AN INTELLIGENT PERSONAL NAVIGATOR INTEGRATING GNSS, RFID AND INS FOR CONTINUOUS POSITION DETERMINATION

Transcription

1 AN INTELLIGENT PERSONAL NAVIGATOR INTEGRATING GNSS, RFID AND INS FOR CONTINUOUS POSITION DETERMINATION Um naegador essoal ineligene inegrando GNSS, RFID e INS ara deerminação conínua da osição. G. RETSCHER Q. FU Insiue of Geodes and Geohsics Vienna Uniersi of Technolog, Ausria gresch@o.uwien.ac.a, fu@mail.zser.uwien.ac.a ABSTRACT Mos of he deeloed edesrian naigaors rel on he use of saellie osiioning (GNSS), someimes also in combinaion wih oher sensors and osiioning mehods. In he rojec Ubiquious Carograh for Pedesrian Naigaion (UCPNAVI) we hae inegraed acie Radio Frequenc Idenificaion (RFID) in combinaion wih GNSS and Inerial Naigaion Ssems (INS) for coninuous osiioning. RFID can be emloed in areas where no saellie osiioning is ossible due o obsrucions, e.g. in urban canons and indoor enironmens. In RFID osiioning he locaion esimaion is based on Receied Signal Srengh Indicaion (RSSI) which is a measuremen of he ower resen in a receied radio signal. The receier can comue is osiion using arious mehods based on RSSI. In oal, hree differen mehods hae been deeloed and inesigaed, i.e., cellbased osiioning, rilaeraion and RFID locaion fingerrining. These mehods can be emloed deending on he densi of he RFID ags in he surrounding enironmen roiding differen leels of osiioning uracies. B inegraing he hree mehods for osiioning ino an inelligen sofware acage and deeloing a nowledge-based ssem i is ossible o deermine he edesrian osiion auomaicall and ubiquiousl. The conce of he inelligen sofware acage is resened and described in he aer. For imroemen of he osiioning urac of cell-based osiioning a modificaion has been deeloed, he so-called imebased Cell of Origin (CoO) osiioning mehod. This mehod uses also he measured RSSI aboe a cerain hreshold which is measured onl if he user is locaed er close o he RFID ag. The es resuls showed ha he urac of osiioning using ime-based CoO is in he range of.3 m. For coninuous Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

2 78 An inelligen ersonal naigaor inegraing GNSS, RFID and... osiioning of he edesrian user, a low-cos INS is emloed in addiion. Since he INS comonens roduce small measuremen errors ha umulae oer ime and cause drif errors, he osiions deermined b RFID would be needed regularl for udae. For he combined osiioning of RFID and INS a ime-aring Kalman filer is emloed. The aroach is esed in indoor enironmen in an office building of our uniersi. For he combined osiioning, an urac of around. m for coninuous osiion deerminaion is achieed. The new aroach and he es resuls are also described in his aer. Kewords: GPS/INS; Inegraion; Naigaion; Mobile; Muli-Sensor.. INTRODUCTION Personal naigaion and guidance serices usuall rel on GNSS osiioning and herefore heir use is limied o oen areas where adequae saellie signals can be receied. If he user moes in obsruced urban enironmen or indoors, alernaie locaion mehods are required o be able o locae he user coninuousl. In our aroach GNSS osiioning is combined wih a MEMS-based Inerial Measuremen Uni for coninuous osiion deerminaion. In addiion, Radio Frequenc Idenificaion (RFID) Locaion Mehods are emloed o relace GNSS in obsruced areas. RFID can also be used for osiioning, because he locaion esimaion can be based on signal srengh measuremens (i.e., receied signal srengh indicaion RSSI) which is a measuremen of he ower resen in a receied radio signal. Then he mobile receier can comue is osiion using arious mehods based on RSSI. Toall, hree differen mehods hae been deeloed and inesigaed, i.e., cell-based osiioning, rilaeraion using ranges o he surrounding RFID ransonders (so-called RFID ags) deduced from RSSI measuremens and RFID locaion fingerrining. In mos common RFID alicaions osiioning is erformed using he cell-based osiioning rincile. In his case, RFID ags can be insalled a acie landmars wih nown locaion in he surroundings. Then he user is carring an RFID reader and is osiioned using Cell of Origin (CoO). The achieable osiioning urac hereb deends on he size of he cell defined b he maimum read range of he signal. Using long range acie RFID his read range can be quie large, i.e., u o 4 o 8 m. Higher osiioning uracies can be achieed using a modificaion of cell-based osiioning which is called ime-based CoO. In his aroach, he locaion of he RFID ag is used o describe he curren osiion of he user onl if he receied signal srengh from he ag is aboe a cerain hreshold. The maimum in signal srengh usuall onl aears when he user is currenl locaed er close o he ag s osiion. If more han one maimum in signal srengh is deeced a differen imes, hen he mean alue for he ime eoch is aen when he user is neares o he ag. This deeloed aroach roides osiioning uracies on he one meer leel and will be inroduced in his aer. Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

3 Rescher, G. and Fu, Q. 79 Aar from cell-based osiioning also rilaeraion and locaion fingerrining can be emloed when he RSSI of more han wo RFID ags can be read a he same ime. Posiioning uracies on he one o seeral mere leel can be achieed for a coninuousl moing user. GNSS and RFID as well are hen inegraed wih INS osiioning for coninuous osiion deerminaion of a edesrian. INS measuremens would be uilized o calculae he rajecor of he user based on he mehod of sra down mechanizaion. Since he INS comonens roduce small measuremen errors ha umulae oer ime and cause drif errors, he osiions deermined b RFID or GNSS would be needed regularl o eliminae and reduce hese errors. All obseraions are hen inegraed in a Kalman filer o esimae he user s osiion and eloci. B inegraing he aboe menioned measuremens ino an inelligen sofware acage he deeloed ersonal naigaor will enable o deermine he mobile user s osiion coninuousl, auomaicall and ubiquiousl. This aer is organized as follows: firs of all, he differen RFID osiioning mehods are briefl described in secion followed b a deailed descriion of coninuous osiioning using RFID and INS in secion 3. Indoor osiioning es resuls are resened in secion 4. In secion 5, he conce for he deelomen of an inelligen sofware acage is inroduced followed b concluding remars in secion 6.. RFID FOR POSITIONING OF A PEDESTRIAN In RFID osiioning of a edesrian, he locaion esimaion is based on RSSI which is a measuremen of he ower resen in a receied radio signal. The receier can comue is osiion using arious mehods based on RSSI. In oal, hree differen mehods hae been emloed, i.e., cell-based osiioning, rilaeraion using ranges o he surrounding RFID ags deduced from receied signal srengh measuremens and RFID locaion fingerrining (Fu, 8). These echnologies can be emloed deending on he densi of he RFID ransonders (so-called ags) in he surrounding enironmen. For osiioning wih RFID eiher readers or ags can be laced a nown locaion in he surrounding enironmen. We hae chosen a low-cos conce where he less eensie ags are deloed in he surrounding enironmen a acie landmars (i.e., nown locaion) or a regular disances. The mobile user is carring a reader in form of a PC-card, which can be lugged ino he mobile deice (e.g. a oce PC or lao). The mos sraighforward mehod is cell-based osiioning. The maimum range of he RFID ag defines a cell of circular shae in which a daa echange beween he ag and he reader is ossible. Seeral ags locaed in he smar enironmen can oerla and define cerain cells wih a radius equal he read range. The urac of osiion deerminaion is defined b he cell size. Using acie RFID ags he osiioning urac herefore ranges beween a few meers u o Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

4 7 An inelligen ersonal naigaor inegraing GNSS, RFID and... ens of meers. Howeer, he urac could be imroed b using he so-called ime-based CoO. In ime-based CoO wo imroemens of he sandard cell-based osiioning hae been made o ge a higher osiioning urac. Firs of all, a hreshold alue is se o reduce he size of he cell. Secondl, he mean alue of he corresonding ime is calculaed for all signal srengh measuremens aboe he hreshold (comare Figure 3). As a resul, he osiioning urac is imroed. The RFID ime for each deeced ID is he mean alue of he corresonding ime. The locaion deermined in his wa ensures ha he calculaed osiion is closes o he rue osiion of he RFID ag. The aroach aes he fac ino oun ha he receied signal srengh is highl ariable in indoor enironmens wih a large number of obsacles and moing objecs which affec he roagaion of he RFID signals. For erificaion of he RFID ime-based CoO measuremens a ool was deeloed under he enironmen of Microsof Visual Sudio 8. If he user asses b an RFID ag a marer can be se in he rogram b a simle mouse clic cauring he ssem ime. This is used as an indicaion for he user currenl being neares o he rue locaion of he RFID ag. The erificaion ool is called ime daa caure ool. The nown RFID ag coordinaes are regarded as rue osiions a his oin of ime. The locaion deermined b he inegraion of RFID and INS a he corresonding oin of ime is he esimaed osiion. The differences beween hese wo osiions are he esimaed errors. As an alernaie o ime-based CoO, rilaeraion and locaion fingerrining hae been inesigaed. Trilaeraion can be emloed if he ranges o seeral ags in he surrounding enironmen can be deermined. Then hese ranges are used for inersecion. The range from he anenna of he reader o he anenna of he ag is deduced from he conersion of signal srengh ino disances. Sraegies for he conersion of he signal srengh measuremens ino disances are disinguished beween indoor and urban oudoor enironmen. I was found ha a simle olnomial relaionshi beween he signal srengh and he range roides reasonable resuls (see Fu and Rescher, 9b). The highes osiioning uracies can be obained wih locaion fingerrining. In case of RFID locaion fingerrining (Kaemarungsi, 5; Rescher e al., 7), RSSI is measured in a raining hase a nown locaions and sored in a daabase. In he osiioning hase, hese measuremens are used ogeher wih he curren measuremens o obain he curren locaion of he user. Locaion fingerrining, howeer, is more cosl and comlicaed in comarison o cell-based osiioning and rilaeraion. For his mehod differen adanced aroaches hae been deeloed and esed (see Fu, 8). For he creaion of he daabase in RFID locaion fingerrining inerolaion mehods can be used, in order o achiee a furher imroemen of he osiioning urac. To es he differen mehods eerimens hae been conduced in a es bed near and in he uniersi building of he Vienna Uniersi of Technolog in Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

5 Rescher, G. and Fu, Q. 7 downown Vienna. The conduced eerimens (see e.g. Fu, 8; Fu and Rescher, 9b) showed ha hese aroaches are suiable o naigae he user wih differen osiioning uracies, i.e., lower osiioning uracies on he seeral meer leel in oudoor enironmen using cell-based osiioning and higher osiioning uracies on he one meer leel in indoor enironmens wih rilaeraion and fingerrining. 3. CONTINOUS POSITIONING WITH RFID AND INS The RFID osiioning is resriced, howeer, o areas where a leas one RFID signal can be deeced. If here is lac of coerage of signals of he RFID ags, he RFID reader will lose is caabili for coninuous osiioning. In order o oercome hese shorages, we hae inegraed a low-cos Inerial Naigaion Ssem (INS) in addiion. In he following, firs of all he deerminaion of rajecories using an INS is elained and hen he fusion of RFID and INS is discussed and resened. The aroach is erified b field ess and he resuls of he eerimen are resened in he ne secion. 3. Trajecor Deerminaion wih Low-cos Inerial Naigaion Ssem (INS) Generall, inerial ssems are caegorized in wo grous (Lawrence, 998; Gabaglio, ): gimbals and sradown INS. The gimbaled ssems are hea and large. Hence, he are unsuiable for edesrian naigaion. In conras o gimbaled ssems a sradown ssem has adances in sensor deelomen (Barbour, ) in erms of size, recision and cos, as well as in comuaion caabiliies of rocessor and can, herefore, be uilized for edesrian naigaion. The erm sradown can be daed bac o he echnique ha modern ssems hae remoed mos of he mechanical comlei of laform ssems b haing he sensors aached rigidl, or sraed down, o he bod of he hos ehicle (Tieron and Weson, 5). Normall, an INS is comosed of hree groscoes and hree eleromeers. All he sensors are mouned orhogonal. The are used o measure he angular rae and o caure he eleraion in one of he hree direcions. I is o be noed ha he measuremens are in he righ handed Caresian coordinae ssem. This coordinae ssem is bod-fied o he deice and is defined as he bod frame. The bod frame changes wih resec o he naigaion frame. The relaionshi beween hese wo frames can be described b aiude arameers which are defined as he orienaions in sace of he INS bod frame (see Figure ). Usuall, he aiude is numericall reresened b a 33 mari R ha is an orhogonal endomorhism. Generall, he aiude can be arameerized in wo was: using Quaernions and using Euler angles. Euler angles ( φ, θ, ) are equialen o roll, ich and aw (see Figure ). Quaernions q, q, q, q 3 are an efficien, non-singular descriion of 3-D orienaion and hae adanages from a numerical and comuaional oin of iew (Shuser 993). Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

6 7 An inelligen ersonal naigaor inegraing GNSS, RFID and... Figure - The INS bod frame (,,z) and he naigaion frame (X,Y,Z) INS obains measuremens for he rae of urn using a groscoe and eleraion using an eleromeer. These measuremens need o be inegraed oer ime o obain orienaion changes and eloci measuremens. Then he curren osiion could be deried b inegraing he obained orienaion changes and eloci measuremens oer ime if he sar osiion could be gien for he inegraion. The sradown mechanizaion in our aroach uses he orienaion daa from he INS (uni quaernions q, q, q, q 3 or Euler arameers φ, θ, ) and he calibraion daa (rae of urn from he groscoe gr _b, gr _b, gr z_b and eleraion from he eleromeer,, z ) o obain he rajecor of he INS. This rocess is diided ino wo ses. Firsl, he inu daa are used o calculae he free eleraion _b, _b, z_b and he roaion mari R nb, whereb, in he free eleraion he grai and cenrifugal force are no included, and he roaion mari is used for he ransformaion beween he bod frame b and he naigaion frame n. In he second se, he osiion of user _n, _n, z_n in he naigaion frame is comued b inegraing eleraion and eloci oer ime. The Euler angles φ nb, θ nb and nb are calculaed b using he quaernions q, q, q, q 3 which can be obained from he sensor direcl: φ (, ) 3 3 nb aan q q q q q q π θ ( 3 ) nb = asin q q q q () 8 (, nb aan q q q q 3 q q ) Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

7 Rescher, G. and Fu, Q. 73 Afer ha he orienaion angles φ, θ and regarding he naigaion frame of he sensor can be obained b inegraing he rae of urn gr _b, gr _b, gr z_b oer he ime: φ φ nb θ = θ nb nb sinφ nb sinθ nb cosθ nb cosφ nb sinφ nb cosθ nb cosφ nb sinθ nb gr gr cosθ nb _b _b sinφ nb gr _b gr _b d cosφ nb gr z_b gr z _b cosθ nb () Then he roaion mari R nb from he bod frame b o he naigaion frame n can be comued: R = R z R R nb θ φ cosθ cos = cosθ sin sinθ sinφ sinθ cos cosφ sin sinφ sinθ sin cosφ cos sinφ cosθ cosφ sinθ cos sinφ sin cosφ sinθ sin sinφ cos cosφ cosθ (3) Noe ha he ransose of he roaion mari R nb is equialen o he roaion mari R bn from he naigaion frame n o he bod frame b. Now he free eleraion will be deduced. The free eleraion here is he second deriaie of he osiion ha does no include he eleraion due o grai and cenrifugal force in conras o he original measured linear eleraion. This is inheren o all eleromeers (Xsens, 7). Therefore, he grai and cenrifugal force mus be subraced from he measured linear eleraion, so ha he free eleraion could be obained. _ b _ b z_ b = z gr _b gr _b _ b gr _b gr _b _ b gr z_b gr z_b z_ b ( R ) T 443 nb R bn z (4) Concerning ha here are drifs in he measuremens of rae of urn and eleraion, one more measuremen was carried ou b eeing he Xsens deice Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

8 74 An inelligen ersonal naigaor inegraing GNSS, RFID and... unmoed for a shor ime eriod, in order o find ou he drifs. These drifs (gr _b, gr _b, gr z_b and,, z ) are subraced from he measuremens while he INS sensor was moed (see equaion 4). Afer he free eleraion and roaion mari hae been calculaed from he original measured orienaion and calibraion daa, he would be inegraed oer ime o obain he orienaion changes and eloci measuremens. Then he curren osiion could be deried b inegraing he obained orienaion changes and eloci measuremens oer ime and using he osiion from he las calculaion. These rocesses occur in he second se of he deerminaion of he rajecor using INS. The curren user osiion mus also be ransformed ino he naigaion frame n. The firs inegraion of he eleraions _b, _b, z_b leads o he elociies _b, _b, z_b in he bod frame b. Then he elociies _n, _n, z_n in he naigaion frame n can be obained. The 3-D coordinaes _b, _b, z_b in he bod frame b can be calculaed b inegraing he aboe obained elociies oer ime. Then he user osiion _n, _n, z_n in he naigaion frame n can be obained. This rocess has o be done recursiel saring from he osiion from he las calculaion. 3. Fusion of RFID and INS Posiioning In our aem resened in his aer RFID ime-based CoO (comare secion ) is combined wih he INS osiioning. Then he absoule osiioning wih RFID can be used o correc he drif of he INS which is caused b he umulaion of errors of he sensors. For he inegraion of INS wih eiher GNSS or RFID osiions usuall a Kalman Filer is emloed. In he algorihm, he sradown INS local geograhic naigaion frame mechanizaion (Tieron and Weson, 5) is combined wih he hree-aes inerial error model (Brown and Hwang, 997) and he RFID ime-based CoO mehod o roduce urae and coninuous osiioning esimaions. A basic 9-sae dnamic model is used as he RFID/INS Kalman Filer model. In he model, he sae ecor conains hree osiion errors, hree eloci errors and hree Euler angle errors. Using such a Kalman Filer a meaningful inegraion of INS wih RFID can be erformed (Fu and Rescher, 9a). In our aroach, we used a so-called ime-aring Kalman filer deeloed in Malab rogramming enironmen which is a generalizaion of he sead-sae filer for ime-aring ssems wih nonsaionar noise coariance and is gien b he recursions. The coninuous-ime sae equaion of he ime-aring Kalman filer is gien b: Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

9 Rescher, G. and Fu, Q & & & & & & S u z_b _b _b B A w w * () w gr a ime Q coariance wih noise rocess gaussian * mari noise ecor inu ) ( mari inu a ime ecor sae ) ( mari ransiion sae a ime ecor sae ) ( * * * * cos sin sin cos = (5) Equaion 5 can be ransferred ino a discree-ime sae equaion: w S u[] z_b _b _b B [] A ] [ ww w w gr a ime Q coariance wih noise rocess gaussian ] [ mari noise or inu ec mari inu a ime ecor sae n mari ransiio sae a ime ecor sae cos sin sin cos cos sin sin cos = (6) The measuremen equaion is: Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

10 76 An inelligen ersonal naigaor inegraing GNSS, RFID and... _ n _ f _ n _ f _ n { [ ] = _ n f n measuremen noise _ n_ f [ ] C wih coariancer (7) obseraion mari f obseraion ecor a ime [ ] wih noise sae ecor a ime a ime The ssem noise coariance mari Q is: Q ww σ = _b, σ _b Q uu σ _b = σ _b σ gr z_b (8) where σ = is he eloci noise in he ais in [m/s], _b σ = is he eloci noise in he -ais in [m/s], _b σ =. is he eleromeer noise in he -ais in [m/s ], _b σ =. is he eleromeer noise in he -ais in [m/s ] and _b σ gr z_b =.7 is he groscoe noise in he z-ais in [rad] wih he bias =. [ m / s ] and drif =.7 [ rad/ s] as gien b he sensor manufacurer. The obseraion error coariance mari R is: R σ _ n = σ _ n (9) where σ >= [m] and σ >= [m] (R mari for RFID is indicaed b R RFID ), _n _n Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

11 Rescher, G. and Fu, Q. 77 σ = [m] and σ = [m] (R mari for INS is indicaed b R INS ). _n _n Using new measuremens he sae ecor can be recursiel udaed in he Kalman filer rocess. 4. INDOOR POSITIONING TEST This secion resens he inegraion of RFID and low-cos INS for coninuous osiioning in an indoor enironmen. In he rojec, a low-cos INS from Xsens, he MTi, has been emloed. For calculaing he osiions from he measured daa of he sensor he sradown mechanizaion is used. Furhermore, a ime-aring Kalman filer is emloed o correc he osiion and eleraion resuled from he sradown mechanizaion. RFID ime-based CoO osiioning is uilized o deermine he curren osiion of he user, when he RFID reader deecs a signal from an RFID ag in he surrounding enironmen. This deermined osiion will be needed o udae and correc he rajecor calculaed b he INS, since he INS comonens roduce small measuremen errors ha umulae oer ime and cause drif errors. Figure - Indoor osiioning es enironmen. Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

12 78 An inelligen ersonal naigaor inegraing GNSS, RFID and... The indoor osiioning es was conduced on he 3 rd floor in an office building of he Vienna Uniersi of Technolog (see Figure ). The rajecor sars in fron of he eleaor (referred o as Lif in Figure ) and leads o a general eaching room along a corridor around he corner. The roue is oall 3.9 m long and can be diided ino wo secions which are recangular o each oher. The second ar of he roue runs along he middle line of he corridor and has a lengh of 5.9 m. The reference rajecor is shown as solid line in Figure. In oal, seen RFID ags were mouned in he es bed eenl disribued. Tags were susended from he ceiling in a heigh of. m aboe he ground. In his eerimen cell-based osiioning was emloed for RFID osiioning. Figure 3 shows he signal receion using cell-based osiioning including he informaion of he RSSI, he ime he ag was deeced, and he ID of he ag. Each ag is idenified b is resence or absence and he measured signal srengh RSSI. As indicaed b he circles, RSSI measuremens which were higher han he hreshold alue of -46 dbm are seleced for ime-based CoO. Figure 3 - Signal receion using cell-based osiioning. The osiion deermined using RFID cell-based osiioning was uilized in order o udae he rajecor calculaed b using he measuremens of he INS. The Xsens MTi sensor bandwidh was Hz. Addiionall, a Kalman Filer is used o correc he osiion and eloci dnamics of he INS sensor. Figure 4 shows he filer resul of he INS rajecor. The dashed line reresens he measured resonse Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

13 Rescher, G. and Fu, Q. 79 b using he sradown mechanizaion wihou filering, while he solid line reresens he resul of he osiions filered b he Kalman Filer. The error in osiion increased oer ime wih a maimum error in -direcion of around.5 m and a maimum error in -direcion of around 5. m. Howeer, he error could be significanl reduced using he Kalman Filer. Figure 4 denoes ha all oins are osiioned wih a maimum radial deiaion of.65 m and wih a mean radial deiaion of around.9 m. I can be seen ha he mean radial deiaion lus sandard deiaion is in he range of - m. Therefore, an urac in his range for coninuous osiioning in indoor enironmens can be achieed. Figure 4 - Filer resul of he INS rajecor. Table includes he esimaed error of osiioning using he inegraed ssem. I can be seen ha he maimum error in -direcion is.4 m, while he maimum error in -direcion is.4 m. The maimum error in osiion is.5 m. In Table, d f is he deiaion beween he filered and rue locaion of he ags along he -ais, d f is he deiaion beween he filered and rue locaion of he ags along he -ais and dr f is he radial deiaion beween he filered and rue locaion of he ags. Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

14 7 An inelligen ersonal naigaor inegraing GNSS, RFID and... Table - Esimaed osiioning error of inegraed ssem. The eerimen showed ha he deerminaion of he rajecor using an inegraion of RFID and he INS achieed an urac of around. m in wo direcions on he - lan. Furhermore, he maimal error of he osiioning was.4 m along he -ais and. m along he -ais. This urac is suiable for mos osiioning alicaions in indoor enironmens. 5. DEVELOPMENT OF AN INTELLIGENT SOFTWARE PACKAGE As resened in he inroducion and in secion, hree differen mehods using RFID hae been inesigaed and emloed for osiioning in indoor and urban oudoor enironmens, i.e., cell-based osiioning (RFID CoO and RFID ime-based CoO), rilaeraion and RFID locaion fingerrining. The conduced ess demonsrae ha hese hree differen RFID osiioning mehods are quie aroriae for osiioning in differen enironmens such as an urban and indoor enironmen and a ransiion zone beween hese wo. When he user wals from one enironmen o anoher, howeer, he mehod of osiioning canno be swiched auomaicall. In oher words, currenl he corresonding sofware acage of he osiioning mehods has o be seleced manuall when he enironmens change. In his secion, a roosed conce is inroduced for an inelligen nowledge-based sofware enironmen for coninuous osiioning in comle enironmens. As menioned aboe, he hree differen osiioning mehods are enironmen secific. For coninuous osiioning in comle enironmens, hese hree mehods should be combined and inegraed ino one sofware acage. In addiion, he ssem should be able o auomaicall idenif he e of enironmen Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

15 Rescher, G. and Fu, Q. 7 encounered b he user a his ime oin. This requires an inelligen inegraion of hese hree mehods b a nowledge-based ssem (Fu and Rescher, 8). Figure 5 shows a reliminar conce for such an inelligen inegraion. The main focus is o deelo an inelligen sofware enironmen. The sofware should suor cauring he measuremens from he emloed sensors in real-ime and rocessing he measured daa wihin he required ime. A he same ime, he sofware should idenif he curren locaion enironmen and sar he corresonding mehod of osiioning auomaicall. Finall, he sofware should hae a user friendl inerface. Wih such a sofware and he nowledge-based ssem, coninuous osiioning in a comle enironmen could be carried ou auomaicall. We roose o use fuzz logic for he deelomen of he inelligen sofware acage. B formulaing a number of condiions he selecion of he suiable RFID osiioning mehod can be erformed. Then he membershi funcions of he fuzz ssem form he nowledge-based ssem. Figure 5 - Inelligen conce for he selecion of he RFID osiioning mehod deending on he enironmen of he edesrian user. 6. CONCLUSIONS AND OUTLOOK This aer addresses he inesigaion of differen mehods and algorihms for osiioning using low-cos acie RFID and INS for urban and indoor Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

16 7 An inelligen ersonal naigaor inegraing GNSS, RFID and... enironmens. For he osiioning using acie RFID oall hree differen mehods hae been deeloed and inesigaed, i.e., cell-based osiioning, rilaeraion using ranges o he surrounding RFID ransonders (so-called ags) deduced from receied signal srengh measuremens and RFID locaion fingerrining. The cell-based osiioning is an algorihm o deermine he locaion of he user in a cell around he RFID ag wih a size defined b he maimum range of he RFID signals. The achieable osiioning uracies deend on he size of he cell, i.e., u o m using our long range RFID equimen. Therefore, his mehod is onl well suied for areas where urac is no ha imor, such as urban oudoor enironmen. Howeer, he urac can be imroed using a self deeloed algorihm ha inesigaes he conribuion of he measured signal srengh around he RFID ag. This aroach is called ime-based Cell of Origin (CoO) and is inroduced in his aer. The es resuls showed ha he urac of osiioning using ime-based CoO is in he range of.3 m. Trilaeraion can be emloed if he ranges o seeral ags are deermined and are used for inersecion. The range from he anenna of he reader o he anenna of he ag is deduced from he conersion of signal ower leels ino disances. Sraegies for he conersion of he signal srengh measuremens ino disances are disinguished beween indoor and urban oudoor enironmen. Using rilaeraion usuall osiioning uracies on he one o a few meers leel can be achieed. For indoor enironmens also he use of RFID locaion fingerrining was inesigaed. For he creaion of he daabase in RFID locaion fingerrining inerolaion mehods can be used, in order o achiee a furher imroemen of he osiioning urac. The es of osiioning using locaion fingerrining showed ha osiioning urac below. m could be achieed. Eerimens hae been carried ou using he hree mehods of RFID osiioning. In general, he eerimens showed hese hree mehods are aroriae for locaing he user wih differen osiioning uracies, i.e., lower osiioning uracies in oudoor enironmen using cell-based osiioning and higher osiioning uracies in indoor enironmens wih rilaeraion and fingerrining. The osiioning is resriced, howeer, o areas where a leas one RFID signal can be deeced. If here is lac of coerage of signals of he RFID ags, he RFID reader will lose is orienaion. In order o oercome hese shorages we roose o inegrae a low-cos Inerial Naigaion Ssem (INS) in addiion. In he rojec, a low-cos INS from Xsens, he MTi, has been emloed. For calculaing he osiions from he measured daa of he sensor he sradown mechanizaion is used. Furhermore, a ime-aring Kalman filer is emloed o correc he osiion and eleraion resuled from he sradown mechanizaion. RFID cell-based osiioning is uilized o deermine he curren osiion of he user, when he RFID reader deecs a signal from an RFID ag in he surrounding enironmen. The deermined osiion will be needed o udae and correc he Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

17 Rescher, G. and Fu, Q. 73 rajecor calculaed b INS, since he INS comonens roduce small measuremen errors ha umulae oer ime and cause drif errors. The aboe conce has been imlemened and esed in a real world enironmen. For he combined osiioning of RFID and INS an urac of around. m for coninuous osiion deerminaion can be achieed using our aroach. From his resul, i can be concluded ha our aroach using an inegraed RFID cell-based and INS osiioning wih a ime daa caure ool is suiable for coninuous osiion deerminaion of a mobile user in challenging indoor enironmens. ACKNOWLEDGEMENTS The research wor resened in his aer is full suored b he FWF Projec Ubiquious Carhogra for Pedesrian Naigaion UCPNAVI (Projec No. P9-N5) of he Ausrian Science Fund (FWF Fonds zur Förderung wissenschaflicher Forschung). REFERENCES BARBOUR, N.M.. MEMS for Naigaion a Sure. In: Paers resened a he Insiue of Naigaion Naional Technical Meeing. Long Beach, CA, Januar -4,. BROWN, R. G., HWANG, P. Y. C.. Inroducion o Random Signals and Alied Kalman Filering. John Wile & Sons, New Yor, 3 rd ediion, 997, 484. FU, Q. Acie RFID for Posiioning Using Trilaeraion and Locaion Fingerrining Based on RSSI. In: Paers resened a he ION GNSS Conference, Seember 6-9, 8, Saannah, Georgia, USA, CD-Rom Proceedings, 4 gs. FU, Q., RETSCHER, G. Using RFID Technolog in Pedesrian Naigaion for Informaion Transmission and Daa Communicaion Recording. In: Paers resened a he Junior Scienis Conference, Noember 6-8, 8, Vienna, Ausria, gs. FU, Q., RETSCHER, G. Anoher Loo Indoors GPS RFID. GPS World, (3), 9a, FU, Q., RETSCHER, G. Acie RFID Trilaeraion and Locaion Fingerrining Based on RSSI for Pedesrian Naigaion. The Journal of Naigaion, 6(),., 9b GABAGLIO, V. GPS/INS Inegraion for Pedesrian Naigaion. PhD hesis a Ecole Polechnique Federale de Lausanne, Swizerland.. KAEMARUNGSI K. Design of Indoor Posiioning Ssems Based on Locaion Fingerrining Technique, Disseraion, 5. Uniersi of Pisburgh, see h://ed.librar.i.edu/etd/aailable/ed353593/unresriced/disser aion8feb5.df (Las essed: Noember 9). Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.

18 74 An inelligen ersonal naigaor inegraing GNSS, RFID and... LAWRENCE, A. Modern Inerial Technolog. in: Guidance and Conrol., Sringer-Verlag. nd ediion. New Yor, gs. RETSCHER G., MOSER E., VREDEVELD D., HEBERLING D., PAMP J. Performance and Accurac Tes of awifi Indoor Posiioning Ssem, Journal of Alied Geodes, Vol., No., RETSCHER, G., FU, Q. GNSS, RFID and INS Inegraion for Pedesrian Naigaion. In: Paers resened a he GPS/GNSS 8 Conference, Too, Jaan, Noember -4, 8, CD-Rom Proceedings, gs. SHUSTER, M.D. A sure of Aiude Reresenaions. Journal of Asronauicals Sciences, 4(4), 993, TITTERTON, D.H., WESTON, J.L. Sradown Inerial Naigaion Technolog. Insiuion of Engineering and Technolog; nd reised ediion, 5. XSENS. MTi and MT User Manual and Technical Documenaion, Produc Manual, Xsens Technologies B.V., The Neherlands. 7. Bol. Ciênc. Geod.,. 5, n. 5 Secial Issue on Mobile Maing Technolog, , 9.