An Interactive System for Exhibitions in a Science and Technology Center

An Interactive System for Exhibitions in a Science and Technology Center Alessio De Angel is, Paolo Carbone, Ma rco Dionig i, Emilio Di Giacom o, Aurelio Stopp ini, Fabio Radicioni Department of Engineerin g University of Perugia , Perugia, Italy , E-mail: alessio.deangelis @unipg.it Enrico Tom besi Perugia Officina Scienza T ecnologia (POST) Science and T echnology Center Perugia, Italy Abstract — This paper presents the dev elopment of a system for realizing interactive exhibitions in the context of a science and technology center. The core fun ctionality of the system is provided by a positioning subsystem comprised of a fixed infrastructure of transmitters and a sensor w orn by a user. The operating pri nciple of t he positioning system is based on inductive coupling of resonators. Information about the position of the user is transferred to an information system f or processing and displaying. Possible use cases include interactive games, information retrieval interfaces and educational scenar ios. Keywords — Science and technology centers, Interactive exhibitions, Indoor positioning. I. I NTRODUCTI ON Science and technology centers play a fundamental role for promoting the educati on of t he general public on scientif ic topics, which are pervasive in today’s society . To stim ulate the interest of a diverse audience, a s cience and technology center typically provides interactiv e exhibitions . In this context, several different electr onic an d inf ormation systems have been de scr ibed in the liter ature, ranging from the immersive augm ented real ity system presented in [1], to pervasiv e social games proposed for cultu ral organizati ons in [2] . Additi onally , traditi onal mus eum exhibition s are benefittin g from the use of technology , such as decision support system s, to plan exhibitions for an o ptimized user experienc e. Th is aspect is enabled by positi oning system s deployed inside the exhi bition center to track visitors and analyze thei r behavio r, such as those pr esented in [3] and [4]. In this p aper, the development of an inte ractive sy stem for engaging and educating visit ors of a scien ce and techn ology center is described . This system employs posit ion an d movem ent in two dimensions as a means for user inter action with an exhibiti on . Therefore, the core of the reali zed system is a positi oning su bsystem based on an infrast ructure o f know n- position nodes d eploy ed in the exhibition area and a sensor, which is worn by the user. The principle of operation of the positionin g sys tem is based on inductive c oupling of resonati ng coils [5] , [6] . The posit ion m easurem ent results are used by a softw are applicati on to implemen t the interacti on fun ctionality and game mechan ics of the interactive exhibition. Such softw are applic ation also prov ides the inform ation fee dback to the user by m eans of a dis play installed in the ex hibition . Potential ly, the educational va lue of the interac tive system lies both in the applic ation and in the princi ple of operati on of the positi oning sy stem its elf. On th e one han d , by pro perly designing the softw are applicati on as a gam e, it is possibl e to demonstrate scientif ic conc epts in an enter taining fashion. On the other hand, by providing additional inform ation material, it is possible to educate the visitor on the underly ing scientific concepts of electromagnetis m and geom etric localization methods, while the visit or is experiencing the effects o f those concepts. II. S YSTEM I NTERA CTIONS AND D ES IG N R EQUIREMENTS The science and technology center is a sys tem of system s and therefo re the interact ions betw een the developed interactiv e exhibition system and the larger system represented by the scienc e and techno logy center must b e analy zed. Such interaction s include social, behavioral, and reliability aspects , as illustr ated in Fig. 1. The social int eraction asp ect is related to the kind of audience the sy stem is targete d to, within the people w ho visit the science and technol ogy center , namely , children, teen- agers, and adults. To stimu late the interests of these differen t types of audience, the system shoul d supp ort s everal differen t activities . In the developed system , this is accom plished by a softw are a pplication, called Positi on C ontroll ed Applic ation , Scie nce and T echnology Cent er Inte r activ e Exhibition S yst em Social int er action Beha vioral int er action Fig. 1 – Inter ac tions of the d evel ope d system with the larger system of which it is a par t, i.e. the S cien ce an d Te chno logy Cen ter. Preprint versio n. Presented at: IEEE Inter national Sy mposium on Syste ms Engineering (ISSE) , Vienna , Austria, October 11-13, 20 17. © 2017 IEEE. Pe rsonal use of th is material is permit ted. Permission fro m IEEE must be obtai ned for all other uses, in any current or future me dia, including reprinting/re publishing this ma terial for adver tising or promotio nal purposes, cre ating new collective wor ks, for resale or r edistribution to servers o r lists, or reuse of any copyrighted compo nent of this wo rk in other wor ks that offe rs vari ous activities /g ames and a s election interf ace where th e user can select th e m ost relevant game . The behavior al interacti on aspect relates to the type of behavior tha t the system requir es from its users, e.g. th e kind of motion and gestu re s that the user performs, and the erg onomics implicat ions on the design. In the developed sys tem, the ergonom ic aspect was considered during the design phase and, as a result, a wearable device was realized, w hich the use r c an wear as a badge during the walking motion or hold in the han d. Finally , it is im portant to consider r eliabil ity aspects such as the possible stress and d amage caused to the realized system due to m isuse, im proper han dling of hardw are compon ents, or incidents by the exhibiti on visi tors. The application, togeth er with the system interactions described above, dictates severa l constraints that must be considere d during the design p hase. In p articu lar, the developed system should be intui tive and easy to use, physically robust, self-cont ained, and responsive . T hese constrain ts m ay be translat ed into technical requirements , to gu ide the design proc ess. Specific ally, the designed sys tem should not require user configuration of hardw are and software components , it should be easily wearable or hel d b y the user without cabling or power cords, and finally it should provide position information with a d ecimeter-o rder accur acy and an update rate of at least 1 Hz. T he develope d system , which is described and charact erized in the followin g sections, complies with these requirem ents. III. S YSTEM A RCHI TECTURE A. High-level a rchitecture The architectu re of the realized system is shown in Fig. 2. The two coordinates ( x , y ) , representin g the position of the user in a c oordinate sy stem relat ive to the known node infrast ructure, are continuously measu red by the position measurem ent system and prov ided as an input to the Position Controlle d Application . This applicati on processes the positi on according to th e game mechanics and provides feedback to the user by display ing the inf ormation g raphical ly on a m onitor . B. Position measuremen t subsystem architecture The architec ture of th e position measuremen t subsystem is shown in Fig. 3. An infrastru cture consisting o f four know n- position transmitting coils ( an chors ) is deploy ed. Each anchor transmits a sinusoidal signal and is chara cterize d by a unique operatin g fre quency; thus , th e sy stem operates in fre quency division m ultiplex ing mode [5] . The user w ears a receiving coil moun ted on a badge ( mobile node ), wh ich is equip ped w ith a microcont roller that d igitizes the received signal and transfers the samples to the PC via a Bluetooth connecti on. The mobile node is self-c ontained, batt ery operated, and enclosed in a 3D- printed case measuring approxim ately 15 × 5 × 5 cm, which may be atta ched to th e use r as a ba dge. Acquired samples are processed by the P C to estim ate the amplitude of the signals receive d by the four transm ittin g anchors. The am plitude estimation algorithm is based o n th e linear least squares s inefit proc edure in [7], extended to t he case of a signal consisting of the sum o f four sinusoids. The estimated amplitude valu es are then use d to compute t he distance be tween each anchor and the m obile node b y inver ting the model that relates induced v oltage V at the receiver t o transmitte r-receiv er dist ance d . Under the assum ption that the receiver and transmitte r coil s lie o n the sam e p lane, it may be sh ow n that such m odel is represent ed by a power law , i.e.     d V , (1) where  and  are constants , diff erent for each ancho r, and dependent on c omponent tol erances and environm ental character istics [6] . The ideal free-s pace value of  is 3. However, if the envi ronm ent surr ounding the coi ls c ontain s metallic structures , the magnetic field might b e distorted and such value mig ht differ from the ideal value. In practical applicati ons, the v alues of  and  should be found by calibration in th e prelim inary deploym ent phase of the sy stem. The power-law model in (1) allows for a low-com plexity processing method to b e implem ented, and for realizin g a low- power 2 -D positioning system . I f the assumption that the receiver and transm itter coils lie o n the same plane is not applicabl e, i.e. f or ar bitrary orientati ons, a differ ent model m ust be used, which is more compu tationally b urdens ome and prone to nu merical issues [5] . I n th e developed system , we apply th e simple power-law mod el in (1), due to its computat ional advantages . T heref ore, we design the hardw are conf igurati on of the system to ensure that all coils lie approxim ately on the same plane. In particular , the anchor coils are mounted horizontal ly on stands at the sam e height and the receiving coil is mounted horiz ontally on the mobile node carrie d by the user. Small deviati ons from th is c onfigurat ion are unav oidable in practice, main ly due to us er height differ ence, p osture, and walk ing motion. However, their influenc e on positionin g accuracy is within acceptable bounds. As shown in [5] , in fact, negligibl e positioning errors occur if the transmitter and P osit ion measu remen t s yst em P osit ion Contr oll ed Ap pl icati on 2D coordina tes (x, y) Fig. 2 – A rch itecture of the re alize d int erac tive e xhi bitio n sys tem. Anchor 1 Ancho r 2 Ancho r 3 Anchor 4 Mobile node worn by user Displa y PC running software application Bluetooth l ink: Mobile node - PC Fig. 3 – A rch itecture of the pos itio ning s ubsy stem . receiver coils do not lie on the same plane, provided that the angle between the line passing through the centers of a transmitte r and the receiver and th e plane of the transm itters is smaller th an 10°. Once distance est imates are av ailable, the two-dim ensional vector containing the x and y coordinates of the user’s position , denoted by   T y x  x , is estimated by trilate ration, i.e. by solving a l ea st s quares o ptim ization pro blem as follow s [5]:        N i i i d 1 2 ~ argm i n ˆ a x x x ( 2) where i d ~ is the m easured d istan ce b etween the m obile no de and the i -th anchor, N is the number of anchors, with 4  N in the realized system , i a is the vector of the known coordinates of the i -th anch or, and  denotes the Euc lidean no rm. C. Architectu re of the Position Controlled Applica tion As explained in Sectio n II , the position of the user continuously measured b y t he position meas urement system is provided as an input to a software application, called Position Controlled Application (PCA). PCA offers v arious activities/games (we will r efer to them a s apps ) that ca n be executed/played b y the user s. The architecture of PC A is shown in Fig. 4. T he Position Receiver m odule is a server listening on a socket; t he positioning subs ystem acts as a client for this ser ver co ntinuously sendin g t he measured coordinates. The received values are forwarded to the Execution Environment module. This communication is r ealized according to the observer de sign pattern: eac h time a p air of coordinates is received , the status o f a suitable obj ect is changed and the Execu tion Environ ment module (the observer) is notified. As the na me suggests, the Execution Environment module r eprese nts the “environment” t hat a llows the various apps to be executed. I t has three main functionalities: 1. It maintains a list of available app s. T his list contains not only the activities/game s to be played b y the user, but also some utilit y apps. In each mome nt, one of the apps in the list is i n executio n. When the app lication is lau nched the app in execution is t he Home App , a utility ap p that presents to the user t he list of available apps, and allows her to select the one she i s interested in. 2. It maps the p hysical coo rdinates received by the p osition measurement system into pixel coor dinates in the application canvas. T he received coordinates ( x,y ) indicate a position in the physical space where the user can move; in ord er for these co ordinates to be used b y the apps they need to be translated into a position ( x’,y’ ) in the virtual space of the ap plication ca nvas. T he transformation is given b y: ) ( ) ( ' ) ( ) ( ' M IN M IN MAX M IN M IN MAX y y y y H y x x x x W x       where W and H are the width and the height of the can vas, respectively, while MA X x and M IN x ( MAX y and M IN y , respectively) indicate the maxi mum and minimum x - coordinate ( y -coordinate, respectively) in t he p hysical space. In order to know the values of MA X x , M IN x , MAX y and M IN y , a Calibration App is available. This app p erforms a simple c alibration proce dure asking the user to m ove to the four sides of the spac e where she can move and registering the val ues r eceived by t he position measureme nt system. 3. It generates event s for the app that is executi ng. Once the coordinates are transformed, the Exe cution Environment module generates an event of type UserMoved , thus indicating to the executing ap p that the us er has moved to position ( x’,y’ ). Each app handles the event acco rding to its logic. F urthermore, if the position received re mains the same for a certain number o f times (five in the current implementation) a UserClicked even t is ge nerated. T his event is m eant to be used b y t he apps as the equivalent of a mouse click. The application has been i mplemented by using the Java FX 8 technology. App List Position Receiv er Execution Envir onment App n App k App 1 … App 2 Executing App Events Notification Selection Position Fig. 4 – T he P CA ar chi tectu re. IV. P OSITI ONING S UBSYSTEM I MPLEMENTATI ON A. Implementation Issues The trans mittin g anch or nod es are implemented as air-co re copper-w ire coils having a radius o f 7 cm and 20 w indings, resultin g in a nominal inductance value o f approxim ately 125 μH. A 160 nF capacitor is conne cted in to each coil, thus realizing a parallel LC resonator. T he resulting nomin al resonanc e fre quency is approxim ately 35 .6 kHz, and the quality factor of the resonat or is on the order of 10. The receiving mobile node is realized using a sm aller coil, since it must be wearable and easily ca rried by the user. The radius of the receiving coil is approxim ately 2.5 cm , and the coil has 60 turns, resulting in a nominal induct ance of approximate ly 200 μH . A 100 nF capacitor is connect ed to the receiving co il, to obtain the s ame nom inal resonanc e frequency as that o f th e transmitte rs . The total pow er consumption of each anchor is approxim ately 100 mW, and each ancho r is suppl ied by a wall power su pply. Each anchor i s programm ed with a unique operating frequency , which is close to the r esonance of the circuit. A quartz oscillato r is used at each anchor to stabiliz e the operatin g frequ ency . The set of f requencies assig ned to t he four anchors is the follow ing : [34482.7 35398.2 36144.5 36922.8] Hz . The coil is driven by a device based on a programm able system on chip (PSoC) microcont roller, of the PSOC 5 LP family , which generates a 5 -Vpp square w ave at the specified frequen cy. Moreover , a transis tor driver circ uit provides the requi red current to the coil , an d the resonance behavior results in a sinuso idal tim e-vary ing magn etic field. A diagram depicting the arch itectu re o f the realize d mobile node, wh ich acts as receiver, is shown in Fig. 5 . T he volta ge signal induced b y the time-var ying magnetic field generate d by the anchors, consisting of the su m of four sinusoidal signals, is first amplifie d using an instru mentation amplif ier with a gain of 100. Then, it is digitiz ed by the analog- to -digital conver ter included in a m icrocontrolle r of the PSoC 4 B luet ooth Low Energy (BLE) family, sampling at 200 kSa/s, 12 bits . A record of 3 00 samples is acquire d and transferre d to t he PC using the Bluetooth lin k for f urther p rocessing . T his procedure is repeated as soon as the BLE data trans mission is completed . The Bluetooth scan, handshaking , an d connection operati ons are performed in the initial deploym ent phase of the system, thus w ithout requi ring use r int ervention . B. Preliminary test setup A p icture of the prelim inary setup use d for developing and character izing the system in controlled laborat ory conditi ons is shown in Fig. 6 . T he ancho rs were p laced at a height of approxim ately 1 .2 m on the corne rs of a 2 .66 m × 4. 66 m rectangle . T he user carried the receivin g coil as a badge, the receiver’ s electronics and battery were contain ed in a box carried by the user, for ease of access and modific ation. This setup allow ed for rapid functional verificati on of the system operation . Before operating the system , a calibrati on procedure was performed, to identify the  and  constants in (1) for each anch or. Th e cali bration w as perform ed by placin g th e receiver at five calib ration points having known distances from each anchor and calculatin g th e valu es o f  and  in (1) tha t best fit the measured values in a l east squares sense. I n the system ’ s operating phase, a position update rate o f 2 Hz w as observed, and the sy stem coul d consistently estim ate the trajectory of a user during walk ing motion in real time , inside the area delimite d by th e four anchors, as c an be se en in Fig . 6. C. Position ing accuracy eva luation To evaluate the positionin g accuracy, a geodetic survey has been perf ormed, obtain ing th e referenc e position s. By m eans of an electro- optical ge odetic theodo lite (Lei ca Geosystem s total station TS- 06 ) the three-dim ensional coordinates of the follow ing points have been determ ined: - centers of the f our transm itting coils A, B, C, D realizing the datum; - center of the receiving test solenoid occupy ing the calibrati on points C 1, C2, C3 , C4, C 5; - center o f the receiving t est solenoid o ccupy ing the c ontrol points P0 1…P27. Fig. 5 - A rc hitec ture o f the mob ile no de ( rece iver) , s how ing i ts w irele ss conne ctio n w ith the PC . Fig. 6 - P ict ure of the prel imin ary ex per iment al se tup. To perform the measurements, the datum solenoids and the test solenoid have been sign alized by m eans of squa re retro- reflectiv e targets (58x58 mm size) position ed over the vertical axis of each solenoid at a vertical offs et m easured at a ±1 mm accuracy . The accuracy obtained by the TS -06 on the des cribed targets in the ex perimental condit ions can be prudently assum ed as ±2 mm. For control purposes, all geo detic measures have been repeated five times, and the d atum poin ts have been measured twice, at the start and at the end of t he survey , for a repeatability check which showed d iffe rences less than 2 millim eters. T he results of the geodetic survey are resumed in Table I , w here A* , B*, C* , and D* denote the coordinates obtained by measuring the datum points the second time, at the end of the survey. After the geodetic surv ey, 10 r epeate d position measurem ents were perf ormed using the realized po siti oning system at each control point . T he positionin g error was defined as the Eucli dean distanc e between the 2D position estimated by the propos ed system and the 2D ref erence posi tion, o btained by projectin g the posit ion provid ed by the geo detic survey on the plane w here the tr ansm itting coil A lies. Experim ental results are show n in Fig. 7 , where it can be noticed that the positi oning error in the central part of the area delim ited by the anchors is sm aller than in the border regions . This is m ainly due t o unfavora ble geom etric conf igurati ons that resul t in a poor ge ometric al diluti on of pr ecision [9], an d t o saturati on effects when the receiver is too close to a transmitte r. By c onsidering all survey ed positi ons, the mean positionin g error w as 21.5 cm. By eliminatin g the six co ntr ol points o n the b orders, thus considering only the cont rol points that are strictly inside the regi on delimited by the transmitti ng nodes, th e mean positionin g error was 12. 4 cm. Therefo re, th e decimeter- order accuracy re quirem ent is s atisfied by t he realize d system . Furtherm ore, the empirical cumulativ e distributi on functi on (CDF ) of the pos itioning error for the control points inside the region delim ited by the trans mitting nodes is show n in Fig. 8. It is possible to notice that, in 90% o f Fig. 8 – E mpir ical CDF o f t he po sitio ning error . Fig. 7 – Experime ntal positio ning results . A verage estima ted positio ns are show n as aster isks for each c on trol point. R el ative positio ns are in the coord inate fr ame cente red a t A in ( 0,0). T ABLE I – C OORDINATES IN THE LOCAL SYSTEM FROM THE GEODETIC SURVEY Points x (m) y (m) z ( m) A 0.000 0.000 1.250 B 2.678 0.000 1.263 C 2.711 4.694 1.242 D 0.009 4.692 1.233 C1 0.700 1.018 1.237 C2 2.037 1.023 1.240 C3 2.044 3.699 1.235 C4 0.701 3.703 1.230 C5 1.367 2.360 1.235 P01 0.699 0.023 1.240 P02 1.371 0.006 1.240 P03 2.030 0.016 1.242 P04 0.703 0.685 1.239 P05 1.357 0.703 1.237 P06 2.030 0.695 1.240 P07 0.701 1.029 1.239 P08 1.371 1.021 1.240 P09 2.035 1.019 1.240 P10 1.039 1.694 1.237 P11 1.374 1.690 1.237 P12 2.044 1.686 1.239 P14 1.368 2.365 1.236 P15 2.041 2.360 1.238 P16 0.697 2.703 1.233 P17 1.371 2.699 1.236 P18 2.041 2.701 1.238 P19 0.371 3.370 1.230 P20 1.370 3.368 1.234 P21 2.044 3.365 1.235 P22 0.703 4.037 1.229 P23 1.378 4.039 1.233 P24 2.052 4.046 1.232 P25 0.697 4.698 1.228 P26 1.384 4.710 1.231 P27 2.045 4.696 1.233 A* 0.000 0.002 1.250 B* 2.677 0.001 1.263 C* 2.710 4.694 1.243 D* 0.009 4.690 1.234 the cases, the error is smaller than 25 cm. T he accu racy and coverage area of the system could be further improved by using additional anchors . However, the obtained levels of accuracy and coverage are satisfacto ry for the applicati on under consi derati on. V. C ONCL USION In thi s paper, an indoor positioning s ystem was p resented and characterized. The aim of this system is to enable the development of interactive exhibitions in a scie nce and technology center. T he architecture o f the devel oped system, comprised of a position measurement subsystem and a position controlled applicatio n, was presented. E xperimental results were provided for position accuracy characterizati on , showing a decimeter -order error . A CKNOWL EDGMENTS This research activity was funded through t he INCOMPASS project, Fondo Ricerca di B ase 2016, by the University of Per ugia a nd through grant PRIN 2015C37 B25 by the Italia n Mi nistry of Instruction, U niversity and Research (MIUR), whose supp ort the authors gratefull y acknowledge. R EFERENCES [1] J. Torpus and B. Toble r, "lifeClipper3 — An augmented walking experie nce field evaluation of an experience design approach for immersive out door augmented reality ," 2011 IEEE International Symposium on Mixed and Augmented Reality - Arts, Media, and Humanities , Basel , 2011, pp. 73-82. [2] D. Economou, V. Bouki, T. Kounenis, M. Mentzelopoulo s and N. Georgalas, "Treasure hunt pervasive games i n cultural organisations," 2015 Internat ional Conferen ce on Int eractive Mobile Communication Technologies an d Learning (IMCL) , Thessalo niki, 2015, pp. 368-372. [3] C. Martell a, A. Miraglia, M. Cattani and M. van Steen, "L everag ing proximity sensing to mine the b ehavior of museum visitors," 2016 IEEE International Conference on Pervasive Computing and Commun ications (PerCom) , Sy dney, NSW , 2016, pp. 1- 9. [4] A. F inkel, A. Harwood, H. Gaunt and J. A ntig, "Op timizing indoor location re cognition t hrough w irele ss fingerprinti ng at the I an Potter Museum of Art," 2014 I nternational Conference on Indo or Positioni ng and Indoor Navi gation (IPIN) , Busan, 2014, pp. 210-219. [5] V. Pasku, A. De Ange lis, M. Dionigi, G. De Ange lis, A. Moschitta and P. Ca rbone, "A Positioning System Based on Low-Frequency Magnetic Fields," in IEEE Trans actions on Ind ustrial Electronics , vol. 63, n o. 4, pp. 2457-2468, A pril 2016. [6] V. Pasku, A. De Ange lis, M. Dionigi, A. Moschitta , G . De A ngelis an d P. Carbone, "Analysis of Nonideal Effects and Performance in Magnetic Positioning Syste ms," in IEEE Transactions on Instrumentat ion and Measurement , vo l. 65, no. 12, pp. 2 816-2827, Dec. 201 6. [7] IEEE Std 1241-2010 (Revision of IEEE Std 1241- 2000), “I EEE Standard for Terminolo gy and Test Methods for Analog- to -Digital Converters," Jan. 14 2011. [8] V. Pasku e t al., "A Magnetic Ranging-Aided Dead-Reckon ing Positioning Syste m for Pedestrian Applications," in IEEE Transactions on I nstrumentation and Meas urement , vo l. 6 6, no. 5, pp. 953-963, May 2017. [9] Y. Bar-Shalom, X. R. Li, and T. Kirubarajan, Estimation with applications to tracking and navigation: theory algorithms and software. John Wil ey & So ns, 2004.