Abstract
Most cyber-physical human systems (CPHS) rely on users learning how to interact with the system. Rather, a collaborative CPHS should learn from the user and adapt to them in a way that improves holistic system performance. Accomplishing this requires collaboration between the human-robot/human-computer interaction and the cyber-physical system communities in order to feed back knowledge about users into the design of the CPHS. The requisite user studies, however, are difficult, time consuming, and must be carefully designed. Furthermore, as humans are complex in their interactions with autonomy it is difficult to know, a priori, how many users must participate to attain conclusive results. In this paper we elaborate on our work to infer intrinsic user qualities through human-robot interactions correlated with robot performance in order to adapt the autonomy and improve holistic CPHS performance. We first demonstrate through a study that this idea is feasible. Next, we demonstrate that significant differences between groups of users can impact conclusions particularly where different autonomies are involved. Finally, we also provide our rich, extensive corpus of user study data to the wider community to aid researchers in designing better CPHS.
Original language | English (US) |
---|---|
Pages (from-to) | 110-117 |
Number of pages | 8 |
Journal | IFAC-PapersOnLine |
Volume | 51 |
Issue number | 34 |
DOIs | |
State | Published - Jan 1 2019 |
Keywords
- Shared control
- autonomous mobile robots
- human robot interaction
- telerobotics
ASJC Scopus subject areas
- Control and Systems Engineering
Access to Document
Other files and links
Fingerprint
Dive into the research topics of 'Inference of User Qualities in Shared Control of CPHS: A Contrast in Users⁎'. Together they form a unique fingerprint.Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS
Inference of User Qualities in Shared Control of CPHS : A Contrast in Users⁎. / Hall, Lucas; Acharya, Urja; Bradley, Justin; Duncan, Brittany.
In: IFAC-PapersOnLine, Vol. 51, No. 34, 01.01.2019, p. 110-117.Research output: Contribution to journal › Article › peer-review
}
TY - JOUR
T1 - Inference of User Qualities in Shared Control of CPHS
T2 - A Contrast in Users⁎
AU - Hall, Lucas
AU - Acharya, Urja
AU - Bradley, Justin
AU - Duncan, Brittany
N1 - Funding Information: Inference of User Qualities in Shared Inference of User Qualities in Shared★ Control of CPHS: A Contrast in Users ★ Control of CPHS: A Contrast in Users★ Control of C∗PHS: A Co∗ntrast in Us∗ers Lucas Hall∗ Urja Acharya∗ Justin Bradley∗ Lucas Hall ∗∗ Urja Acharya ∗∗ Ju∗stin Bradley ∗∗ Lucas Hall UBrrjiattAancyhaDryuancaJnu∗stin Bradley LucasHall ∗ UrjBriattAancyhaDryuan∗caJun ∗∗stin Bradley ∗ Brittany Duncan∗ ∗All authors are affiliated with the Department of fiomputer Science & All authors are affiliated with the Department of fiomputer Science & EAnlglinaeuetrhionrgs, aUreniavfefrilsiiatytedofwNitehbrtahsekaD,eLpainrtcmolenn,tNoEf f6i8o5m8p8u,tUerSSAci(eenmceai&l: ∗Engineering, University of Nebraska, Lincoln, NE 68588, USA (email: Engineering, University of Nebraska, Lincoln, NE 68588, USA (email: {lhall, uacharya, jbradley, bduncan}@cse.unl.edu). Abstract: Most cyber-physical human systems flCPHS) rely on users learning how to interact Abstract: Most cyber-physical human systems flCPHS) rely on users learning how to intteeract Awibthstrtahcet:syMstoesmt c.yRbeart-hpehry,siacalcohlulambaonrastyivsetemCsPfHlCSPHshSo)urledlyleoanrnusefrrsomleartnhienguhseorwatnodintaedraapctt with the system. Rather, a collaborative CPHS should learn from the user and adapt Abstrtahcet:syMstoesmt c.yRbeart-hpehry,siacalcohlulambaonrastyivsetemCsPfHlCSPHshSo)urledlyleoanrnusefrrsomleartnhienguhseorwatnodintaedraapctt to them in a way that improves holistic system performance. Accomplishing this requires toolltahbeomratiinonabwetawyeetnhatheimhpurmovaens-rhoobloistt/ihcusmyastne-mcompperuftoerrmianntceera. cAticocnomanpdlisthhinegcytbheisr-prehqyusiicreasl system commuunniittiiees in order to feed back knowledge about users inttoo the design of the sCyPstHemS. Tcohme rmeuquniistiitees uisnerosrtduedrietso, hfoewedevebra,cakrekdniofwficleudltg,etimabeocuotnsuusmerisngin, taondthmeusdtesbiegncaroeffutlhlye CPHS. The requisite user studies, however, are difficult, time consuming, and muusst be carefully designed. Furthermore, as humans are complex in their intteeractions with autonomy it is difficult dtoeskignnoewd,. aFuprrtiohreir,mhorwe,masanhyumusaenrss amreusctompaprlteixciipnattehetior ianttteariancctioonncsluwsiitvhe areustuolntos.my it is difficult to know, a priori, how many users muusst participate to attain conclusive results. designed. Furthermore, as humans are complex in their interactions with autonomy it is difficult In this paper we elaborate on our work to infer inttrriinnssic user qualities through huumman-robot Inntetrhaicstpioanpsercowrreelealtaebdorwatitehornobooutr pweorrfkormtoainncferininotrridnesrictousaedraqputatlihteiesautthornooumghy haunmdainm-rporbovoet holistic CPHS performance. We first demonstrate through a study that this idea is feasible. hNoelxistt,iwc eCdPeHmSonpsetrrafotremthaantces.igWnifeicfairnsttddifefmeroenncstersabteetwthereonuggrhouapstoufduysetrhsactanthiims pidacetaciosnfcelaussiibolnes. Next, we demonstrate that significant differences betweeeenn groups of users can impact conclusions Npaerxtti,cuwleardlyemwohnesrteradtieffethreanttsaiguntiofincoamntiedsifaferereinncveoslvbeedt.wFeeinnagllryo,uwpes aolfsuosperrsovciadneiomuprarcitchc,oenxctluensisoinves particularly where different autonomies are involved. Finally, we also provide our rich, extensive pcoarrptiucsuloafrulysewrhsetrueddyifdfearteanttoatuhteonwoimdeirescaormeminuvnoiltvyedto. Faiindarlelys,eawrechaelsros ipnrodveisdiegnoiunrgrbiceht,teerxtCePnsHivSe. corpus of user study data to the wider community to aid researchheers in designing better CPHS. corpus of user study data to the wider community to aid researchers in designing better CPHS. © 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: Shared control, autonomous mobile robots, telerobotics, human robot interaction Keyworddss: Shared control, autonomous mobile robots, telerobotics, human robot interaction Keywords: Shared control, autonomous mobile robots, telerobotics, human robot interaction 1. INTRODUCTION Additionally, humans are complex and 1. INTRODUCTION Additionally, humans are complex and small numberhsuomf aunsserasremcaoymbpelexinaadned-Allowing autonom1.ousINTcoRnOtrDolUCTIOto beNsupervised by and smalAdditiol nunamlblyer, hsuofmaunsserasremcaoympbelexinadande-quamuaatelel ntouomdbrearws ocfoncucsluseurssioomnsa,ymbaekinginagditet-AshlalorwedingwiathutuonseormsocuasncgornetartollytiombpreovsueppeerrvfiosremdabnyceanbdyAllowingautonomouscontroltobesupervisedbyand dmuifafaitclelulnttuotmodbirdearewsntociffoynuhcsloeuwrssiommnasan,yymbuaeskeiirnnsagadrieet-shared with users can greatly improve performance by difficult to identify how many users are Allowing autonomous control to be supervised by and neifefidceudlttotoenidsuenretisfiygnhiofiwcamntarneysuultsse.rEs vaerne allowing users to complete complex tasks and assure needed to ensure significant results. Even shared with users can greatly improve performance by nfetehdeeddatotaenissucroellesicgtnedif,icdaenstigrneseursltso.f ECvPeSn thatalltasksarecompletedsafely.Inthesecases,most if the data is collected, designers of CPS allowing users to complete complex tasks and assure if athyendoattcaonissucoltllethcteedd,atdaestiogninercsoropfoCraPtSe designs rely on users “learning” or being trained on may not consult the data to incorporate that all tasks are completed safely. In these cases, most usaeyr sntoutdycoinnsfuolrtmtahteiodnaitnatotothinecirordpeosirgantse. how the autonomy behaves and adapting their behavior usaeyr snotudyt coinfonsultrmtheatiodan itantototheincirorpdeosigrantes. hacocswoigrtndhsinegraleyul.ytoWnoeonmcouynsteberenshda“vtlehesaartanntihndigsa”pdaaorpratdibnigegimntghiseitbrraabcinekhewdaavriodonsr. user study information into their designs. accordingly. We contend that this paradigm is backwards. We have been correlating user interac-how the autonomy behaves and adapting their behavior We have been correlating user interac-Collaborativeautonomiesshouldinferuserqualitiesand Wtioenshavnedbreoebnotcoaruretloantoinmgyuasenrdinpteerrfaocr--accordingly. We contend that this paradigm is backwards. tions and robot autonomy and perfor-use this knowledge to adjust autonomy in an effort to mtioannsceanwdithroibnotrtinasuictounsoemr qyuanlidtiepsewrfiotrh-umsoepllratohbvioesrtakhtnievoepwealreufdotgromenoatmonciaedsojsfuhtsohtuealdcuotimonnfbeoirnmeuydseiurnsqearu/naaliuetftifoeosnrotamntdoy mance with intrinsic user qualities with improvetheperformanceofthecombineduser/autonomy tions and robot autonomy and perfor-umseprtohvies tkhneopwelrefdogrme atoncaedojfutshteacuotmonboinmeyd iunsear/naueftfonrotmtoy the goal of building adaptive robot au-system.Inthiswaycyber-physicalhumansystemsflCPHS) mance with intrinsic user qualities with improve the performance of the combined user/autonomy tonomies as reported in Acharya et al. Accomplishingthisobjectiveisverydifficultasitrequireslearnandadapttogether. ftl2tou0n1do8ym).wieiWsthaes2i8rnepiptaioarrtltliyecdipcaionntdsAuiccnhtvaeerdsytaiagaeuttisnaegrl. Alearcconmanplisdhingadaptthistogetobhjeerct.ive is very difficult as it requires study with 28 participants investigating Aa ctcigohmtpfleisehdibnagctkhlios opbjienctwivheicihs vaerCyPdHifSficluealtrnass aitnrdeqinufiererss the impact of the intrinsic quality Locus Aactigcohmtplisfeehingdbacthisk looopbjeinctwivheicishvaeryCPHdiffiScleultarnsasaitndrequireinferss ofhtudyfeCiomnwptraitholcltaso2sf8dtpaescrhsecrticrinibtipaedridnsnbitycsqRotiunoavtletiserttigyrflLl1966)ao9tingc6u6s)thtuediymwpaitchto2f8tphaeritnictirpinasnitcsqiunavleisttyigLaotcinugs a tight feedback loop in which a CPHS learns and infers ofnCtohnetrpoelrafosrdmeascnrcibeeodfbayDRoouttbelreflt1e9l6e6p)-shortperiodoftimeusinginteractions.Whilethehuman- on the performance of a Double telep-intrinsic qualities about a user and adapts to them in a oensetnhcee proerbfootrmflsaenecneinofFaigDuroeu1b)l.e Ttehleespe-robotinteractionflHRI)andhuman-computerinteraction resence robot flseen in Figure 1). These frlHohboCroItt)picneotremioramdcuotinfoitntiimeflsHeRhuaIsv)ineagnaidnrtihecurhamckatninoo-nwcsol.meWdpguhetileberatsihneet,eshrtauucmdtiiaoenns-, supplemented flHCI) communities have a rich knowledge base, studies, woesenriitghintceawlor2obm8ootpraeflrseenftoicrip30ainnttoFst,iagusl,urwpepe1)rleem.iThnesnuteseeffdi-Fig. 1. Double fsalHynsCdteId)macsto,amthomensuenudistaeirteasqmhuaaavlyietniaeostriamcnhadkketnhotehwierleirdgrewleaaytbioatnosesd,heispstudieigwneitrhss, wciietnht ttwoodrmaworefirfmorc3o0nctloutsaiol,nws edrueeintosutfhfie-Fig. 1. Double systems, these data may not make their way to designers with two more for 30 total, were insuffi-Telep-oyf sctyebmesr,-pthyesseicdalatsaysmteamysnfolCtPmSa).keMtohreeiorewveary, tino tdheesigevneenrst limitations of the platform and our au- Fig. T1e.eslDeenpoc-ueble ofsystcyems,ber-pthhyesesicaldatsyastmemsay nflotCPmSa).keMortheieorewveary,tinotdhesiegneveernst tioiemnnoittmatotyi.odWnrsaewotfhfirmtehnercepoclrncautiluftoesrdiomansnandueadddoiutotirothenaaul-resence tifgchytbceorl-lapbhoyrsaictaiolnsyisstaecmhsiefvlCedPSb)e.twMeoerneotheveseer,cionmtmheuneivteienst,tonomy.Wethenrecruitedanadditionalreesleenpc-e tofigcyhtbcoler-plabhorysiatcalionsyisstaemschieflvCPedSb)e.twMoreeneotheveseer,cionmtmheuneivteies,nt 3tio0mnuoitmsaetyri.soWntsoeoptfhaterhtniecrieppclaratuteiftoierndmtahnaenadsdtduoidutiyronaaaulll-owing usRobtootdraw tioglhletcctoinllgabtohreatidoantais atochicelvoesde btehtewedeensitghneselocoopmmonunuitsieers-, RRobeosebnootcte collecting the data to close the design loop on user- 30 users to participate in the study allowing us to draw cioaglshleetdcctdoinellgsaibgtnohreiastifdoranatuaigshattochwicietlvhoesddeibfftiehcteuwletdieeens.itghFneosreleocxooapmmmopnluen,uiutsiseersr-, much stronger conclusions. The differences betwReeonbotthese baseddesignisfraughtwithdifficulties.Forexample,user much stronger conclusions. The differences between these coalsleedctdinegsigtnheis fdraatuaghttowcitlhosdeifftihceultdieess.igFnorleoxoapmopnle,uusseerr-msuecrhgsrtoruopnsgewracsonpcrlounsoiounnsc.eTdhaenddiffienrsetnrucecstibvee.twTehenistdhaetsae studies are time consuming, require institutional review user groups was pronounced and instructive. This data stasedouadridesdeflIaRsiregnB)tiimaspefpraurcovgnahsltu,mwireitcnhrgu,ditrifficumeqeuniltrteioes.finpsFatoirtrtuiexctiipoampannatlslre,,etvuiigserehwt wmusauescrhdgifstrfoircuonupltsgertwoacosoncbtpariluonnsiaoonnudns.creeThdquaeirnedddifferienxspenterrucestcisteibvfeer.towmTehetnihsotdsheaesetina boardflIRB)approval,recruitmentofparticipants,tight was difficult to obtain and required expertise from those in studies are time consuming, require institutional review wthaes HdiRffIicaunltdtCo PobStacoinmamnudnrietyquainreddreexppresretnistes farolmargtehoesfefoirnt monitoringofthestudyprocesstoensuredataintegrity,etc. the HRI and CPS community and represents a large effort board flIRB) approval, recruitment of participants, tight the HRI and CPS community and represents a large effort monitoring of the study process to ensure data integrity, etc. ★oThisnitowringork wofasthesuppstudyortedproin partcessbtoy NensSFureawadardta#1638099integrity, etc. the HRI and CPS community and represents a large effort ★ This work was supported in part by NSF award #1638099 This work was supported in part by NSF award #1638099 ★C2o40p5y-r8i9g6h3t ©©22001189 ,I FIFAACC (International Federation of Automatic Contr1o1l)0 Hosting by Elsevier Ltd. All rights reserved. CCPooepperyy rrriieggvhhiettw ©© u22n00d11e88r IIrFFeAAspCConsibility of International Federation of Automa111t1ic00 Control. Copyright ©2018 IFAC 110 C10.1016/j.ifacol.2019.01.047opyright ©2018 IFAC 110 toward our ultimate objective of a userffiadaptive autonomy for robots. Publisher Copyright: © 2019
PY - 2019/1/1
Y1 - 2019/1/1
N2 - Most cyber-physical human systems (CPHS) rely on users learning how to interact with the system. Rather, a collaborative CPHS should learn from the user and adapt to them in a way that improves holistic system performance. Accomplishing this requires collaboration between the human-robot/human-computer interaction and the cyber-physical system communities in order to feed back knowledge about users into the design of the CPHS. The requisite user studies, however, are difficult, time consuming, and must be carefully designed. Furthermore, as humans are complex in their interactions with autonomy it is difficult to know, a priori, how many users must participate to attain conclusive results. In this paper we elaborate on our work to infer intrinsic user qualities through human-robot interactions correlated with robot performance in order to adapt the autonomy and improve holistic CPHS performance. We first demonstrate through a study that this idea is feasible. Next, we demonstrate that significant differences between groups of users can impact conclusions particularly where different autonomies are involved. Finally, we also provide our rich, extensive corpus of user study data to the wider community to aid researchers in designing better CPHS.
AB - Most cyber-physical human systems (CPHS) rely on users learning how to interact with the system. Rather, a collaborative CPHS should learn from the user and adapt to them in a way that improves holistic system performance. Accomplishing this requires collaboration between the human-robot/human-computer interaction and the cyber-physical system communities in order to feed back knowledge about users into the design of the CPHS. The requisite user studies, however, are difficult, time consuming, and must be carefully designed. Furthermore, as humans are complex in their interactions with autonomy it is difficult to know, a priori, how many users must participate to attain conclusive results. In this paper we elaborate on our work to infer intrinsic user qualities through human-robot interactions correlated with robot performance in order to adapt the autonomy and improve holistic CPHS performance. We first demonstrate through a study that this idea is feasible. Next, we demonstrate that significant differences between groups of users can impact conclusions particularly where different autonomies are involved. Finally, we also provide our rich, extensive corpus of user study data to the wider community to aid researchers in designing better CPHS.
KW - Shared control
KW - autonomous mobile robots
KW - human robot interaction
KW - telerobotics
UR - http://www.scopus.com/inward/record.url?scp=85061115142&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85061115142&partnerID=8YFLogxK
U2 - 10.1016/j.ifacol.2019.01.047
DO - 10.1016/j.ifacol.2019.01.047
M3 - Article
AN - SCOPUS:85061115142
VL - 51
SP - 110
EP - 117
JO - IFAC-PapersOnLine
JF - IFAC-PapersOnLine
SN - 2405-8963
IS - 34
ER -