+ All Categories
Home > Documents > Web interfaces for mobile robots in public places

Web interfaces for mobile robots in public places

Date post: 22-Sep-2016
Category:
Upload: ab
View: 217 times
Download: 1 times
Share this document with a friend
9
Web Interfaces for Mobile Robots in Public Places Addressing Issues Such as Shared Control, Safety, and Ease of Use During Web Teleopwation of Museum Tour-Guide Robots 1070 9932/00/$10 D0@~0001EEE rolmts iri ptihlic plnccs tlirougli by DIRK KHULZ, WOLFRAM BURCARD, DIETER FOX, SEBASTIAN THRUN, and ARMIN E. CREMERS . . thc wcb. 'I'lic dcsign of thesc intcrlnccs spccifi- c:illy addrusscs issucs swli as low bL~nrlwirltli oi'lntcrriut conncccinnr, coiitrnl brokci-iiig, :uid sharcd colltrol as wcll :Is Web-Based vs. Conventional Teleoperation Wcb-bascd ~lecoi 1 trol tlificl-s honi convciiti 01 i;iI tcl crqicra- tiuii in sevcral nsl,ct:ts: MARCH 2030
Transcript
Page 1: Web interfaces for mobile robots in public places

Web Interfaces for Mobile Robots in Public Places

Addressing Issues Such as Shared Control, Safety, and Ease of Use During Web Teleopwation of Museum Tour-Guide Robots

1070 9932/00/$10 D0@~0001EEE

rolmts i r i ptihlic plnccs tlirougli

by DIRK K H U L Z , WOLFRAM BURCARD, DIETER FOX, SEBASTIAN THRUN, and ARMIN E. CREMERS

. . thc wcb. 'I'lic dcsign of thesc intcrlnccs spccifi- c:illy addrusscs issucs s w l i as low bL~nrlwirltli oi'lntcrriut conncccinnr, coiitrnl brokci-iiig, :uid sharcd colltrol as wcll :Is

Web-Based vs. Conventional Teleoperation Wcb-bascd ~ l e c o i 1 trol tlificl-s honi convciiti 01 i;iI tcl crqicra- tiuii in sevcral nsl,ct:ts:

MARCH 2030

Page 2: Web interfaces for mobile robots in public places

Figure 1. O n h a r d interlace of the interactive museum four-guide robot Rhino.

MARCH ZOO0

Page 3: Web interfaces for mobile robots in public places

hbfid+*W .”; :. . . . ~ ’ ’ . . - . ’ ’

. . . . . . . . . . . . . .

. . . . .

Figure 3. Minerva’s exclusive web control interface. Userr give target posirions by cficking inside the map. The rohot movcs to targets in a first-corne, first-setve manner.

. -.. ~ - ~ * r a m 9 p : - ~ . . ,. . , L:’’ :::.: , ~

. ; , i : : . . . : . , ~ , , : - - . . . . :. ; . - . . . . . . . .

::. ’, . . . .

,. .. , , . . . . - . . . . . . . . . . . ’ . I ,..., ::.:

. . . . . . . ,

. . . . .

yure 4. Control page of Rhino’s web interface. Users ran choose exhibits rhcy want the robot to go to as well as vicwpoMs.

Thc last two iritcrbces, whose coniuianda arc cxtretiiely Iii~11-leid with a n cxccutioii tinic lkatiic of sevcral iiiitiutcs, art: ptrticuliirly well suited fbr sharing coiitml hetween pliysi- cal and virtual iiitiseuiii visitors.

This articlc dcscribcs a l l t h e e iritetfices in rlcpth as wcll ;is tlicir atlva~~tagcs anti limitations. 13uring three wccks of on-hie n p e r x i r m (oiic rvcck in Gcrmatiy arid two weeks in tlic US), approxiiilately 4000 web usen cuutrollcd rlic robots. At ccrtairi tiriics, die roboLs were untler exclusivc control of tlie Wurld Widc Wcb, wliilc a t other tiiiies wrl) users mid coiivcntinnal Iiiuseiiiii visitnrs SILIFCCI contlrol. Unfortunately, the effcctivcricsr of rhe iliffere~it iuterfilccs i s dificult to cvaiu- ate systcmaticnlly, a s most fccdback we received i s anecdotnl iii nature. This article also diacmses iisability aspccts of the inter- hcis , and people’s experience of usirig tlieni.

Related Work Web-controlled robots linvc gained serious iiirerert in the last few ye;irs. First, thcrc is a vsricty of niatiiptilatoi-s that CIH lie used to tclcmanipula tc itunis located iri their workiiig area. Thc web intcrfnccs oftlicsc systcnis allow mcrs to issue single uoiritiiaiids that arc cxccuted iinniediately. Tlirec ewly system arc tlic M c r c ~ i y l’rojcct [3] itistnllerl ill 1394; hustrn1i;i’s ‘i’clcrobot 011 tlie web 1121, which cuiie on-line iic:trly nt tlic sniiie time; arid tlie Telegarden [4], which rcplaccd the Mcr- cury robot in 1‘395. Whilc thc Mercury I-nbot a i d t he T e l e g a ~ d e ~ ~ enxblc wcb mcrs to purforrii diffcreiit. types of dig- ging td i s , such as CxcnxtioIi of artifacts ss well as watering niid seeding fluwcrs, thc Tclcrobot oii the wcb gives web t i s e n

the oplmrttinity to build coiiiplcx structtires froiii toy Mocks. Thc intcrhccs cif thcsc robots iisc diffcretit tecliniques t o visu- alizc tlic actions of tlic robot and he i r eflwts. While iiiost of chc systciiis providc still iniagcs nf tlie robot’s olxrxiclii ni:e:i, chc Tclcgarderi system additiorially IWS CAD drawings to a n i - m a t e thc state of the triatiipulator. 111 atldition to thcsc mmili- ~ilatorr, there is it gmwiiig iiu~nber ofweb cniiicrx that dclivcr iiiiagc scrwiiis to tlte we l ) . Some nfthese cmicras, siich as [SI, arc installed on a robol m i 1 a i d can even be contro!lctl over tlic wcb by issuing sitiglu coimiarids.

I n orcler to let iii~lividunl I I S C ~ S tclcprogrritii a robot or to pcrforiri scqticnces ofactiotis, tiiore coiiiplex i n t d x x s ;ire rc- quirotl. A typical cxaiiiplc is thc 1’tliiial’:iint l’wject, ivliich al- lows people tr) draw n coinplctc painting by controlling ;i PUMA 760 mhot mil. The web interhcc pruvidcs a virtual c3tiv;Is iniplciiicntcd itiJsva mi wliicli the user dl.aws his p;biIit- ing. Tho iiitcrhcc forwards all actions oftlie user to the robot, s o that iilruost the saiiic iniagc appcaix 011 thc real cmvas. ‘l’hc systcni also provides visual feertliack in tlie f o r m ofpcriorlically updrtetl live im:igcs from thc i-obot. klirukawa CC al. 171 dc- scribe a well operatiorb it iterEice where web iiscrs mti pcrfororrn rnnnipulatioii tasks using a three-ditrieibsioHnl (3-11) graphics siiriiilntion contained in the web I)rowwt. ’The specified tasks arc first tcstcd on ;1 sininlator. A vcrificd scqiiciicc of actiuns cap be ti:niisi~iitted to the rr;d rohot.

MARCH 2000

Page 4: Web interfaces for mobile robots in public places

Additioiially, theru is a lorig list r,fniobik r o h t s that ciiii \)e cimtrollrd over thc web. Most uftlirsc systeins provide cxclu- sive remote coiitrol tu ;I siiiglc pcrwii or provide qucucs to sclicddc user reqocsts. 1CliepOnl'hcWeb [9] is P typical rcprc- srntativc ofmobilc rcihnts with a web ititcrhcu. Users c m givc elcmctitnry iiinveiiiCiit ; s t i o n s to thc rolmts and ubserwr them using several wel~-catiici:as. Xnvier I H] wits proliably tlie first: iiiobile robot to nperntc in :I pqxilntcd u f i c e building CDII-

trolled tlirnugh the weh. Xavicr call be advised by wcb iiscrs

tu niove to MI office and to tclla "kiiock-ktiock" joke sftcr ar- rival. The robot collects reqmsts Loth off-liric a n d ott-liric and proccsscv tlietii during special workuig liourr. A t c I the EUC-

cessfiil execution of the mission, Xnvier iiiforms the web Liser via c - i i d Xavicr's web iiirethcc rclics 011 clicn t-pull arid scrvcr-puslt tccliniqiies to providc iniages taken by the robot. Fitrthcrniore, it provides a map of thc r~iviroii~tici~t ;md i d - catcs die robot's ciirrciit pi)siitioii in rcgulnr iritctvnls.

I n dytiairiic Ciivirotiiiieiits, like 2 poptilatd iimseuni. unforeseen events iiiay occiir during task execution. Mnrr- over, since the task of a toiir-gtiidc robot i s to interact with peoplc in tlie ~ ~ i i i ~ e t i i i i , tlic nctioiis of tlic ruhnt cmiiot 31-

ways he prcdictcti, na is tlic casc i r i I I I O S ~ of thc systeni< dc- scribed abovc. Therefore, thc curitlo1 iritcrhces (if our tour-piidc robots, whicli arc dcscribetl in this cirticie, pro- vide vuious forins of ii1stnrlt fccdback tc the tiscrs. For ex- nrnple, tlicy curitiiiuously inforin users about ntlier L I S ~ I ' S

interncritig with tlie robot, tlie ciii'rerit goals to lie rcaclicd, ill>tl about thc actions currently hcing cnrricd ont. Siiicr tlicy w e basrcl on Java npplrtc, tlicy provide frequcnt updates ofthc rubut's state CVCII ovcr low-bandwidth cwiticciiotis. The in- tcrhccs a l ~ o tieliver iinngcs tnkcii with thc robot's cmicras ns well a i with a canwm ~nnrintcd 011 a ceiling. Fitinlly, they arc ;hie to serve scvcral penple at the saiiie tilw by different ~ypcs of brokering tcchniqucs.

Software Architecture of the Museum Tour-Guide Robots 'I'hc Ability to tiiove safcly tliroiigh a dyti;miic eiiviroiiiiie~it plays a f~~iidaiiietitnl role f c x web-bascti coritrol. Ilc~ice, we will begin by bricfly outlining oiir suftware architecture far niitunoiiious navigation, referring thc iiitercstcd rexier to iiiocc detailed literature. Tliif software :trcliitccturc plays ail

impwt;irit role, as it giiarmitees safcy a i i d iiiisrioti comp1i;irice undcr arbitrary latency conrlitiotis.

T h e ovrrall softw;we architccture of tliu muse~iin tour-giiidu robots Rhino arid Miiierva consists of deceritral- i A irinduies that ;ire execu~cd it1 pcirallel o n - h o d atid off-lmird, tising sockcts arid B tcthcrlcss Etlicrnct bridge for cc~iuiii~iiiication 1.1, 2, 'I 31. At tlic lowest levcl, various ijiod- ules i~iterfacc dircctly to the robot's sctisors arid cffcctors (la- sers, sonars, caiiicrar, i>iotors, pltiltilt titiit, Gcc, spcrcli t init , touch-senritivc d i sphy , Iriteriirt server, etc.). O n top of that , several navigation modules providu attti~Iioiiiom nsvig;uioii capabiiitics by carrying out fi~nctions l ike mapping, Iocaliva- tion, cnllisiori avoidnnce, a d p x h plnntiing. The intc~tctioii

leve l contruls the robot's voicc, hcad direcIion, ant1 other intcractivz nieilii~ such as h c Gcial cxprcssioti. F ~ ~ ~ a l l y , both robnh also employ high-level plauriers that coordinate critirc tours a t an abstract, symbolic level.

!

Figure 6. The monitoring paje provides smooth visrralizarion of the robot's current posirion in a previously learned two-dimensional map of the crrvironment. It also incirirles a ticker displaying rhe current ac- tions of ihr! robot.

MARCH 2000

Page 5: Web interfaces for mobile robots in public places

clesigii o f ;ipproprintc wcb intei-Encis, I~owcvcr, is cvcn m i r e ch;iIlc1tgitig, siiicc paoplc or1 d i e wel) i i r e r l additional infoi-iiin-.

t i m to iiiidcrstanrl w h a t is hapjiciiiiig ;it thc rc1iIocc sitc.

Web Interfaces of the Museum Tour-Guide Robots Interface #I: Exclusive Control by the Web T l ie fiirt wch i i itcrhcc dcscribcd 11c1.c i s clesigncd for midilc robots nildcr cxrlrrsiw control o f people cl11 tllc wcb. Thc p ~ r - p s c oftlic intcrhcc is to cnable people ;dl arrwnd tlic world to

gather iiiformation iiitcractiwly a t ;i rcmote site (;I 11111-

sciim). Thu main \vel) pige oftlic iiitcrficc i s shown in

the u w interfxes iiiiist he iiitiiitivc, so that iiiitrniricrl and iwntcchtiicnl iiscrs can opcriltc thc systcin with cnsc. I t iiimt also Lie iqqmlii~g, so that pcnplc will be ;ittr;ictcrl tu thc robot ;old will w,int t o p;vricip;ite in ;I tour . Figiircs 1 ; i n d 2 show thu oil-board Lisrr iiiterhcea o f thc r(i1mts I t h i n o ancl Miiicrva, whi cli eiiabl et1 i t iiiscii ni visitors t o ci iter col iii i i x i r ls simply by clickitig buttons o r by toucliiriji tlic robot's tr)id-suiisitivc iiioijilor, Iti nilciitiriii to displ:iyiiig tcxts :ind gqdi ics about thc uxhibits 011 thc S C ~ C U I I , thc robots plnycd prcrccordcd sounds; for cxainple, tc cxplaiii esliil~its o r to ask for clc;il;uic:c. ' I ' l i c

, &10 W. 0 NrhCepe. JaIn B t a r UlmO 1hS NYAll y l W d by MI1IERVh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v M/NERVAQ

Figure 7. Minema's rhared user and web interface. The robot's user interface is replicated an the we6 page (the applet with the yellow btittons).

Intetface #2: Perpetual Monitorhg with Exclusive Web Control Tlic second intcrhlcc, usorl with tho cubor LLliiiin, is dcsigncrl fix rncmitotiiig tlic actions of tlic robot over chc Ititcriicr. Jus t

52 MARCH 2000

Page 6: Web interfaces for mobile robots in public places

Figure 8. Pictures taken wilh lhe robot’s on-Boar-d cnmcrd.

I -

Figure 9. Typic<?/ on-hncl images recordctl with the ceilirty mounted camera lrackinq the robot

’ \ MARCH 2000 53

Page 7: Web interfaces for mobile robots in public places

iii;iticin wliilt. dit. r t h t is i i i motion, alnng w i h iiiorc specific iiiforiiiatioii w l i c i i the rntmt h a s rcachcd an exhibic. 'I'his i y

achicvcd b y iiitegratijig aJava applct, which receives dic U l i L of a correspriiidiiig web pge froin tlic roliot coiitrol syateiii via ;I U1>1' socket coriiicccioii. A third p ~ g c (SCC Pig. fj) i s tlc- sigiicrl to provide ;I siiiootli visu;ilin:iticiii of the rohot's pith i i i

the inap. This page npemtes siiiiilar tc dit. i i i ; i p displ:iy o f Mi- iicrva's cxclusive coiitrol iiitcrEu tlescrilietl ;ibovc. It cuiitaiiis a tickcr clisplaying thc ciirrcnt intcntions (plan) of thc rohot. For cxainplc, j, iniiglir riiiiio1iiicc thr iimic ofthc iicxt cxliibit. ]iist as above, this page is also iniplcnioiitcd iisiiig a Java applcc

IicIw fullows D siniplc approach. I r i coiitr;iat ti) physical visitors in tlic iiiLiscLiii1, whicl i sclcct tuurs on a first-conic, first-scrvc basis, wt.1) mcrs vntr. For this puqio~c, a11 npplcts are coiiucctctl to tlie ruhot coiicrol sywiri via a TCl' st,ckct conncctici~i. Tlic task-lcvcl

Sinw the delay and throughput of the Internet varies drastically, the cammunication be extremely low bandwidth.

t h a r crraldisties a UI31' socket co imcct ic~r to tlic robot's coil- trol systuni. For exaniplc, it tires tlic ctirreiit pnsitioii rccejvcrl finm thc control systcni to sinoothly visualize the IiiovcriieIits o l the robot. l ' h e status iiihmii:~io~i cliaplayctl in the tickcr is rcccivccl via dit. smic coniiccticiri.

Interface #3: Shared Control by Visitors and People on the Web I h t h iritcrfaccs dcscribcd abovc on ly suppore exclusive coIitrol of thc robro~. T h y provide 1 1 0 iiizaiis fiir sli;iriiig tlic robot's fi~ictionality bctwccn IIFCI'S in clic i i i i is t ' i t i1 i ; ~ r d vii-tual (reniotc) visitors. ' 1 ' 0 ciiablc physical md I-einotc visitors to upcratc tlw robot simriltancously, wc cicvelopcd a w c h ititerface tlint allows visitors to choosc complctc tours i r i skxd of sirigle target Ioca- tioris. T~i11i-s contaiIi bc~wcc~i tlitcc atid cix exhibits. This iiiw- f ; ~ , which i s implcii~ctitcrl iii Java, is executecl cl11 I)o:ilcl the robot aiid dirplnyeci oii tlie robot's touch-scrccii (SCC big. 2). Iiiteriiet usci-s ;nitoiii:itiually download thc intcrhcc as a n applet,

Figure I O . Minerva's pntli during a web-control smsion (exclusive contrail.

coutiul i i i o t i u l c coiiiits all incomiiig vote< of \vel) LWI-L;

ni id executes i r i rt.gul;ir t i n i c intcrvcils a tniir h i - the wcb iiwrs bawd oii the 1iI;ymity votc. T h c scorcs ol thc differ- clit tuurs nrc displayud mi all npplcts.

Res u Its Ihritig t 1 w x weeks cif oil-liiic inoliilc robot opcratioii (oiie wcck in Gt.riii;uiy : u i d two wccks in tlic 'US), npproxiiii;itcly 400IJ wc1) users conti-cilled tlic lohots. (.)tir cvaluario~i i s strictly :incctlotal ill iiatiiw, since we did iiot obtain systcmntic feedback froiii ;1 suficiciitly I;irgc fraction oE.ttie wcb users. Ncvcrtticlcss, we I)clit.vt. tlir wsh iiitcrhccs wcrc vcry attrac- tivc to p"lc 011 die web. A iiuiiibcr of-web I i s m lcft coni- iiiciits i t i our p i c ~ r book 01- otlierwiw suit u s fcccdbnck. A largc iiiiiiibcr oEpcopIc, oftrii iioviccs to tlic ficld, liked to control tlic idlots but also ciijoycd seeing d i e r pcopic intcrxting witli thc robot. Figures 8 n n d c) sliow typical imrgcs recorded with tl le oii - l io: ird a i d ccilin g-iiiou iitcd ~'a i~ i t 'ra i .

Iii ~)iic cxpcriiiiciit, thc cxclusive cmitrnl i i i t c r h x of Mi- I ICTVJ (intcrEicc f f l ) was on-litic f i x 3 l i iurs a i i d 40 iiiiriutcs.

I h r h g t l i i s tiinc, 6 0 tiscrs Ioadcrl h e cxclusive coiitrol p;ig:", Thcy c t i tc rd 278 t a r p p locations. Figurc 10 shows a typical tr;ict:tury of Miiicrva when i i VVBS m d e r exclusivc coiitrol. This partictilw 1" wx recorder1 tltu-irig ;III Iiitcrnct scssioti I;istiiig 73 niitiiirus 011 Scpteiiiber 4, 'I W R . T h c spccific riiii is 21 44 tinctcls Ioiig. Tlic niaxiitiuiii velocity of M i n c r v i was 163 crn/s. The ci>iitrcil systuiri rcccivud 103 target locaticriis. (:nrisequcii tly, thc avenge dismncc bctweeii cmsccutive t;lr- get I~icntioiis EKIS 20.8 ii1ctcr.S.

IUiiiio's wcb i i i t e r f k e (intcrhclcc #2) coulcl be accesseil for ;i total O F 32 Ii(iiirs ; i i id 18 iiiinuccs. Ihiriiig t h i s tirile, 2060 people visited I<hino's wcb p q p . l<Iiino travclerl ovcr 18 !mi titiring its 6-day duploynicnt pei-ind. Siiiiultnrier~usly witli thc wzh visitors, llhiiio gtiidcd OVIY 2000 pligsical visims tjirougli tlic I I ~ ~ I S C ~ I I I I . l h r c wci:u 2400 requesa to cxplairi cxliibits. Itliiiio iiianagcd to succcssfi~lly x n v e a t irs exhibits aiid dcliwr thc iiifonrmation to the visitors i n 239.1 c;iseq, which corresponds to iiii ovurall S ~ I C C L W rate Ot'1)1).75'%-iil tlic othcr

MARCH 2aaa

Page 8: Web interfaces for mobile robots in public places

six c:ises, Khiiio’s opcwtioii w a s disrupted by :I linttury chat ig t that let1 tn a rcstli:t o f lihitio’c quciie. At peak t r d k Iiotirc, ovcr I50 pcople siiiiiihttmusly tclcuper.llcil tlic rc>t)ot through tlie iiiiisciiiii, using thc wet) intcrfkce dcsciibcd. 11cl.e.

Mi iiut.v;i’s sharctl-control in tei-fJcc (I ntcrhcc #$ W:IS

ori-liric h r 91. liours ; t ~ i d wns accesscd by 2NH5 pcoplc. We cnuiitcd 45151. votcs for tlic six diiheiii tolit’s the roboc oitircci to thc visitors. Minewa trawled 3K.5 kni iuitler. sharctl w c h and on-line contrtil, which is 87.3% o f tlic ovei-all distance travclutl Iiy tlic rubut during the critiri cxtiibitiiw. I juring 620 tours, Minerva provicleti iidoriiiatioii :ibaut 2390 e x t d k . A total nuiiilw of 87 tours w c r c schcdulerl for virtual (wc-b-basal) visitors. The tiid\xiiiiliiii ii~iinhcr of wte’i for ;I tour at niiy pnitit in timc MUS 25‘). If one wants more than a single person

to use the interface, interfaces that support exclusive control may overly Lessons Learned

While the origiIia1 design nf I I IC three iiiicrticcc w i s souicwhat cvoliition:wy in iiattirc, they arc ncvcrtlie- less distiiict i i i tlw typc ofcnntrol tlicy provide, a n d the way thcy ;illow for rh:iririg control ainoiig rvcb iiscls and bc- [wccti wcl) uccrs :tnd peoplu who direr:tly iiitcriict with the ro- bot. Amoiig tlic most iniporcuit Ir .suoi is tliat WVLL lunrncrl clitiing thc cvili~aticin plinscs wcrc tlie hllowing:

1 , Since thu ilchy atid throughpiit o fhu Inceriiet varicr d ~ s - ticnlly, tlie cotiiiiiii nic;itioi I iliust be cxtrei iiel y low-tui dwidth. I Iigk initid <tart-lip costs, inferrcd by tlic javn-based siliiulatnr chat web iisers had to dowidoxl, arc prcfuddc ovcr liigl~-l);iti~l- width coiiuiiiinication diiriiig qwratioii (e.g., pcriodiciilly downloitiing (;IF imagou to display thc kitest robot po~iticiii). IUiino’s iiitcl-ficz tisod fixctl gnphics for clispl:iyiiig thc ml)ot’s Incatirm iii i c s i i iap, [tic size oFwhich i n n t l c tlic displny vciy slvw (typical update f i q u c u c y : 011cc i i i 1.0 sccoiids). I i i Mincrvn’a case, :I silnulacor was hwiiloadctl lip front, koni wl~iah poitit wi iliily chc robnt’s actital position I d to be c:oiiiiiitiiiic~tcti. Hcre we nchicvrd uptlatu ~cq~wiic:ics of 1 tc 5 Hci-tz.

2. Ifone w w t s t~ iurc thnii a single pci-ion to iisu tlic ititer- hcc, iiitcrtkef that supprwt exclurivc coiitl-ol may ovcrly re- strict tlic access to thc riibot. Ncvurthclcss, a i i intcrkicc dial gives wcb users inoii icnmy c:oiitrol ( e .g . , t)y schcduling ;I tar- get locxtirm i i i tlic que~ie) ;ire iiiorc cfticclivc than vot- ing-bascd iilt~‘rhccs, wlicl-e tlic robot’s ;ictions arc cliosci) 11y tlic tiInjority. The voting intcrEnce rlirlii’r give iiidividiial iisci-s sufficici~t cciiitml to bc truly excitcd; clicrcfoi-e, only coiisitler- alrly few wcb usen used this intei-hoc for r t h w control.

3. rntet’lcaving web-lrascd concrol with control by tlic pcoplc that dircctly intci-act wiili chc robot is a tricky issnc. Thc tcchtiiquc rlevclopcd for Minc ivn turned out tu he effective in pv id ing the Saiiir scivicc to visitoir in tlic niiisuiin and virtid viriiois. Forti- ri;\tcly, thc visitoir ttuiicil out to be vel-p coopcntive ;iiicl ~icvcr coiiiplairiud wl i c i i Mincrva chosc B tour Lx thc wcli visitorr;.

4 . l’lacing all itiipi)rt:int oii-line infori1i:itioii oil :I sir& web luge is csscntiil. In h t , Ithino’s intcrhce, which wi is Iiismri- c:illy tlie first, did not coiiiply with this r d c . Miiicrva’s iiitei,-

fkec wcrc nimc cfcctive hi this x p e c t , rwasurvtl l)y (ancctlr,tal) uwr ratirhctioii .

restrict the access to the robot-

MARCH 2000

Page 9: Web interfaces for mobile robots in public places

Dirk Scliulz is a l’Ii.l>. student a t thc I.)clmtIncnt ofC;oni- ptitcr Scicnce I11 ;it tlic Uiiivcmity of Ihiiib. H i 5 iiiain rcscarcli intcrcstr arc in the f ie lds uf. survicc inlmtics, Intclnct-l,ncctl telcprcwice, a n d probabiliqtic stiitc cstiination.

Dieter P o x lias a postdoctoral poriticlii :It thu r)cpartmcni of C:otnpiitcr scicncc a t Cariicgic hrlcllmi Univcrsity, L’ittrhui~gh, [’A. I-lis rcscaidi in tcrests w v c r robotics, probaldistic re;lsoli- ing, ai ic l wtificial iiitclligcncc. Fox carried his 1’11.11. degruc koni tllc Uiiiversity uf’llonii in 1998.

MARCH 2000


Recommended