Electromagnetic modular Smart Surface architecture and control in a microfactory context

HAL Id: hal-02120819 https://hal.utc.fr/hal-02120819

Submitted on 6 May 2019

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

and control in a microfactory context

The Anh Tuan Dang, Magali Bosch-Mauchand, Neha Arora, Christine Prelle, Joanna Daaboul

To cite this version:

The Anh Tuan Dang, Magali Bosch-Mauchand, Neha Arora, Christine Prelle, Joanna Daaboul. Elec-

tromagnetic modular Smart Surface architecture and control in a microfactory context. Computers

in Industry, Elsevier, 2016, 81, pp.152-170. �10.1016/j.compind.2016.02.003�. �hal-02120819�

Electromagnetic modular Smart Surface architecture and control in a microfactory context

The Anh Tuan Dang *, Magali Bosch-Mauchand, Neha Arora, Christine Prelle, Joanna Daaboul

Sorbonneuniversite´s,Universite´ detechnologiedeCompie`gne,CNRS,LaboratoireRoberval,CentrePierreGuillaumat,CS60319,60203Compie`gneCedex, France

Contents

1. Introduction... 000

2. Microfactoryconceptandconveyingsystemtrends... 000

2.1. Microfactorycontext... 000

2.2. Newtrendforflexibleandreconfigurableconveyingsystems:ModularmicrofactoryandSmartSurface(SS)concept ... 000

2.2.1. Modularmicrofactories... 000

2.2.2. SmartSurfaceconcept ... 000

2.2.3. Synthesisonthecharacteristicsandperformancesofconveyancesystems... 000

3. Proposedflexibletransferplatformformicrofactory:AnelectromagneticmodularSmartSurface ... 000

3.1. Principleofactuation... 000

3.2. Controlofroutingflexibility... 000

4. ProposedframeworkfortheelectromagneticmodularSmartSurface(emSS)managementandcontrol... 000

4.1. ManagemanufacturingprocesswithintheemSS(A0)... 000

4.2. MonitorandcontroltheemSS(A2)... 000

4.3. Generatepallettrajectory(A22)... 000

4.4. ModelemSS(A23) ... 000

4.4.1. Physicalsystemmodel ... 000

4.4.2. GridbasedemSSmodel... 000

4.5. SimulateemSS(A24) ... 000 ARTICLE INFO

Articlehistory:

Received30April2015

Receivedinrevisedform3February2016 Accepted8February2016

Availableonlinexxx Keywords:

Microfactory SmartSurface

Reconfigurablemanufacturingsystem Flexiblemanufacturingsystem Modularmanufacturingsystem Routingflexibility

Collisionavoidance

ABSTRACT

This paper presentsanelectromagnetic conveyance systemcalled electromagnetic modularSmart Surface(emSS)permittingtomovepalletsonaplanarsurfaceinamicrofactorycontext.Theproposed surfaceconceptallowsflexibilityinreconfiguringthesystemlayoutalongwithproductrouting.The possibilitiesofaccuratepositioningofthemovingpalletandcontrollingmultiplepalletsonthesurface maketheemSSsuitableforreconfigurableandflexiblemanufacturingsystems.However,theemSS controlneedstoberobustandscalabletoadaptthechangesinmanufacturingsystems.Aframeworkis thereforedefinedtomonitorandcontroltheemSSbysimulationorin-line.Itallowstodefineproduct routingontheemSSbysatisfyingnumerousrequirementssuchasreductioninenergyconsumption, collisionavoidance,etc.,andtominimizethehumaninterventionsbychangingproductroutingwhen emSScomponentfailuresoccur.AfirstexperimentrealizedonanemSSprototype,allowedtocompare twopathsstrategiesregardingcostfunctionlinkedtoenergyconsumptionandvelocities.Twoother studiesexploit theemSS modelingin termsof pallet path generationand simulation ofcollision avoidance.

ß2016ElsevierB.V.Allrightsreserved.

* Correspondingauthor.Tel.:+33344237357.

E-mailaddress:the-anh-tuan.dang@utc.fr(T.A.T.Dang).

ContentslistsavailableatScienceDirect

Computers in Industry

j ou rna l h ome p a ge : w ww . e l se v i e r. co m/ l oc a te / c om pi nd

http://dx.doi.org/10.1016/j.compind.2016.02.003 0166-3615/ß2016ElsevierB.V.Allrightsreserved.

5. Experimentsanddiscussion... 000

5.1. Case1:Validationofthephysicalmodelusedforsimulation... 000

5.1.1. Experimentalsetup... 000

5.1.2. Simulationoftheexperimentsetup... 000

5.2. Case2:Validationofthegenerationprocessoftherefinedtrajectory ... 000

5.3. Case3:Managingthecollisionavoidancebetween2?pallets... 000

6. Conclusionandfutureworks ... 000

Acknowledgements... 000

References... 000

1. Introduction

Nowadays,micromanufacturingsystemshavebeenadoptedasa newconcepttohandlenumerouschallengeslinkedtosustainability, globalizationandincreasedcompetition[1–3].Inordertoanswer the turbulent business environment requirements [4], micro manufacturingsystemsaredesignedtobeflexibleandreconfigur- ableatbothhardwareandsoftwarelevels[5,6].Insuchacontext,the microfactoryconcepthasemergedandwasfirstdefinedasasmall manufacturingsysteminwhichproductionequipmentisminiatur- ized to match product dimensions in order to reduce energy, material and space consumptions [7]. Furthermore, modular manufacturingsystemsarchitecturesatmachinelevelareapplied formostofthemicrofactoriescurrentlydeveloped.Theyprovide flexibleandreconfigurablemanufacturinglines[3].Insuchmodular microfactorysystems,eachworkstationisamodularitemthatis designedtoworkasastandaloneunitorasapartofaproductionline [8].Withastandardinterface,thisworkstationcanbepluggedat multiplelocationsinthemicrofactoryandconnectedwithothers modularitems.Thanks toexchangeableworkstations,themicro- factorycanproduceawiderangeofproducts.Furthermore,Wang et al. have [9] demonstrated that the design of manufacturing systemsbasedonthe sixRMSkeyrequirements(customization, convertibility,scalability, modularity,integrabilityanddiagnosa- bility[10])ismorecost-effectiveandprovidesabetteradaptability tomarketdemands.Forinstances,themanufacturingcapacitycan beeasilymodifiedbyadding,subtractingand/orchangingmodular items;the manufacturingsystemcanbecustomizedfora single productfamily.

At the sametime,smart surfacesusedtomanipulatemicro- objectshavebeendevelopedbothonthetechnologicalpointofview andon the controllevel[11,12]. Smartsurfaces canbeusedas conveyingsystemformicrofactory[13,14]andaredefinedasasetof identicalmodules,oftendistributedinaplanarmatrixlayout[13], that can berearranged in different 2D shapes. The goal of the presentedworkistobenefitfromtheSmartSurfaceconceptandthe modernmodularmanufacturingsystemdesigntoachieveflexibility andreconfigurability requirements in microfactories. Thispaper focuseson the advantage of using an electromagnetic modular SmartSurface(emSS)toflexiblyconveymicro-products,locatedon pallets,overaplanarsurface.Hence,ifproductionplanningchanges orperturbationsoccurrenceimpliesaproductroutingmodification, thepalletroutingcanbeautomaticallyreconfiguredwithoutthe needofchangingtheproductionline.Therefore,themainresearch focusoftheworkpresentedinthispaperistoproposeaframework to control and monitor the emSS developed in the Roberval laboratory,andtoperformfirstnumericalandexperimentaltests tovalidatethephysicalmodelandthemodulesaimingtogenerate palletpathandsimulatecollisionavoidanceofthisemSS.

Thispaperisorganizedasfollows.First,microfactoryconceptis introduced;researchworksdealingwithmodularmicrofactories and smart surfaces are analyzed; and a synthesis is proposed relatedtorequirements.Thenphysicalprinciplesandcharacter- istics of the proposed emSS are described in Section 3. The

proposedmonitoringandcontrolframeworkoftheemSSenabling reconfigurabilityandflexibilityisexplainedinSection4.Finally, Section 5 describes three validation steps of the proposed framework. In the first step experimental results obtain with therealprototypearecomparedtosimulationresultsinorderto evaluatethe coherency of theemSS physical model. The steps 2and3aretwonumericalusecases,performedwiththisvalidated physical model, allowing to evaluate strategies for trajectories refiningandforcollisionavoiding.

2. Microfactoryconceptandconveyingsystemtrends

2.1. Microfactorycontext

Earlierresearchworksinmicrofactorieshavefocused,onone hand, on downscaling manufacturing system components [7,15,16] (e.g.: miniaturized machine units and/ormicro-press, small-size manipulator, transfer arm or conveyor system to transport the components, and small-size assembly unit or micro-manipulator to assemble the components) and, on the otherhand, on integratingthe microfactory componentsintoa singleportablebox [17].Then, theresearchinterestsrelatedto microfactoryhaveevolvedtorespondtoflexibilityrequirements bydevelopingmodularconceptsinordertoorganizethehardware part of the microfactory [5,8,18]. Now, the majority of the microfactoriesaredesignedattheoutsetforrapidchangeofthe structurein termsofhardwareaswellassoftwarecomponents suchasTUTmicrofactory[5,37].Theexistingmicrofactorieshave demonstratedsomebenefitsinsavinginvestments,space,energy and resources.The nextstepis toenhance theperformance of micro manufacturing systems such as the design of a fully automatedsystemswithlimitedhumantasks,aswellasmanaging theemergingmodularmicrofactoriesinordertoinsureflexibility inbothroutingandproduction[3,16,19].

Simultaneously,organizationsystemsfordataandinformation control have been developed to manage the microfactory according to different approaches found in the literature. For example,Fatikowetal.[20]andGendreauetal.[21]haveproposed amethodologytodesignmodularcontrolarchitectureadaptedto micromodularmanufacturingsystems.Inaddition,Descourvie`res etal.[22]andMauchandetal.[23]havedefineddatamodelsto managethereconfigurationofmodularmanufacturinglayout.This meansthatincaseofreorganizationandreconfiguration,support functions such as product/process modeling, process planning, production and capacity planning, control of processes and production,andlogisticshavealsotobeadaptableandchangeable [24,25].

Inthecurrentcontextofmass-customization,whereproducts have to be personalized to answer customer needs, modular manufacturingsystemsareameantoimproveindustrialsystem flexibilitybyenablingarapidchangeofproduction.Thus,insucha context of production change, material handling and product transferprocesses havetobeadapted (reconfigurable),and the

producttrajectorieswithinthemicrofactorymust beflexible to producemultiplevariantsofoneproduct.

2.2. Newtrendforflexibleandreconfigurableconveyingsystems:

ModularmicrofactoryandSmartSurface(SS)concept 2.2.1. Modularmicrofactories

Existing modular microfactories can be classified into two categoriesbasedontheirarchitectureforaggregatingmodules.In thefirstcategory,themodularmicrofactoryismadeofmodular blocksincludingworkstationsandpartofthetransfersystemas illustrated in Fig. 1. In the second category, machines and/or componentscanbepluggedorremovednearaflexible,butfixed, transfersystem(planarsurface)asshowninFig.2.

Forproductconveyance,thefirstcategoryusesconventional downscaledlineartransferlinesplitintoindependentelementary unitsthatcanbecombinedinsideamoduleinordertosetupthe desiredlayout.Themodularmicrofactoryoperatingsystemisthen composedofasetofseveralmodulesarrangedsidebyside.The interfacesandsetupaswellasthegeometryofsuchmodulesare standardizedinordertoguaranteesmoothintegrationorexchange withoutcostlytime-intensiveramp-up.Thus,themodulescanbe freelyandeasilysetupondifferentlinelayoutsandthecontrolof transferunitsisrelativelysimplesinceitispredefinedandlinear.

Furthermore,thelineartransferlineallowshighthroughputsand accuracy of pallet positioning. However, in case of machine breakdown or full capacity, the entire production process is blocked.Therefore,inthistypeofmodularmicrofactory,flexible and efficient production can by realized with ‘‘bypassing’’

conveyorthatcanbeaddedtoallowthecarriers/palletsalternative routes[26].Besides,incaseofreconfigurationofthelayoutfora newprocessoranewproduct,manualinterventionisrequiredto

adapt thelayout,takinginto accountthepredefinedshape and orientationofeachelementarytransferunit(inline,withcorner, etc.).

The second category of modular microfactory is ‘‘flexible transfer line’’. In this case, the conveyance system is a planar surface on which a number of carriers or pallets can move independentlyandfreelyasshowninFig.2.

Thecarrierorpalletisdefinedas‘‘active’’whenitisenergizedto poweritsownactuatorsand‘‘passive’’intheothercases.Forthis typeofmodularmicrofactories,modularityisduetotheabilityto plugorremovemachines/stationsanywherearoundtheconvey- ancesystemsaccordingtotheproductionneed.Thisisachievedby usingstandardizedinterfaces.Inthiscase,abaseunitortableacts as a service module that supplies power, network connection, pressuredair...tootherconnectedmodules.Thisconceptprovides routing flexibility and reduces time of layout reconfiguration.

Nevertheless, this type of modular manufacturing system is limited to the dimension of the planar surface that is not adjustable suchas AMMS [8] and Miniprod [27]. Moreover, in thistypeofsystem,thewiredpowersupplyofthecarrierslimits the area of displacement. Contrary to AMMS and Miniprod systems, the AAA systemis built froma collectionof modular baseunits,platentiles,andbridges.Theworkstationmodulesare setuponthebridgesatanylocations.Theaimofthissystemisto enablerapiddeployment.

Thissecondcategoryrespondstotheflexibilityrequirements, nevertheless,thereconfigurabilitycouldbeenhancedbyincreas- ingmodularityoftheplanarsurface.

2.2.2. SmartSurfaceconcept

Uptoourknowledge,thereisnoconsensusintheliteratureon the concept of ‘‘Smart Surface’’(SS) [13,28–30]. In the present Fig.1.Exampleofmodularmicrofactorywithlineartransferline.

Fig.2.Exampleofmodularmicrofactorywithflexibletransferline.

paper,a‘‘SmartSurface’’oran‘‘activesurface’’isdefinedasamicro conveyancesystemthatcancontrolthepositionofamicroobjectand allowsitsdisplacementwithinaplanar surfaceinordertoreacha desiredpositionandorientationasshowninFig.3.

InFig.3[13],aSSisbasedonamatrixofcellswhereeachcellis composedofa microactuator,amicro sensorand aprocessing unit. Indeed, by exploiting airflow [31], or electromagnetic [29,30,32]principle,anarrayofmassivelyparallelactuationcells can be successfully employed for micro conveyance of small objects.[31,32]showedthatSmartSurfacearchitecturesarewell adapted to flexible conveyance platform since it is possible to orientateand displacein parallelseveral objectsorpallets that carryobjects.

Yahiaoui et al. [12] recently showed the development of actuators array on a 9mm9mm surface composed of 88pneumaticmicro-conveyorsthatcangeneratetiltedair-jets infourdirections.Inthe‘‘SmartBlock’’project,aself-reconfigur- able conveyor based on sliding blocks uses the pneumatic actuators described in [12]sothat each blocksupportarrayed actuatorsonitsupperface.In[14],electro-permanentmagnetsare usedtoslideoneblockalonganotheroneinordertoconfigurethe conveyancesystem.Inthefuture,theelectro-permanentmagnets willallowtoautomaticallyandeasilyreconfigurethesurfaceina newgeometricallayoutwithouttheneedofthehumaninterven- tion.

Finally,theconfigurationofaSScanbesetupasalinearor flexibletransferlineormixed,thankstoitsdesign.Ononehand,

the matrix structure of the SS allows its scalability and modularity.Ontheotherhand,theintegrationoffunctionssuch as recognition, conveyance and positioning increases the SS robustness.

2.2.3. Synthesisonthecharacteristicsandperformancesof conveyancesystems

Fromtheprevioussections,itisobviousthattheconveyance system plays a significant role to enhanced flexibility or reconfigurabilityofmicrofactories.Ifaconveyancesystemreacts autonomouslyandimmediatelytochanges(forinstances:module breakdowns,aswellasintroductionofnewmodules,etc.)without crash,thiscanreducetheeconomicimpactsintermsoflogistic costs and delivery time. The capability to support dynamic workflow reconfiguration without shutdown the entire system appears as an important requirement in a flexible conveyor system. Therefore, specific criteria (e.g., flexibility, reconfigur- ability, pallet (carrier) type and workstation setup) have been chosentoidentifytherequirementsforabetterimprovementof theconveyingsystemconcept.

Regardingflexibility,utilizingthe possibilitiesof flexiblyre- routing for alternative paths could enable avoiding interrupt production,rapidresponsetounexpectedequipmentfailuresor limitthehumaninterventionsforchangingstructurelayout.

Regarding reconfigurability, the interface connecting the transfer units/modules have to be able to integrate modules rapidlyand precisely. Themicrofactory design has totake into accountthewaytoconnectrapidlymodulesandtooptimizethe materialflowsthroughthemicrofactory.

A synthesis of the characteristics of existing conveyance systems according to thedefined criteria is shown in Table 1.

[18]hasshownthattheAAAsystemisflexiblenotonlyinthatit dealswithvariabilityintheassemblyprocess,butalsothatitcan rapidlyrespondtochangingmarketpressures.Contrarytoothers existingconveyancesystems,SSseemstobecomparabletoAAA sincebothofthemenablemodularity,scalability,integrability,and customization.

Asstatedin[33],thenextfactoriesgenerationmustbenotonly reconfigurablebutalsoflexible.Forthat,conveyingsystemshould alsobeflexibleandconfigurableforquickalternativeprocessflow.

Theaimofrecentresearchesisthentodevelopamodularconveyor thatisthencomposedofaseriesofsimilarcellsthatcandetect objectswithsensors,movethemwithactuatorsandcommunicate witheachother(suchas SS)toforma flexibleconveying path.

Currently, all modularconveyors are connectedto a baseunit, whichis anelectroniccontrolboard thatcommunicatedwitha hostcomputer.Afurtherideaisthatthemodularconveyorwillbe Fig.3.SmartSurfaceconceptinmicrofactorydevelopedfrom[13].

Table1

Conveyancesystemcharacteristicsofexistingmodularmicrofactories.

Conveyance systemtype

Concept Flexibility Reconfigurability Pallet

type

Workstation setup Integrability

(interfaces)

Conveyor structure Lineartransfer

lineseeFig.1

TUTmicrofactory[37] Linecustomized Plugandplay(module-module) Adjustablemodular Passive predefined Boschleandesktop[38] Linecustomized Plugandplay(module-module) Adjustablemodular Passive predefined Modulebased

microfactory[39]

Linecustomized Plugandplay(module-module) Adjustablemodular Passive predefined

Flexibletransferline seeFigs.2and3

AMMS[8] Routing Plugandplay(module-table) Fixed Active fixed

Miniprod[27] Routing Plugandplay(module-baseunit) fixed Active fixed

AAA[40] Linecustomized,

Routing

Plugandplay(module-baseunit) Adjustablemodular Active predefined SmartSurface[13] Linecustomized,

Routing

Plugandplay(module-baseunit) Adjustablemodular Passive predefined

able to self-reconfigure to form manufacturing line [14]. That means that a standard communication interface has to be developedforlinkingamodularconveyortoanother.Thepurpose istobuildamodularconveyancesystemthatcantransportitems independently and simultaneously to different workstations, allowingdifferentprocessestobecarriedoutoneachitem,and resultinginindividuallycustomizedproducts.

TheelectromagneticmodularSmartSurface(emSS)basedon conveyors developed by [29,30] appears as a well-structured combinationofmodularconveyorsandSmartSurfaceformodular microfactory.Theresearchaimstodesignamodularconveyance system adapt to different transfer system requirements (e.g., positioningaccuracy,rapidity,flexibility) andotherapplications (e.g.,microcoordinatemeasuringmachine)insuchmicrofactories.

Electromagneticactuatorscombinedwithanactiveplanarsurface seemtobesuitableforbuildinglightmobilemodules.Withpassive palletsandmatrixdesign,numerouslayoutconfigurationscanbe studiedforbestcostoptimization.

3. Proposedflexibletransferplatformformicrofactory:An electromagneticmodularSmartSurface

In this section, an electromagnetic modular Smart Surface (emSS)isproposedbased onprevious worksaimingtodevelop electromagneticplanarpositioningsystems[29,30,34].

3.1. Principleofactuation

Long timeresearchworks oftheRoberval laboratoryaimto develop electromagnetic planar positioning devices [34] and recentlyextendtheconcepttoaconveyanceplatformincluding flexible planarelectromagnetic conveyors [29,30]. The working principle of the electromagnetic conveyor system is based on

Lorentz force. As depicted in Fig. 4(a), when two sinusoidal currents,I1andI2,witharelativephaseshiftof

p

=2aresuppliedto twofixedPlanarElectricDriveCoils(PEDCs)placedunderneatha PermanentMagnetsArray(PMA),electromagneticforcesoccuron thePMA,duetotheinteractionbetweenthemagneticfieldfrom thePMAs andthecurrentsinthecoils,and generatesthePMA motion[34].Basedonthisprinciple,a3DOFxy

u

positioningdevice has been developed which is capable of performing linear displacements along x- and y-axis on a few millimeters stroke andarotation

u

alongtheaxisperpendiculartotheplane.Thecross sectionalviewofasingleaxisLinearMotor(LM)isshowninFig.4 (a).Themovablepartconsistsofasiliconpalletdesignedinthe formofacrosstointegratefourPMAsarrangedatthefouredgesof thestructure[34,35].EachPMAconsistsof14PermanentMagnets, each of dimensions 6mm1mm1mm (see Fig. 4(b)). Two fixedPEDCsareplacedunderneatheachPMA.Thenthepositioning deviceintegratesfourLMswhereLM1andLM2areusedformotion alongx-axiswhileLM3andLM4areusedformotionalongy-axis.

Inthispositioningdevice,eachPECDisdedicatedtoonePMA, formingaLM,andthestrokeofeachLMislimitedduetothesizeof eachPEDC(eachPEDCisspatiallyseparatedfromeachother).In ordertoovercomethislimitation,atwolayercoilsbaseddesign2D actuatorhasbeendeveloped[29,30].Thisnewdesignuses two layersofcoils:Layer1andLayer2.ThecoilsinLayer1andLayer 2areoverlappedandarrangedorthogonallywithrespecttoeach other.Withthisnewdesign,accordingtotherelativepositionof thePMAinfrontofthecoils,eitheratranslationofthePMAalong x-axisisrealizedsupplyingthecoilsoflayers1,oratranslationof thePMAalongy-axisisperformedsupplyingthecoilsoflayer2 (seeFig.5).Withthiscoildesign,thedisplacementstrokecanbe increasedinproportiontolengthofthecoils.Nevertheless,the strokeincreaseimpliesanincreaseonthecoilsresistanceandthen anincreaseofenergyconsumption.

Therefore,thetwolayerprincipleisfurtherextendedtodevelop alongrangeelectromagneticconveyorsystembasedonamatrix design. Each cell of the matrix is made of overlapped and orthogonallyarranged2Dcoilsaspreviouslydescribed.

Each elementary cell of the matrix can be managed and controlledindependentlybyactivatingtherequiredcoilsdepend- ingonthepositionofthepallet.Inthesamemannerasforthe previousstage,fordisplacementofthesiliconpalletalongx-axis, two opposite LMs (LM1 and LM2) in Layer 1, are supplied simultaneously with currents(see Fig. 6(a)). In the same way, LM3andLM4inLayer2aresuppliedwithcurrentsatthesametime toperformmotionalong y-axis(seeFig.6(b)).AllthefourLMs (LM1,LM2,LM3andLM4)aresuppliedindependentlyinorderto performplanarmotions(seeFig.6(c)).Thatmeans,ateachinstant and foragiven positionof themobilepart, specificcells ofthe matrixarerequiredtobeactivatedin ordertogeneratemotion while the remainingcells are no more activated.This strongly reduces theelectricalresistance oftheactivatedcoilsand thus, energyconsumptionislowespeciallyoverlargesurfaces.Besides, the matrix design configuration ensures the possibility of controllingandmanagingseveralpalletsatthesametimewithout anymanualintervention.

Moreover,theemSSisdesignedtobesuitableforreconfigurable manufacturing systemparadigm[10] as arguedin Table 2 and Section2.2.3.

3.2. Controlofroutingflexibility

In order to analyze the emSS control, an example of manufacturingagearisconsidered(Fig.7).Arawmaterialblock isplacedabovethepalletattheinputstation(Station0).Thefirst task is machining, realized by the milling machine located at StationA.ThepalletishencetransferredtoStationAthroughpath Fig.4.Electromagneticactuationprinciple(a)Singleaxislinearmotor(b)3DOF

electromagneticpositioningdevice.

Fig.5.Twolayercoilsbased2Dactuator.

Fig.6.Workingprincipleoftheconveyanceplatform.

Table2

RMScharacteristicsenabledbytheelectromagneticmodularSmartSurface.

Flexibility Convertibility Diagnosability Reconfigurability Pallet

type

Workstation setup Integrability(interfaces) Conveyorstructure

Linecustomized, routing

Linearor/andflexible transferline

Detectionofcell failureand pathredefinition

Plugandplay(module-baseunit), distributedcontrol,information systemmodules

Adjustable modular

Passive Predefined

Fig.7.Flexibleelectromagneticconveyorplatform[Arora2014].

(1a).The machinedgear is geometrically controlledby a micro measuringmachinelocatedatStationB.Thepalletistransferredto StationBthroughpath(1b).Thenthe finalproductissenttothe outputstationatStation1followingpath(1c).Therefore,theoverall processpathtobefollowedis(1a)!(1b)!(1c).Theshortestpath (3)!(1c) can befollowed ifonly measurement task has to be performed.Usingthisshortestpathinrequiredmanufacturingplan savesenergyandtimetofinishtheprocess.Moreover,ifdamaged cellmakesitimpossibletousepath(1b)thenanalternativepath (1a)!(2)!(1c)canbefollowedforthesmoothfunctioningofthe process. The whole process can be thus continuously realized withoutanymanualintervention.

Thisexampleillustratestheneedofaframeworktomonitor andcontrolthepalletrouteswithintheemSS.Inparticular,the framework needs to coordinate the pallets activities on the platform,tohandleunpredictedconstraintssuchascellsfailure, and to manage complex trajectories. To ensure thescalability, flexibility and reconfigurability requirements of the emSS, the frameworkmustincludethefollowingfunctionalities:

Easilyadd,modify,eraseanobject(acell,apallet,aproduct,etc.) inthedatabase;

AllowuserstodesigntheemSSviaagraphicalrepresentation;

Updatethedatabaseconcerningthechangesinthelayoutofthe emSSafterreconfiguration;

Identifycellfailures,andrecalculatetheoptimalpallet’srouteif needed;

Evaluate and optimize the virtual preconfigured platform to reduceengineeringcosts.

Themainissuesarisingfromsatisfyingthesefunctionalitiesare:

Having an extendable database, and based on it a module allowingreconfiguration oftheplatformtakingintoconsider- ationchangesinthenumberofavailablecells;

Theneedforaccurateandrealtimecommunicationbetweenthe physicalplatformandthecontrolsystem;

Therecalculationtimeofanewrouteshouldbefasterthanthe palletdisplacement;

Takingintoconsiderationtheroutesofallpalletstoavoidcollision.

4. Proposedframeworkfortheelectromagneticmodular SmartSurface(emSS)managementandcontrol

In this section,and in order toanswer theissues identified above,anovelframeworktoachievetheemSSmanagementand controlispresented.Thisconsistsof:

communication with Enterprise Resource Planning (ERP), in ordertoreceivetheproductionplan;

layoutdefinitionandreconfigurationdueto:cellsmodifications (addedordeletedcells),newproducts(hencenewproduction line);

pallets monitoring and path optimization to generate pallet trajectory;

dataexchangeforemSScontrol.

4.1. ManagemanufacturingprocesswithintheemSS(A0)

Withinanindustrialcompany,theproductionprocessplanning is usually managed by the ERP. In the activity A1—Schedule production(seeFig.8),theproductionplanningisproposedbyERP andcanbemodifiedbytheproductionsupervisorwhenthereare inventoryadjustments,scrappedparts,lostpartsorpriorityorders.

Thisproduction planningtogetherwiththeemSSlayout are consideredastheinputstomonitorandcontroltheemSStaking intoaccountspecificobjectiveslikedefiningtheshortestpaths,the least energy consumption, or avoiding collision.An application called the Monitoring & Control Software (MCSo) has been developedtomanageandsupervise themonitoring andcontrol ofthepalletsactivitywithintheemSSviaauser-friendlyinterface.

The nextsection focuses on functionalities associated withthe monitoringandcontroloftheemSS.

4.2. MonitorandcontroltheemSS(A2)

The monitoring and control activity is structured into four tasks: update database, generate pallet trajectory, model and simulatetheemSS(Fig.9).

Through the activity A21—Update database, the database collects the production planning (operation/workstation sequences) and layout definition (emSS dimension and shape;

cells size and state; workstations footprint, dimension and location;palletcharacteristics...).Aclassforeachtypeofobjects suchaspallet,cellandworkstationisdefinedwiththeirattributes suchastheidentityIdandthestate.Thisstateattributecanbe active, inactive or defected. The database is automatically or manuallyupdatedincaseofreconfigurationoraftermaintenance intervention(Fig.9).

Takingintoaccountdatabaseinformationandobjectivessuch asproductiontime,energyconsumptionandcollisionavoidance, thedisplacementofeachpalletisgeneratedandoptimizedwith activityA22—GeneratepallettrajectorybyMSCo_algorithmsand MSCo_collisiondetectionmodules(explainedindetailsinSection

Fig.8.IDEF0A0ManagetransferprocesswithintheemSS.

4.3). Taking into account database information and geometric model, the activity A23—Model emSS (explained in details in Section4.4), definestheemSSinstantiatedmodel (locationand stateofthecells,workstationsandpallets).

ThisemSSinstantiatedmodelandthedefinedtrajectoriesare simulatedintheactivityA24—Simulate theemSS(explainedin detailsinSection4.5),byincludingphysicalmodel.Theresulting refinedtrajectoriescanbevalidatedorinvalidated.Inthislastcase, theinvalidatedtrajectoriesareeliminatedforthenextcalculation stepofthepallettrajectory.

4.3. Generatepallettrajectory(A22)

Thegeneratepallettrajectoryactivityisexecutedfollowinga workflowasshowninFig.10.

Inordertocontrolthepalletpathfromitscurrentpositiontoits destination(Fig.11,stage1),thefirststepA221aimstodefinethe palletrawtrajectoryasasetofviapoints(locatedatcellcenters) takingintoaccounttheobjectives(Energyconsumption/shortest path)andthedefectedcellsasavoidedcells(Fig.11,stage2).Atthe secondstep,thecalculationofpalletpathsisconsideredasagraph path-planningproblem.TheactivityA222—Interpolatetrajectory with different methods, refines the raw trajectory in order to obtain a smooth desired trajectory with interpolated points.

Differentinterpolationfunctionscanbecompared(spline,Beziers, etc.)andchosen(Fig.11,stages3).Therefinedtrajectoriesofeach palletarethenusedintheactivityA223tofindtheintersection pointsbetweenapalletandothers(Fig.11,stages4).Intheactivity A223—Choosetrajectoryanddefinevelocity,theinputisfiltered byeliminatingtheinvalidatedtrajectoryC3thatisdefinedinthe Fig.9.IDEF0A2Monitor&controloftheemSS.

Fig.10.IDEF0A22Generatepallettrajectory.

activity A24—Simulate emSS. The ‘‘MCSo_Collision detection’’

module is used to detect the possibility of collision between two pallets and disqualify the undesirable trajectories before validatingthefinaltrajectory(Fig.11,stages5).Thismoduleisalso usedtodetectinrealtimethecollision.Inthecaseofapossible

intersection,thismodulecouldtakedecisionsonthepalletcontrol (forinstances:varyingthevelocityofthepallet,respectingthedue dateandthepriorityofoneproductorder).

4.4. ModelemSS(A23)

Inthissection,themodelingoftheemSSisdefinedindetailsto givetheassumptionsandthemodelcharacteristics(corresponding totheactivityA23inFig.9).Inasecondtime,thegridbasedemSS representationprincipleisexplained.

4.4.1. Physicalsystemmodel

TheemSSmodelingisdividedinto2stagesasfollows:

TheLorentzforceanalyticalandmechanicalmodels:

Knowingthattheelectromagneticforce,whichistheresultof theinteractionbetweenmagneticfluxdensityfromthePMAand currentsinthePEDCs,dependsonthenumberofturnsinacoilof thePEDCbelowthePMA,thisforcewillimmediatelydropwhen apalletpassesoverthetransitionzonebecauseoftheabsenceof coils(asshowninFig.12andseeSection3.1)betweentwocells.

Inthiscase,themathematicalmodelneedstotakeintoaccount thiseffect.Thedimensionofthetransitionzonesisdesignedto havea fixednumberofthepairofturncoilsforensuringthe continuousmotionofapallet.AmatrixIdenotesthecurrentsent totheemSScell.ThedimensionofthematrixI_xforthexdirection andthematrixIyfortheydirectionincludesthenumberofturn coilsofallthePEDCandtheonesoftransitionzonesforeach direction.Themagnitudeofsinusoidalcurrentsinjectedintothe turn coils at the transition zone equals 0A. The maximum allowable magnitude of the sinusoidal currents injected into coilsis0.8A.Thefrictionandadhesionforcesinthetranslationof themobilepartinvolvethefinalcomputingforceduetothetotal netweightofpalletwhiletransportingproductsoritems.

Thegraphicalmodel

The ‘‘MCSo_Geometric modeling’’ module is written in Matlab/Simulinkfor modellingthe emSSand thepallets. The current layout model will be updated and displayed by the

‘‘MCSo_HMI’’.

4.4.2. GridbasedemSSmodel

DuetotheemSSlayout,a2Dgridbasedmodelisapplied.The modelisbinary(eachgridcellcontainseitheranobstaclelikethe footprintofworkstations,etc.orfreespace)andassociatestoeach cellacostreflectingthedifficultyofmovingapalletfromittoits neighborcells.Forresolvingtheproblemoffindingtheshortest Fig.11.Theactivitiesfromcalculatingtrajectorytocheckingcollisions.

Fig.12.Palletatthetransitionposition.

path,thisgridisconvertedintoagraphinwhicheachnodeofthe graph denotes the cell center coordinates. The cost value for movingonecelltoitsneighborsrepresentsthelinkofthegraph.

Manyalgorithmsexistforgraphshortestpathidentificationlike Dijkstra’salgorithm,A*,Floyd–Warshallalgorithm,etc.[36].Those algorithmsareusedtofindtheappropriatesolution.

To simplify thecalculation, theindex of a cell in theemSS matrix represents its identity attribute in the MCSo database.

There,theemSSismodeledbyauniformgriddatastructureandits cellsareindexedasshowninFig.13.Inordertobeadaptedtoall emSSlayouts,theemSSmatrixisdefinedasa squarematrixof ordernwherenisthemaximalvaluebetweenthenumberofcells inthexdirectionandinthey direction(see Fig.13). Thezone withoutcellsisrepresentedasdefectedcellsintheemSSmatrix

and these cells have their own identity. The index system is automaticallyrecalculatedwhenthelayoutismodified.

4.5. SimulateemSS(A24)

TheemSSsimulationactivityincludesthesimulationofcells state (A242)andthepalletdisplacement(A243)asdetailedin Fig.14.

IntheactivityA241—Generatecells control,the‘‘MCSo_con- troller’’ moduleis usedtoconvert theinputparameters tocell control(currentmatrix).Eachpalletneedsatleastoneortwopair ofcellsformovinginonedirectionanditonlyworksifcellsare activeandpower.Forprecisepositioning,thefrequency

v

ofthe injectedsinusoidal currentsI1 and I2 in thecells is adjustedin Fig.13.Differentplatformshapeswithindexedcells.

Fig.14.IDEF0A24SimulateemSS.

function of travel time and the distance between the current positionandthedestination.

The activity A242—Simulation cell state defines cells layer statesinthe‘‘MCSo_HMI’’moduleasfollows:

Activestatewhencurrentsaresenttocell.Eachlayercellhasits ownstate(theredcolorforthexdirectionandthebluecolorfor theydirectionasshowninFig.15);

Inactivestate;

Defectedstatewhencellisdefectedoroccupied.

‘‘MCSo_HMI’’alsodisplaystheformanddimensionoftheemSS and thepallets given by the activityA23 at each time step of simulation.Whenthecurrentsaresenttothecell,the‘‘MCSo_- Physicalmodel’’modulethatisimplementedinsidetheactivity A243,isusedtocalculatetheelectromagneticforcesappliedtothe pallets.SincetheemSSisamatrixofcells,twomatrixofcellstates are needed for modeling the two layers of emSS as shown in Fig.15.Theidx,dimXdenotetheindexofacellinthesquarematrix andthenumberofcellsinthexaxis,respectively.

Inordertocontrolsimultaneouslyseveralpallets,thecontrol signal is defined as a combined matrix. For each step time of simulation, the pallet position and cell state are sent to the databaseformonitoringthepalletalongitstrajectory.

TheactivityA244isusedtoselectthevalidatedtrajectoriesthat respecttheimposedtraveltime.Simulatingthedynamicbehavior ofemSSanditscontrolsystemallowsustostudyandtochoosethe appropriatetrajectoryforeachpallet.

The ‘‘MCSo_Position checking’’ module has the function to controlandcheckthepalletpositionduringthesimulation.Dueto somecontrollimitations,thepalletcanbelocatedinazone(for instances:transitionzone,andsoon)whereitisimpossiblefor pallettoleave.Inthiscase,the‘‘MCSo_Positionchecking’’module willtakedecisiontostopthepalletsimulationandstatetheinvalid trajectorytotheactivityA22.

Supervisorcanalwaystakethecontrolofonepalletbychanging thecontrolmode(frommanualmodetoautomaticmode).

5. Experimentsanddiscussion

InordertojustifytheworkflowintheactivitiesA22,A23and A24,threestudycasesarepresented.Itallowsnotonlytocompare thesimulationresultswiththefirstexperimentalsetupbutalsoto improve the layout design and pallet control in the future.

Hereafter,differentcaseswillbepresentedtovalidatethedifferent controlfunctionalities.

5.1. Case1:Validationofthephysicalmodelusedforsimulation In this study case, an experiment is setup to compare the experimentalresultswithsimulationresultsandtoallowcontrol andobservationofthebehavioroftheemSSwhenapalletcross overthetransitionzone.ThisaimstovalidatetheemSSphysical modeldefinedforsimulationpurposes.

In simulation, some relevant characteristics are taken into account:

Fig.15.Apalletcontrol.

Fig.16.Experimentalsetup.

a highcurrent maydamage thecoilsor degradetheresidual magnetizationofthePMAs;

theairgapbetweenthemobileandcoilinthelayer2isbigger thantheairgapbetweenthemobileandcoilinthelayer1as showinFig.12;

the unbalance of electromagnetic forces applied on the two extremitiesofapalletatcertainareasintroducestherotationof thepallet;

5.1.1. Experimentalsetup

TheexperimentalsetupshowninFig.16hasbeenrealizedin ordertovalidatetheemSSmodeling.

InordertovalidatetheemSSmodel,aprototypeoftheemSS formedofa55squarematrixhasbeendeveloped.Theoverall surface dimensions of the conveyor platform are 130mm130mm.Thethicknessofcopperofthecoilsineach layeris35

m

m.Thedistancebetweenthetwolayersis282

m

m.A 130

m

mthickglasslayerhasbeengluedtothebottomsideofthe PMAsupportstructure.Thisglasslayerisusedtoachievesmooth motionofthepalletovertheemSSandtopreventdirectcontact between the current carrying coils and magnets that helps to minimizethefriction.Therefore,theairgapd1andd2(asshownin Fig.12)arerespectively130

m

mand447

m

m.Thetotalweightof thepalletis3.7gandtheadhesioncoefficient

m

adhesionbetween theglasslayerandthePCBsurfaceequals0.23.

A computerequippedwithadataacquisition board(NIPCI- 6733)withaLABVIEWinterfaceisusedtogeneratetwocontrolling voltages for each axis. These voltages are then converted into currentsusingvoltagetocurrentconverters.Thevoltageinputand currentoutputofboththeconvertersareintherangeof[10V, 10V] and [3A,3A] with a bandwidth of 50kHz. For the displacementmeasurement, acamerahasbeenfixedabove the emSSinordertoobtainimagesofpalletwhilemoving.Theimages ofsize1024768pixelshavebeencapturedbythecamera.

Currentsneededforthemotionsalongy-axisarehigherthan theonesusedformotionsalongx-axisduetothelargerdistance betweenthePMAsandLayer2inordertoequalforcesgenerated during the motion along both axes and obtain similar motion characteristics.Therelationbetweenthesetwocurrentsisgivenin thefollowingequation[29,30]:

Ix¼0:53Iy (1)

Anexperimenthasbeencarriedouttoshowtheabilityofthe experimental emSS to perform long stroke displacements by crossingseveralcellsandalsotodeterminethevelocityinthex andydirectionviadecoupledorcoupleddisplacement.Inorderto

minimizethenumberofcontrolsignals,ithasbeendecidedto onlycrosstwocellsineachdirection.Asthemaximumallowable magnitudeofthesinusoidalcurrentsinjectedintocoilsis0.8A, themagnitudeofsinusoidalcurrentisIx=0.4Aformovinginthex direction and I_y=0.74A for moving in the y direction. The frequency of these currents for both directionis fx=fy=4Hz.

Fig.17showsthetwotypesoftheexperimentaldisplacementsof thecrossstructurefromcell(2,2)tocell(3,3).Theinitialposition is[x0,y0]=[7.3;8.03].Thefinalreachedpositionsarerespectively [xfd;yfd]=[38.96;38.83] for the decoupled displacement and [xfc;yfc]=[39.65;38.33] for thecoupled displacement (diago- nal). The travel time in the decoupled displacement and the coupleddisplacementisrespectively7.2sand3.73s.Theresults demonstrate the possibilities to move in different directions withintheconveyanceplatform.Thedifferenceobservedforthe finalpositionisduetothefactthatexperimentshavebeencarried outinopenloop.

VelocitiesinbothdirectionsaredisplayedinFig.18.Forthe decoupled displacement,mean velocity is 8.25mm/sfor x axis displacementand8.42mm/sforyaxisdisplacement.Themeans velocity for the diagonal displacement is obviously ffiffiffi

p2 times higher than the decoupled velocity and is experimentally 11.76mm/s.

In the case of coupled displacements, totally, 16 cells are poweredwhileonly8cellsarenecessarytobepoweredinthecase ofdecoupleddisplacementtoperformthedisplacementfromthe startingpointtothearrivalpoint.Inordertoestimatetheratioof energyconsumptionbetweenthesetwotrajectories,thenumber ofsuppliedcoilateachtimehasbeentakenintoaccountandshow approximately energy consumption two times higher for the coupleddisplacement.

5.1.2. Simulationoftheexperimentsetup

In the emSS model, the electromagnetic forces applied on 2diametricallyopposedextremitiesofthecross-shapedpalletare consideredtobesymmetric.Thatimpliesthattheeffectofrotation onthepalletcanbeneglectedandtherearenodeviationsofthe directionofapalletwhileitmovesalongonesingleaxis.

Fig.19illustratestheHMIofthemodeledemSSusedbythe supervisor for controlling one pallet in open loop. In order to validatetheemSSmodeling,afirstsimulationisrealizedinopen loopasintheexperimental setup.Inthis simulation,apalletis movedin thediagonaldirectionbysupplying currentsintothe emSS cells withthe amplitude of I_x=0.4A, I_y=0.74A and the frequency of fx=fy=4Hz.Itmoves for a distance of30mm as illustratedinFig.20.Alldesignandmechanicalcharacteristicsof Fig.17.Coupledanddecoupleddisplacements.

Fig.18.Velocitiesforcoupledanddecoupleddisplacements.

theemSSsuchastheair gaps,theadhesioncoefficient,etc.are respectedinthemodel.

The simulation shows that the pallet moves in the desired directionandreachesthearrivalpointatthetimeof3.702sandit hasthesameaveragespeedforthexandycomponents,equalsto 8.1mm/s.Thissimulation resultis ingoodagreement withthe experimentalresult.Itallowstovalidatetheelectromagneticand mechanical model and enables the simulation of different trajectories. The velocity curves in the x and y directions are showninFig.21.Whenthepalletcrossesthetransitionzone,the velocityvalues aredroppedas expected.Due tothemagnitude differenceofcurrentandtheairgapsinterval,thevelocitycurvesin thexandydirectionsaren’tsimilar.

5.2. Case2:Validationofthegenerationprocessoftherefined trajectory

AsmentionedintheactivityA22,thepallettrajectoryinthe emSS is the curve thatlinks the starting point [x0, y0] tothe

arrival point[x_d,y_d].Inthiscase study,agivensquare matrix platformthatispartitioned intoauniformgrid squarecells,is considered as the emSS simulation. As the emSS is a square matrix, theindexof eachcell will beautomatically generated and assigned as the identifier of cell. The ‘‘defected’’ state denotes the occupied areas or failure cells and they are represented in whitein theemSS representation as shownin Fig.22withtheirown identifier.

For demonstration purpose, the cell centers correspond to potential via points and a graph is defined for finding a path linkingthestartingpointtothearrivalpointcrossingthesevia points.Thedistancebetweenacellcenteranditsneighborsand thetraveltimearetheoperationcostsforevaluatingthetransfer fromapointtotheothers.Inthesimulation,theFloydWarshall algorithm [36] is used to find the shortest path via a graph becauseitissuitableforrapidexecutioninMatlabwhileusing matrix representation. For simplification, the cost values are calculatedinfunctionofthedistance.Thecalculationof pallet path is repeated andtested for differentemSS configurations.

Fig.19.HMIformanualcontrolofapalletinopenloop.

Fig.20.Palletdisplacementinopenloopwithfx=fy=4Hz,Ix=0.4A,Iy=0.74A.

Fig.22(a)showsthefoundpaththatlinksallthecellcenters.In thecasewherethecellstateattheindex20isdefected,thepath iskeptasthepreviouscasebecausethiscellisnotusedtomove thepallet(Fig.22(b)).However,whenthecellattheindex19is removedfromtheemSS(Fig.22(c)),thealgorithmrecalculates automaticallyanewpath.Fig.22(d)showstherecalculatedpath whentheremovingofacellattheindex9occurs.Theuseofthe graphmethodfor calculatingthepath allowsthe flexibilityof pallettravelswithintheemSSsinceitenablesautomaticrouting reconfiguration.

Astherawtrajectoryisalistofcellscenters,thesesimulated pathsshowthepossibilitytofindrawtrajectoriesfromthecurrent positiontoa defined destination.The previous experiment has proventhatthesetrajectoriesareaccessible.Inordertooptimize thepalletpaths,interpolationmethodsareusedtocalculatethe refined trajectories considering specified constraints such as energyconsumptionortraveltime.

Fig.23illustratesthewaytosmooththerawtrajectoriesusing the interpolation method—Cubic Hermit and then define the refinedtrajectory.Inthefuture,differentinterpolationmethods will be studied to respond to the specified requirements (for instances:minimizethe pallet’sstrokeor energy consumption, etc.).

5.3. Case3:Managingthecollisionavoidancebetween2pallets Asmentionedintheprevious Section4.2.,pathrecalculation allowsavoidingthecell at‘‘defected’’state.Thenextstepis to avoidcollisionsbetweenpallets.Thetraveltimeofapalletfroma workstation to another is defined by production schedule. For simplification,inthiscasestudy,thecollisionavoidancestrategyis

appliedbasedontherawtrajectory.Asexplainedpreviously,the rawtrajectoryisconstitutedofasetofconsecutivesegmentsthat linkthecellcenters.Foreachsegment,apalletmovesinstraight line in a travel time ti(i=1, 2, ..n), n denotes the number of segments. In ordertorespectthedesireddirection, thecurrent frequencyneedstobeadaptedforeachsegment.

A frequency fi denotes the reference for calculating the frequency in x and y direction (fxi,fyi). Thisfrequency defines the travel time ti to reach the destination and si denotes the distancethatpalletwilltravel.Asthepalletdisplacesonaplanar surfaceandthepalletspeedisrelatedtothefrequencyofinjected currents,toreachthedestinationatthesametimet_iinthebothx andydirection,thevelocityinthexandydirectionneedtobe coherent.Thevelocityinthexdirectioncouldbedeterminedin functionofthevelocityintheydirectionandviceversa.Therefore, thereare twooptionsfor calculatingthefrequency fiasshown below:

si=min(

D

xi,

D

yi)andfi=min(fxi,fyi) si=max(

D

xi,

D

yi)andfi=max(fxi,fyi)

Apalletmoves from[x₀,y₀]to[x_d,y_d](Table3). s_i=min(

D

x,

D

y)=30mm and fi=fx=4Hz. The travel time is 3.702s. A program calculatesthe frequency fy withthe smallesterror of position,lessthan0.06mminthiscase.Thesimulationresultsare showninTable3.

As shown in Fig. 24, the variation of the frequency is approximatelylinearbut notperfectlybecauseof theinfluence ofthetransitionzonesnumberthatthepallet crossesover.The resultsshowthepossibility tomanagethedesireddirectionby definingthecorrespondingfrequencyfiforthedistancesi. Fig.21.Variationofvelocityinthexandydirectionsinfunctionoftime.

Forthecollisionavoidancedemonstration,asimulationoftwo palletspathsisrealized.Thetwopalletsp1andp2aredisplacedon the 1010 square matrix emSS in straight line as shown in Fig.25.Thepalletp1goesfromaposition[x1₀,y1₀]=[40,40]mm to[x1d,y1d]=[230,230]mmwhilethepalletp2movesfrom[x20, y20]=[210,70]mmto[x2d,y2d]=[40,180]mm.Inordertocontrol

thetwopalletsalongthedesireddirection,thepoweredcurrentsto theemSSareshowninTable4accordingtothepreviouslydefined rule.

Fig.25shows the emSS simulation withoutcollision avoid- ance.Thesinusoidalcurrentsarechoseninordertomoveeach pallet in the desired direction from the starting point to the Fig.22.Calculationofpalletpathindifferentconfigurationlayouts.

Fig.23.RefinedtrajectorywiththecubicHermitmethod.

arrival point. In this first simulation, the ‘‘checking collision’’

module is not used.With these parameter inputs, when two pallets reach at the same time the point of intersection, a collisionishappening.

Inordertoavoidthecollision,a‘‘checkingcollision’’module isimplemented.Inthiscasestudy,threeareasaredefinedjust asshowninFig.11,includingthecollisionareaaroundthepoint of intersection and the secured areas of pallet 1 and pallet 2.Whentwoamongthreeareasintersect, acontrolstrategy is

defined.AsseeninFig.25,thepalletp2isnearertothepointof intersectionthanthepalletp1,itmeansthatthepallet2enters the collision area before the pallet p1. When the pallet p1reachesthecollisionarea,itgetsaninstructionthatrequires thepalletreducesitsspeedandthenstopsatasecureposition.

Hence,thepalletp1hastowaitforpallet2leavingthecollision areaas shownin Fig.26.During thesimulationtimeline, four images capturedat timet respectively4.78s,6.5s,8.58sand 9.63sareshownonFig.26andsimulatedthestoppositionof thepalletp1andtheon-goingofthepallet p2.

Fig. 27 illustrates the displacement of the two pallets during the simulation. There have no collisions of the two pallets at the intersection point. The figure shows the coordinatesof thepositionofthepalletp1andp2centersin functionofsimulationtimeand itshowsthesecuredistance between two pallets to avoid collision. When the pallet p2reachesthearrivalpoint,thepalletp1restartsitsdisplace- ment. It means that the two pallets are controlled simulta- neouslyandindependently.Thesimulationisstopped atthe timeof22.249s.

Thevelocitycurveofthepalletp1(Fig.28)showsthatthepallet p1reduced itsspeed beforestopping itsdisplacementduringa waiting time and restarted with slow speed. Furthermore, the effectofcrossingthetransitionzonesisillustratedbysmalland timelyspeedmagnitudedecreases.

Inthissection,the‘‘monitorandcontroloftheemSS’’activity interest hasbeen demonstratedand showstheabilitytoadapt trajectoriesincase ofdefectedcellsalsothepossibilitytoavoid collision.

Fig.24.Calculatingthefrequencyfy. Table3

Simulationresultsincaseofdisplacementtodifferentdestinationsinafixedtravel timet=3.702s.

[x0,y0] [xd,yd] Dx(mm) Dy(mm) fx(Hz) fy(Hz)

[40,40] [70,70] 30 30 4 4

[40,40] [70,80] 30 40 4 5.4

[40,40] [70,90] 30 50 4 6.72

[40,40] [70,100] 30 60 4 8.09

[40,40] [70,110] 30 70 4 9.45

[40,40] [70,120] 30 80 4 10.08

Fig.25.Controlling2palletsinopenloop.

Table4

Theparametersofthesinusoidalcurrentsformoving2palletsp1andp2.

Pallet Ix(A) Iy(A) fx(Hz) fy(Hz)

p1 0.4 0.74 6 6

p2 0.4 0.74 6 3.89

Fig.26.Palletpositionssnapshotsattimet=4.78s,t=6.5s,t=8.58s,t=9.63s.

Fig.27.Displacementtrajectoryofpalletp1andp2.

6. Conclusionandfutureworks

Thispaperpresentedthetechnicalprincipleanditsassociated framework for managing an electromagnetic modular Smart Surfaceasthetransferunit orconveyancesystemofa modular microfactory. This paper argued that the proposed emSS is a relevant candidate for the paradigm of reconfigurable micro- manufacturingsystems.Aframeworktomodelandsimulatethe emSScontrol is presented. First experimental resultsshow the abilityoftheproposedframeworktoperformplanartrajectories firstlywithdecoupledxandydisplacements thenwithcoupled displacements.Thechoiceofthedesiredtrajectorydependsonthe objective:fastdisplacementorminimalenergyconsumption.The abilitytogeneraterefinedpathtrajectoriesandtoavoidcollisionis respectivelydemonstratedbyusingpath-planningalgorithmsand interpolationmethods and by simulation using simple rule for prioritydefinition.

Futureworkswillfocuson improvingtheframeworktotake into account more complexity in managing pallets and on improving the strategy of trajectories determination. On the technologicalpoint ofview,alargeremSSwillbedesignedand manufacturedtobeabletotestmulti-palletstrategies.

Inthefuture,thecollisionavoidancestrategycouldbedecided byanautonomouspalletwithoutfollowingapredefinedprotocol andwithoutneedingsupervision.Inthiscase,thedifferentcells formingtheemSSaswellasthepalletscouldbemodeledasagents basedsystems.Alltheseagentscollaboratetorejectdisturbances (forinstances:reducevelocityofapallettoavoidcollisionwith another, adjust the pallet’s trajectory in case of miss target location).Thecombination ofglobal andlocal schedulerscould improvetheperformanceofthecontroller.Futureframeworkwill includethesimulationofproductionsystemstoenhancethedaily operationsandinvestigatethefuturealternativesofsolutionsfor decision-makingandevaluation.

Fig.28.Thevelocityofthe2palletsinfunctionoftime.

Acknowledgements

This work is a part of the MICROCOSM (Micro-Coordinate MeasuringMachine)project.Itwasfundedintheframeworkofthe ConseildelaRe´gionPicardie(2014–2017).

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound, inthe online version, at http://dx.doi.org/10.1016/j.compind.

2016.02.003.

References

[1]R.Reichwald,ChristofM.Stotko,T.P.Frank,Distributedmini-factorynetworks asaformofreal-timeenterprise:concept,flexibilitypotentialandcase studies,in:ThePracticalReal-TimeEnterprise,Springer-Verlag,Berlin Heidelberg,2005,pp.403–434.

[2]Y.Okazaki,Microfactories—anewmethodologyforsustainablemanufacturing, Int.J.Autom.Technol.4(2)(2010)82–87.

[3]R.Tuokko,R.Heikkila¨,E.Ja¨rvenpa¨a¨,A.Nurmi,T.Prusi,N.Sitala,A.Vuola, MicroandDesktopFactoryRoadmap, TampereUniversityofTechnology, DepartmentofProductionEngeering,Tampere,2012.

[4]Y.Koren,M.Shpitalni,Designofreconfigurablemanufacturingsystems, J.Manuf.Syst.29(October(4))(2010)130–141.

[5]N.Siltala,T.Prusi,A.Vuola,R.Heikkila,R.Tuokko,Modularmicrofactory systemforgassensorassembly,in: IEEEInternationalSymposiumon AssemblyandManufacturing(ISAM),2011.

[6]S.Scheifele,J.Friedrich,A.Lechler,A.Verl,Flexible,self-configuringcontrol systemforamodularproductionsystem,in:SecondInternationalConference onSystem-IntegratedIntelligence:ChallengesforProductandProduction Engineering,2014,398–405.

[7]N.Kawahara,T.Suto,T.Hirano,Y.Ishikawa,T.Kitahara,N.Ooyama,T.Ataka, Microfactories,Microsyst.Technol.3(2)(1997)37–41.

[8]T.Gaugel,H.Dobler,M.Bengel,C.Weis,J.Schliesser,Buildingamini-factory fromatechnologyconstructionkit,in:InProc.3rdInt’l.Workshopon Microfactories,September16–18,2002,Minneapolis,MN,(2002),pp.5–8.

[9]W.Wang,Y.Koren,Scalabilityplanningforreconfigurablemanufacturing systems,J.Manuf.Syst.31(2012)83–91.

[10]KorenY.andUlsoyA.G.,ReconfigurableManufacturingSystemhavingA ProductionCapacityMethodforDesigningSameandMethodforChangingits ProductionCapacity,UnitedStatesPatent,US6349237B1,2002.

[11]Y.-A.Chapuis,Y.Fukuta,Y.Mita,H.Fujita,Autonomousdecentralizedsystems basedondistributedcontrolledmemsactuatorformicroconveyance application,J.Inst.Ind.Sci.Univ.Tokyo56(1)(2004)109–115.

[12]R.Yahiaoui,R.Zeggari,J.Malapert,J.F.Manceau,AMEMS-basedpneumatic micro-conveyorforplanarmicromanipulation,Mechatronics22(2012)515–552.

[13]K.Boutoustous,G.J.Laurent,E.Dedu,L.Matignon,J.Bourgeois,N.LeFort-Piat, Distributedcontrolarchitectureforsmartsurface,in:ProceedingsoftheIEEE InternationalConferenceonIntelligentRobotsandSystems,October18–22, 2010,Taipei,Taiwan,2010.

[14]B.Piranda,G.Laurent,J.Bourgeois,C.Cle´vy,S.Mo¨besb,N.Fort-Piat,Anew conceptofplanarself-reconfigurablemodularrobotforconveyingmicroparts, Mechatronics23(October(7))(2013)906–915.

[15]Y.Okazaki,N.Mishima,K.Ashida,Microfactoryandmicromachinetools,in:The firstKorea–JapanConferenceonPositioningTechnology,Daejeon,Korea,2002.

[16]Y.Okazaki,N.Mishima,K.Ashida,Microfactory—concept,history,and developments,J.Manuf.Sci.Eng.126(4)(2004)837–844.

[17]K.Ashida,N.Mishima,H.Maekawa,T.Tanikawa,K.Kaneko,M.Tanaka, Developmentofdesktopmachiningmicrofactory,in:Proc.J-USASymposium onFlexibleAutomation,2000,175–178.

[18]R.L.Hollis,J.Gowdy,A.A.Rizzi,DesignandDevelopmentofaTabletop PrecisionAssemblySystem,MechatronicsandRobotics,September13–15, 2004,Aachen,Germany,(2004),pp.1619–1623.

[19]E.Descourvie`res,S.Debricon,D.Gendreau,P.Lutz,L.Philippe,F.Bouquet, Towardsautomaticcontrolformicrofactories,in:FifthInt.Conf.onIndustrial Automation,June2007,Montre´al,Que´bec,Canada,2007.

[20]S.Fatikow,J.Seyfried,ControlArchitectureofaflexiblemicrorobotbased microassembly,in:ProceedingsoftheseventhMediterraneanConferenceon ControlandAutomation(MED99),June28–30,1999,Haifa,Israel,1999.

[21]D.Gendreau,M.Gauthier,D.Heriban,P.Lutz,Modulararchitectureofthe microfactoriesforautomaticmicro-assembly,J.Rob.Comput.Integr.Manuf.26(4) (2010)354–360.

[22]E.Descourvie`res,D.Gendreau,P.Lutz,Datarepresentationforthecontrolof full-automatedmicrofactories,in:ProceedingsoftheFifthInternational WorkshoponMicroFactories,Besanc¸on,France,2006.

[23]M.Bosch-Mauchand,C.Prelle,J.Daaboul,T.A.T.Dang,S.B.Labato,Preliminary contributionsofindustrialmanagementmethodstomicrofactorycontext:

Caseofmicrofactoryconveyorsintegration,in:ProceedingsofJointConference onMechanical,DesignEngineeringandAdvancedManufacturing,June18–20, 2014,Toulouse,France,2014.

[24]E.Descourvie`res,D.Gendreau,P.Lutz,F.Kiefer,Specificationoftechnicalof dedicatedtoareorganizableandreconfigurablemicrofactory,in:Fourth InternationalWorkshoponMicroFactories,Shanghai,China,2004.

[25]H.A.ElMaraghy,Reconfigurableprocessplansforresponsivemanufacturing systems,in:ProceedingsoftheCIRPInternationalDesignEnterpriseTechnology (DET)Conference,18–20September,2006,Setubal,Portugal,2006.

[26]E.Jarvenpaa,R.Heikkila,R.Tuokko,Logisticandcontrolaspectsforflexible andreactivemicroanddesktopassemblyatthefactorylevel,in:Proceedings of2009IEEEInternationalSymposiumonAssemblyandManufacturing, November17–20,2009,Suwon,Korea,2009.

[27]D.Schmidt,J.Kegeler,3, MiniProd–MiniaturisiertesProduktionssystem,wt- online,95, Springer-VDI-Verlag,2005,pp.146–149.

[28]D.R.W.Barr,D.Walsh,P.Dudek,Asmartsurfacesimulationenvironment,in:

Systems,Man,andCybernetics(SMC),2013IEEEInternationalConferenceon, 13–16Oct.2013,(2013),pp.4456–4461.

[29]N.Arora,M.U.Khan,L.Petit,C.Prelle,Aplanarelectromagneticactuatorbased ontwolayercoilassemblyformicroapplications,in:IEEE/ASMEConference onAdvancedIntelligentMechatronics,July8–11,2014,Besanc¸on(France), 2014,6pages.

[30]N.Arora,T.A.T.Dang,L.Petit,M.Bosch-Mauchand,J.Daaboul,C.Prelle, Contributiontomicrofactorytechnologies:aflexibleconveyorandits dedicatedcontrolsystem,in:9thInternationalWorkshoponMicrofactories, October5–8,2014Honolulu,U.S.A,2014.

[31]Y.Fukuta,Y.A.Chapuis,Y.Mita,H.Fujita,Design,fabrication,andcontrolof MEMS-basedactuatorarraysforair-flowdistributedmicromanipulation, J.Microelectromech.Syst.15(4)(2006)912–926.

[32]H.Nakazawa,Y.Watanabe,O.Morita,M.Edo,E.Yonezawa,Thetwo- dimensionalmicroconveyer:principlesandfabricationprocessofthe actuator,in:ProceedingsofInternationalSolidStateSensorsandActuators Conference,Junel6–19,1997,Chicago,1997.

[33]H.P.Wiendahl,H.A.ElMaraghy,P.Nyhuis,M.F.Za¨h,Changeable manufacturing—classification,designandoperation,CIRPAnn.—Manuf.

Technol.56(2)(2007)783–809.

[34]M.U.Khan,N.Bencheikh,C.Prelle,F.Lamarque,T.Beutel,S.Bu¨ttgenbach,A longstrokeelectromagneticXYpositioningstageformicroapplications,IEEE/

ASMETrans.Mechatron.17(5)(2012)866–875.

[35]M.U.Khan,N.Bencheikh,C.Prelle,F.Lamarque,T.Beutel,S.Buttgenbach, Siliconconveyorbasedelectromagneticdeviceforlineardisplacement(July 2010),in:IEEE/ASMEInternationalConferenceAdvancedIntelligent mechatronicsAssociationInformationandManagement,Vol.no.2,Montreal, QC,Canada,2010.

[36]R.K.Ahuja,T.L.Magnanti,J.B.Orlin,NetworkFlowsTheory,Algorithms,and Applications, PrenticeHall,UpperSaddleRiver,NJ,1993p.07458.

[37]R.H.Heikkila,I.T.Karjalainen,J.J.Uusitalo,A.S.Vuola,R.O.Tuokko, PossibilitiesofaMicrofactoryintheAssemblyofSmallPartsand Products—FirstResultsoftheM4-project,in:InternationalSymposiumon AssemblyandManufacturing,Proceedingsofthe2007IEEE,July22–25, 2007,AnnArbor,Michigan,USA,2007.

[38]O.Klemd,Desktopfactory—newapproachesforleanmicroassembly,in:

InternationalSymposiumonAssemblyandManufacturing,ISAM,22–

25.07.2007,AnnArbor,Michigan,USA,(2007),pp.161–165.

[39]S.Nakano,K.Ashida,J.Akedo,OnDemandManufacturingSystem,aSolution ofEcologicalMEMSFactory,in:SixthInternationalWorkshoponMicro- factoriesIWMF2008,NorthwesternUniversity,October5–7,2008,Evanston, Illinois,USA,2008,pp.73–76.

[40]A.A.Rizzi,J.Gowdy,R.L.Hollis,Distributedcoordinationinmodularprecision assemblysystems,Int.J.Rob.Res.20(10)(2001)819–838.

TheAnhTuanDANGreceivedhisMaster’sdegreein

‘‘AdvancedSystemsandRobotics’’fromUPMC,Paris6, Francein2013andhisengineer’sdegreeinIndustrial EngineeringfromPolytechLyon,Villeurbanne,France in2012. He is currently workingtoward thePh.D.

degree in a microfactory project at the Roberval Laboratory,Universite´ deTechnologiedeCompie`gne, France.

MagaliBOSCH-MAUCHANDreceivedherPhDinMe- chanicalEngineeringfromtheUniversityofNantesand EcoleCentraledeNantesin2007onthemodellingfor thesimulationofvaluecreation chainsinindustrial enterpriseasadecision-makingaidtoolintheproduct design and production engineering phases. She is Assistant Professor in Department of Mechanical SystemsEngineeringoftheUniversite´ deTechnologie deCompie`gne—UTC(France).Sheisalsomemberof UMRUTC/CNRS7337 Roberval.Herresearchtopics focusonProductLifecycle Management,Production ProcessManagementandDigitalManufacturing.

Neha ARORA received the Bachelor of Engineering degree ininstrumentationand controls engineering fromtheMaharshiDayanandUniversityinHaryana, India.In 2010,shereceivedher Master’sdegreein mechanicandsystemswithspecialisationinmecha- tronicsystemsfromtheUniversite´ deTechnologiede Compie`gne—UTC (France). She is currently working towardthePh.D.degreeinmechanicalengineeringat theUniversite´ deTechnologiedeCompie`gne,France.

Christine PRELLE received the Ph.D. degree in industrialautomaticcontrolfromInstitutNational desSciencesApplique´esLyon,Villeurbanne,France, in1997.SheisaProfessorincontrolengineeringand mechatronicsatUniversite´ deTechnologiedeCom- pie`gne,Compie`gne,France.Sheiscurrentlyincharge oftheemergentresearchteam‘‘IntegratedSystemsin Mechanics’’atRobervallaboratoryofUniversite´ de Technologie de Compie`gne. Her research interest focusesonmicromechatronicsandcontrol.

JoannaDAABOULisAssistantProfessoratUniversite´ de TechnologiedeCompie`gneandresearcherinRoberval MechanicalLaboratory(UMRCNRS7337).Sheobtained her PhD in Mechanical Engineering from Centrale Nantesin2011.HermainresearchtopicsareEnterprise Modeling,SystemsPerformanceevaluation,Product/

Processintegration,andMassCustomization.