human visual system (HVS) can be segmented an image into the region of interesting (ROI) which is considered important and the background which is

Stage simulator was responsible to simulate the environment with the robots and ob- stacles, and Ptolemy for simulate the control algorithm for avatar and real robot.. With

A silicon retina is a particular image sensor in which an analog andor digital signal processing circuitry is integrated next to the photosensitive or image sensing

In this section, we describe some of the characteristics that are essential in the design of situated reasoning systems for diagnosis and malfunction handling in process

Furthermore, evaluating our descriptor with an object recogni- tion system using Harris-Laplace and Laplacian interest point detectors on VOC2007 database showed a slight decrease