Home

MScMS (Mobile Spatial coordinate Measuring System)

1. 0 10 20 30 40 50 60 70 Misalignment Angle degrees TOF Standard Deviation Vs Misaligment Angle d 2034 mm 10 20 30 40 Misalignment Angle degrees 50 60 70 Average TOF Vs Misaligment Angle d 3671 mm Average TOF Error us TOF Error Standard Deviation us 0 10 20 30 40 50 60 70 Misalignment Angle degrees TOF Standard Deviation Vs Misaligment Angle d 3671 mm 10 20 30 40 Misalignment Angle degrees 50 60 70 Fig 4 11 TOF average value and standard deviation depending on the misalignment angle 0 and the battery level V for different transceivers distances 46 Chapter 4 As already noticed TOF Error increases depending on 0 and d Also the TOF standard deviation is slightly increasing with the angle this behaviour is more definite for large distances between transmitter and receiver With respect to the experimental data in Fig 4 11 there are some measurements not included in the factorial plan They are TOF Error measurements related to misalignment angles of 70 degrees which cannot be performed for all the possible distances For in stance considering the long transceivers distance d3 3871 mm transmitter and receiver are not able to communicate because of the strong signal attenuation It can be noticed that
2. b Reply RF and authorization WWW for signals transmission RFE US a 1 INIIV E c Concurrent transmission of RF NW and US signals ee Fig 2 2 Communication scheme implemented by Cricket devices Priyantha et al 2000 d The receiving devices measure the time lapse between reception of RF and US signals see Fig 2 3 transmitting device receiving device RF c speed of electromagnetic radiation LELETTTTTTI a TTT TT ree LEEETETT query RF signal ee RF authorization for 45 rareropkaidip ipo niikanmns pinasara next transmission none anannnnnnun mansunnunnn PTT TTL seueeneeeee snnansanunnn concurrent transmission of RF and US signals At TDoA time lapse between reception of RF and US signals Fig 2 3 Time evolution of RF and US signals qualitative scheme 12 Chapter 2 The distance between two devices is calculated by the following formula a 2 1 S C where c is the speed of electromagnetic radiations s the speed of sound and At is TDoA Gustafsson and Gunnarsson 2003 Due to the large difference between c about 300 000 km s and s about 340 m s in air with temperature T 20 C and relative humidity RH 50 d s At 2 2 2 3 0 Crickets communication Cricket devices build a wireless network of cooperating sensor nodes To preserve net work scalability that is to make sure that the amount of information stored by each node is independent from network dimension in ter
3. Experimental evaluation of the MScMS ultrasound transducers 41 section 4 4 3 presents other minor experiments aimed at deepening the factorial plan analysis 4 4 1 Results of the factorial plan Analysing the factorial plan experimental outputs the first interesting result is that the TOF Error standard deviation o changes depending on the TOF Error value In other words the population of TOF Error cannot be considered as homoscedastic that is to say with a constant standard deviation Dispersion of TOF Error 100 r r r ee e w 6O SO a a a a ee er og aoe aoa bewweereeee Tae dal E 5 T a a ale eae ae eae zwar g eE r3ig 11 l A SSS E ee ae eee EEE a id Sp ste gust js 3 eset f L a e UL fea o Eeit COATTA 7 Hide i dita ig v Hi His a ake at T HH E Ti a TRA jp i SIE i IEE gt ee ae ara ror rrr r i Tr rr rrr rtr oe aa ee aS Se H ie 41 ER g D i ecco tC 0 T AEA i fee ae ee ae S a T i se aa er ee ar 2 e ho sao ses 222 Aa eee e026 6 4 6 6 shoe ca se ree 26222 4 s ees sees ees es ee ss 6 26 a 2s a nee ah manm T OE E An R R R iat Sm g j E 60 a a a a E E S NNS i ae i eee See rer errr cages tears arenes ep oe ce ee cana cn antenatal este Tunes ces Centon estos ee eo fenios E s nae of eve EE A cates A E celeste L gt 100 0 50 100 Average TOF Hror ps 250 300 350 Fig 4 6 TOF Error standard deviation vs average TOF Error For each of the 525 factors combi natio
4. Politecnico di Torino Porto Institutional Repository Doctoral thesis Design and development of a new large scale metrology system MScMS Mobile Spatial coordinate Measuring System Original Citation Maisano Domenico 2008 Design and development of a new large scale metrology system MScMS Mobile Spatial coordinate Measuring System PhD thesis Availability This version is available at http porto polito it 2544537 since May 2014 Published version DOI 10 6092 polito porto 2544537 Terms of use This article is made available under terms and conditions applicable to Open Access Policy Article Public All rights reserved as described at http porto polito it terms_and_conditions html Porto the institutional repository of the Politecnico di Torino is provided by the University Library and the IT Services The aim is to enable open access to all the world Please share with us how this access benefits you Your story matters Article begins on next page POLITECNICO DI TORINO DOCTORATE IN PRODUCTION SYSTEMS amp INDUSTRIAL DESIGN DESIGN AND DEVELOPMENT OF A NEW LARGE SCALE METROLOGY SYSTEM MSCMS MOBILE SPATIAL COORDINATE MEASURING SYSTEM Supervisor Ph D Student Prof F Franceschini Domenico MAISANO May 2008 Executive summary This thesis arises from the research activity developed at the Industrial Metrology and Quality Engineering Laboratory of DISPEA Politecnico di
5. 5 MScMS and CMMs a structured comparison sssssssssssessceccccccccccsososssssscceceeceeosoo 1 S TO Oe E A N aaoemereone 1 32 CMMs manr Car aC Cer SCS a s een Aone eases sidienashg heceen beast aemanstlges ehaeerenicee 1 J Comparison Ceran a tata e a a a E 1 P PO D a E E seat 1 vi Contents I2 OW ORKING V OLUIMC woos sate tsi a ei sssagelopaniwlsiasssaceocnn sia 1 tos DE UP e a my Sree ma arr Rea a 1 534 Metrological performanc s ssisscucacananeaiacecavorsnpianda EEE 1 5 3 5 Measurements System diagnostics eesesssssssssssssssseeteeererssssssssssssseeeeereee 1 DG EDSO ISE aa a a 1 I lex DN a N 1 Desh D E L E A E N A T E E T 1 De Sy emn Manar emne i eaa E E A bee RRE 1 3A IMAC ONGIC CRALIONS a3 a a a sditabanstasatacatenabensnsiavessumiaoanceeaaies 1 6 iGPS performance evaluation and comparison with MSCMS sesccossosssssssssceccecseee 1 E MOG CIO a5 esse reese E EE ats T EET EEE 1 GD IGES SUP NL sa a a a l 6 3 iGPS technology and operating features ccccccccccccccccccececeaeeeseseesseeeeeeeeeeeeees 1 6 4 Factors affecting MeasureMent ccccssseeessssssesseeeeecceeeccccccceecccssecesssssssseeeeeeeees 1 6 5 Experimental work for 1GPS preliminary performance analysis ccccceeees 1 60 Sy StCiINS COMPAL IS OM sie dasa a a E 1 6 7 Summary and final considerations cecceesseeeececceeceeecacaaeeeesesesseeeeeeeeeeeeeeeeqaas 1 7 Future wireless sensor networkK
6. Comparison results between MScMS and 1GPS Localisation of the constella Semi automated procedure Semi automated procedure tion devices Technical feature iGPS Two angles among each cou Measured variables l l tion devices ple of sensor and transmitter Localisation technique during Triangulation measurements Transmitter s communication More than 30 m range Number of constellation de 4 or 5 per every 400 m vices Sample rate About 50 points per second Sensibility to environmental Temperature light vibrations conditions Diagnostic function to filter Use of fixed sensors to de l i wrong measurements andto termine whether measure System diagnostics i correct parameters ment system is going out of tolerance Automatic calculation of the Real time adjustments of the System calibration check speed of sound during meas scale urements Position accuracy of about Position accuracy of about 10 20 mm measurement of a 0 5 mm measurement of a Metrological performances single point by a single sam single point by averaging a pling number of scanned in 2 sec onds Working volume size Scalable Scalable Estimated at 10k About 200k for a typical Cost system with four transmitters iGPS performance evaluation and comparison with MScMS 85 In the following paragraphs some of the previous results are individually analysed in order to emphasise the most interesting similarities and differences between the tw
7. No 2 pp 137 143 SAFIGIANNI A S TSOMPANIDOU C G 2005 Measurements of Electric and Magnetic Fields Due to the Operation of Indoor Power Distribution Substations IEEE Transactions on Power De livery Vol 20 No 3 SANDWITH S PREDMORE R 2001 Real time 5 Micron Uncertainty with Laser Tracking Inter ferometer Systems using Weighted Trilateration Proceedings of 2001 Boeing Large Scale Metrology Seminar St Louis SAVVIDES A HAN C STRIVASTAVA M B 2001 Dynamic fine grained localization in ad hoc networks of sensors Proceedings of ACM IEEE 7 Annual International Conference on Mo bile Computing and Networking MobiCom 01 pp 166 179 114 References SCHURGERS C SRIVASTAVA M B 2001 Energy efficient routing in wireless sensor networks Military Communications Conference MILCOM Communications for Network Centric Op erations Vol 1 pp 357 361 SHNAYDER V HEMPSTEAD M CHEN B WERNER ALLEN M WELSH M 2004 Simulating the Power Consumption of Large Scale Sensor Network Applications Proceedings of SenSys 04 Baltimore Maryland November 3 5 2004 SIRBU M LEHR W GILLET S 2006 Evolving Wireless Access Technologies for Municipal Broadband Submitted to Forthcoming in Government Information Quarterly SOTTILE F SPIRITO M 2006 Enhanced Quadrilateral based Localization for Wireless Ad hoc Networks In Proceedings of IFIP Fifth Annual Mediterranean Ad Hoc Networking Worksho
8. Points are defined on a Cartesian coordinate system and then coordinates are processed by specific algorithms in order to determine the surface geometrical features angles distances other objects shapes etc The remainder of this chapter is organised into six sections Sections 6 2 and 6 3 pro vides an introduction to the iGPS technological features and modus operandi Section 6 4 analyses in detail the most important factors affecting measurements Section 6 5 reports on the system performance and the most important factors affecting it as evaluated by a number of initial tests carried out in collaboration with the University of Bath For the experimental work described in this section an 1GPS system equipped with 4 transmitters was used Section 6 6 compares the 1GPS with MScMS emphasising their many common aspects and their differences The comparison is carried out according to a structured set of evaluation criteria Finally Section 6 7 gives the conclusions and future directions of this research 6 2 IGPS structure Before describing the iGPS characteristics here we present a classification of large scale metrology measurement systems see Fig 6 1 These systems can be divided into cen tralized and distributed In the case of centralized systems measurements can be obtained by a single stand alone unit which is a complete system like a Laser Tracker While the distributed instruments are made of two or more distributed uni
9. The resulting wave is given by the sum of the single waves If the receiver is directly facing the transmitter case a the two individual waves are in phase and the resulting wave amplitude has the maximum value If the transmitter is misaligned with the receiver case b the resulting wave is attenuated because of a phase cancelling effect due to the phase difference between the two individual waves Lamancusa Figueroa 1990 The resulting ultrasonic transmitter radiation pattern depending on the transmitter mis alignment angle with the receiver is shown in Fig 4 3 As represented the transmitter US signal strength drops along directions that are away from the direction facing the ultra sonic transducer Similarly the received signal strength can be influenced by the receiver orientation In particular considering the same signal strength from the transmitter the received signal strength is maximum when the receiver s surface is facing the transmitter On the other hand the received signal decreases when the receiver s surface is angled Experimental evaluation of the MScMS ultrasound transducers 37 Fig 4 3 The radiation pattern of the Cricket ultrasonic transducer on a plane along its axis de pending on the orientation Signal strength drops along direction that are away from the normal direction to the transducer surface Several signal methods have been developed for detecting US signals Thresholding It is the simp
10. focusing on e the wireless Crickets devices Subsection 2 3 1 e the measuring method to evaluate mutual distances among Crickets Subsection 2 3 2 e the mobile probe Subsection 2 3 3 2 3 1 Cricket devices Cricket devices are equipped with radiofrequency RF and ultrasound US transceivers Working frequencies are respectively 433 MHz on RF and 40 kHz on US Cricket de vices are developed by Massachusetts Institute of Technology and manufactured by Crossbow Technology Each device uses an Atmega 128L microcontroller operating at 7 4 Mhz with 8 kBytes of RAM 128 kBytes of FLASH ROM program memory and 4 kBytes of EEPROM as mostly read only memory Alimentation is provided by two AA batteries of 1 5 V Balakrishnan et al 2003 10 Chapter 2 Cricket devices are quite small see Fig 2 1 easy to be moved and cheap each unit would cost about 10 20 if mass produced Due to these characteristics they are optimal for ad hoc WSN applications Priyantha et al 2000 Ultrasound Transmitter m m Integrated antenna for RF transceiving Ultrasound Receiver x 4cm photo perspective view orthogonal projection Fig 2 1 Cricket Device Crossbow Technology The US transceivers equipping Crickets are quartz crystals which transform electric energy in acoustic and vice versa piezo electric effect They generate receive 40 kHz ultrasound waves Transmitters excited by electric impulses vibrat
11. the most important effects are due to the US signal attenuation which may have three major sources 1 transceivers distance 2 transceivers misalignment angle 3 transducer battery charge level In particular the paper shows that transducers misalignment and transceivers distance are the most significant This statement is supported by the results of an organic experimental factorial plan It is important to remark that this source of error 1s directly caused by the method of tresholding US detection method Typically attenuation may produce an overestimation of several centimetres up to 3 4 cm in distance evalua tion Also these results can be useful to identify the major MScMS sources of inaccuracy and to determine how the error in TOF evaluation changes in the different points within the Cricket transmitters cones of vision An organic analysis of the combined effect of the transmitter and receiver orientations on TOF error will be the object of a future work Regarding the future Cricket s accuracy could be improved using more refined ranging methods for example based on phase detection with fixed frequency signals and with frequency modulated signals Unfortunately these detection methods are more expensive Another possible solution to the error derived by the transmitter misalignment is the use of omnidirectional ultrasonic transducer like the cylindrical polyvinylidene fluoride PVDF film transducers Toda 2002 The
12. 1 D2 25 1999 Draft Standard for a Smart Transducer Interface for Sensors and Ac tuators Network Capable Application Processor NCAP Information Model Institute of Electrical and Electronics Engineers Inc Piscataway New Jersey January 25 1999 INTEL CORPORATION 2005 Expanding Usage Models for Wireless Sensor Networks Technolo gy Intel Magazine http www intel com technology magazine research sensor networks 0805 pdf INTEL 2006 Improving Life and Industry with Wireless Sensors http www intel com research exploratory wireless_promise htm ISO 1993 International Vocabulary of Basic and General Terms in Metrology 2 edition Ge neva ISO 10360 PART 2 2001 Geometrical Product Specifications GPS Acceptance and reverifica tion tests for coordinate measuring machines CMM ISO 5725 1986 Precision of test methods Determination of repeatability and reproducibility for a standard test method by inter laboratory tests JACOBS S 2004 Battery Power for Remote Wireless Sensors Sensor Magazine Online March 2004 http www sensorsmag com articles 0304 42 main shtml JOHANSSON J GUSTAFSSON M DELSING J 2006 Ultra low power transmit receive ASIC for battery operated ultrasound measurement systems Sensors and Actuators Vol 125 pp 317 328 KANG S TESAR D 2004 A Noble 6 DOF Measurement Tool With Indoor GPS For Metrology and Calibration of Modular Reconfigurable Robots IEEE ICM Internation
13. 2003 Sensor networks Evolution opportunities and chal lenges Proceedings of the IEEE Vol 91 No 8 pp 1247 1256 CHEN M CHENG F GUDAVALLI R 2003 Precision and Accuracy in an Indoor Localization System Technical Report CS294 1 2 University of California Berkeley USA CRATER K C 1992 When Technology Standards Become Counterproductive http www ctc control com tutorials language counter htm Cross N R DOTSON J R FLANK D R PEGGS G N Cox M G FORBES A B CORTA R O DONNELL J AND PRIETO E 1998 A Large Reference Artefact for CMM Verification National Physical Laboratory NPL Report CLM 6 May 1998 110 References CROSSBOW TECHNOLOGY 2008 http www xbow com CURTIS M A FARAGO F T 1994 Handbook of dimensional measurement third edition Indus trial Press Inc New York DEA 2007 http www dea it DELEPAUT C VANDENDROPE L AND EUGENE C 1986 Ultrasonic three dimensional detection of obstacles for mobile robots In Proceedings of 16th International Symposium of Industrial Robots ISIR 86 Brussels pp 483 90 Doss R C CHANDRA D 2005 Reliable Event Transfer in Wireless Sensor Networks Deployed for Emergency Response In Proceedings of the 17th ASTED International Conference on Parallel and Distributed Computing and Systems PDCS 2005 Phoenix AZ USA FIGUEROA J F LAMANCUSA J 1992 A method for accurate detection of time of arrival analy sis and de
14. Aieeeaniaceuenaaenens l 24 MSeMS sofware architecte os csctcc ac sschis wi taadoaiinagess axeculnnscneasv qeseaweaGeaereiaeets 1 2 4 1 Location of Crickets mounted on the mobile probe sssseeesessssseeerssssss 1 2 4 2 Location of points touched by the probe tp eessseesssssssseerssssssseeeressssse 1 24 y C ncke Gwae ee a a a a 1 2 4 4 Semi automatic location of the constellation 00000soooooooeeeeeesssssssssssssee 1 De IVESOMS Protot Pe sisina a a cedevessvesetss 1 JE ANE O CUCU ON e E A A E A ROER 1 3 2 Description of the first MSCMS prototype cccccccccccccsssssssssessesseeeeeceeeeeeeeeaaas 1 3 3 MScMS actual performance critical aspects and possible improvements l J Alia CONSIGSTAUIONS a A a O l 4 Experimental evaluation of the MScMS ultrasound transducers 0000 1 A ali nrod ON osics wseesans esrescesuasdsan ep saacnsy NN 1 4 2 Piezoclecine US 1ransduce isen a beauesasnacmcen saunesacumocenosmaneesncsnes 1 43 Factors affectine US transceivers wate sacle ie E cake lie cated 1 4 4 Analysis of the experimental results 0 0 0 cccccccsssssssssssssessseeeeeececeeeeeeeeeeaeaeeessees l 4 4 1 Re sults of the factorial plam icxcsenteiiessseealaasdactiataisesbasadsthesssdeuadaasdaatsaaneateseacs 1 4 4 2 Interpretation of the results cccccssseeccccceeccceceeeeeeeeesssesseeeeceeeeeeeeeeeaaas 1 AD LAGIION Al CX PMIINENUS e A 1 4o UVa VOUS S and MUE WOK asanes as a l
15. TOF measurement error can be also reduced by implementing proper compensation techniques 5 MScMS and CMMs a structured comparison 5 1 Introduction The goal of this chapter is comparing MScMS with well tested and widespread instru ments such as classical Coordinate Measuring Machines CMMs MScMS and CMMs have many common aspects For both the systems measurements are taken touching few points on the objects surface with a probe tip points are defined on a Cartesian coordinate system and then coordinates are processed by specific algorithms in order to determine geometrical features angles other objects shapes etc On the other hand MScMS and CMMs have many different characteristics such as their physical structure size cost etc This comparison will be carried out according to a structured set of evaluation criteria The chapter is organised in five sections Section 5 2 refers to CMMs main characteris tics Section 5 3 illustrates the comparison criteria with which MScMS and classical CMMs will be compared Section 5 4 shows the results of this comparison Finally the most important results are summarized 5 2 CMMs main characteristics The CMMs are complex mechanical devices to determine the coordinates of the points touched by an electromechanical probe CMMs can be controlled either manually or by Computer Numerical Control CNC systems they are available in a wide range of sizes and designs offering a variety of differ
16. WANG L AND XI F 2006 Challenges in design and manufacturing International Journal of Computer Integrated Manufacturing Vol 19 No 5 pp 409 410 WARRIOR J 1997 Smart Sensor Networks of the Future Sensor Magazine Online March 1997 http www sensorsmag com articles 0397 net_mar main shtml WEISS L 2002 Developing Tangible Strategies Design Management Journal Vol 13 No 1 pp 32 38 WELCH G BISHOP G VICCI L BRUMBACK S KELLER K 2001 High Performance Wide Area Optical Tracking The HiBall Tracking System In Presence Teleoperators And Virtual Environments Vol 10 No 1 pp 1 21 ZAKRZEWSKI J 2003 Error and Uncertainty Reduction Challenge for a Measuring Systems Designer Measurement Science Review Vol 3 Section 1 pp 31 34
17. a diagnostic tool with the purpose of filtering wrong distances among Cricket devices US reflection diffraction or other measuring accidents Moore et al 2004 Wireless device 1 Protes range of vision 1000 g Wireless device z2 Fig 5 3 Graphic representation of the probe range of vision The right part of this screenshot shows the constellation devices seen by the mobile probe Crickets On the other hand CMMs do not offer on line diagnostics for single point measure ments but only for shape measurements if the reconstructed shape does not reasonably fit the measured points then a warning signal is reported This kind of diagnostics is only possible when there is a significant measurements redundancy for example five or more points to construct a sphere or four or more to construct a circumference Similar diag nostic tools can be implemented for MScMS Off line measurements diagnostics Both CMMs and MScMS can provide very similar off line diagnostic tools These diag nostics are based on the concept of measurement replication if variability is higher than expected measurements are considered not reliable Franceschini et al 2007 II During a measurement cycle some known points are repeatedly touched by the probe at regular MScMS and CMMs a structured comparison 63 intervals If the variability of these points measurements is larger than expected the measurement cycle stops because this i
18. accurate measurements on objects which are even complexly shaped are widespread On the other hand CMMs are generally bulky and not always suitable for measuring large size objects for example longerons of railway vehi cles airplane wings fuselages etc because the working volume is limited ISO 10360 part 2 2001 In general for measuring medium large size objects portable systems can be preferred to fixed ones Transferring the measuring system to the measured object place is often more practical than the vice versa Bosch 1995 Systems as theodolites tacheometers photogrammetry equipments Laser Trackers or GPS rather than CMMs can be easily installed and moved Pozzi 2002 However they can have some other drawbacks as mentioned in the remaining of this thesis Section 2 2 2 Chapter 1 1 1 The Mobile Spatial coordinate Measuring System MScMS This thesis introduces a new measuring system called Mobile Spatial coordinate Measur ing System MScMS developed at the Industrial Metrology and Quality Laboratory of DISPEA Politecnico di Torino MScMS has been designed to perform simple and rapid indoor dimensional measurements of large size objects large scale metrology An essen tial requirement for the system is portability that is the aptitude to be easily transferred and installed MScMS is made up of three basic parts 1 a constellation of wireless devices Crickets 2 a mobile probe and
19. be tween measured distances evaluated by means of US transceivers and calculated dis tances determined on the basis of the localised position We have experimentally veri fied that the minimum value of the EF is generally of the order of the tenth of mm When one or more measured distances are wrong due to systematic effects the EF minimum value explodes becoming 3 or 4 orders of magnitude greater In practical terms during the location of devices A and B if the EF minimum is included below a threshold value say 70 mm then the position is considered to be reasonable Otherwise it is rejected Speed of sound dependence on environmental conditions Speed of sound s value makes it possible to turn US time of flight into a distance Eq 2 2 It is well known that the speed of sound changes with air conditions temperature and humidity which can exhibit both temporal and spatial variations within large work ing volumes As a consequence s requires to be often updated depending on the time and the position A partial solution to this problem is to use the temperature T informa tion evaluated by embedded thermometers at the Cricket receivers and to periodically up 30 Chapter 3 date s using an experimental relation s s T Bohn 1988 As a better alternative we implemented an optimization procedure which makes it possible to estimate measure ment by measurement the optimum s value using the
20. by conventional instruments like Coordinate Measuring Machines CMMs For in stance some large size objects can not be transferred to the measurement systems due to their dimensions or other logistical constraints Therefore it is required for the measure ment system to be moved to such components For the system operator 1GPS can poten tially be considered as a faster and easier solution compared to conventional CMMs theodolites or Laser Trackers Transmitters use laser and infrared light to determine the relative angles from the transmitters to the sensors The sensors used for measuring the workpiece have photodi odes inside their modules that can sense the transmitted laser and infrared light signals see Fig 6 2 Based on the known location information of the transmitters which is normally obtained in an initial setup phase the position of the sensors can be calculated iGPS performance evaluation and comparison with MScMS 73 The signal is transferred through a wireless network connection providing mobility to the operator Similar to a satellite based GPS a one way signal path is created from transmit ters to each sensor This approach allows an unlimited number of sensors to continuously and independently calculate positional data Y NS transmitter 1 transmitter 2 measured portable probe measured point 5 sensors transmitter 3 transmitter 4 Fig 6 2 Representation scheme of an iGPS meas
21. circuits and radio technologies have made the use of dis tributed wireless sensor networks WSNs possible for many applications Neil 2005 Recently the attention towards the utilization of systems based on distributed sensor de vices in manufacturing is increasing Since sensor devices do not need cables and may be easily deployed or moved they can be practically utilized for a variety of industrial appli cations factory logistics and warehousing environmental control and monitoring sup port for assembly processes industrial dimensional measuring and real time surveillance are only some possible applications While outdoor localization applications are wide spread today e g Global Positioning System GPS indoor applications can also benefit from location determination knowledge Gotsman and Koren 2004 To make such ap plications feasible the device costs should be low and the network should be organized without significant human involvement To give a concrete idea of the potential of the systems based on WSNs in manufactur ing here are briefly introduced some of the most interesting research issues with the cor responding bibliographic references Support for final assembly Ultrasonic sensors are mounted on power tools for example screwdrivers to detect their real position and activate them if they are in the right posi tion during final assembly Pepperl Fuchs 2005 Industrial control and monitoring S
22. conditions described in Section 3 3 and Section 6 5 Test repeatability reproducibility Men oao o amp oz standard MseMS 4 8 a5 7 8 4 1 deviation mm 0 057 0 056 0 036 0 157 0 162 0 081 Due to its optical technology iGPS metrological performance is considerably better than MScMS Considering these results iGPS is approximately 2 orders of magnitude more precise than MScMS The technology employed in particular the use of US transceiver to calculate the dis tances between the sensor devices is responsible for MScMS s low accuracy compared to iGPS Franceschini et al 2008 II Chen et al 2003 The US speed may change with the environmental conditions depending on time and position Furthermore US signals may be diffracted and reflected by obstacles interposed between two devices This is a nega tive effect for the measurement accuracy however it can be limited by the use of soft ware compensation tools Working volume size MScMS and iGPS introduce an important difference in the typologies of measurements The big difference from the traditional frame instruments like CMMs is that their struc ture is not rigidly connected but it is constituted by separate components that can be eas iGPS performance evaluation and comparison with MScMS 87 ily moved and arranged around the measuring area depending on the requirement There fore these systems are scalable or modular since the number of constellat
23. consumption Since network nodes should be tiny unobtrusive low cost and wireless they can carry only a small battery as energy supply As a result low power operation is a must and computational and communication capabilities are limited Many devices that are broadly defined as wireless because of their method of data transmission are not truly wireless in that they may require hardwiring to an AC power source In remote sensor installations however it is often impossible or expensive to connect to the power grid In situations where a self contained power supply is required design engineers have traditionally relied on two options primary lithium battery power or photovoltaic systems with rechargeable batteries Warrior 1997 Photovoltaic systems are naturally suited to sunny temperate climates but they tend to be large comparatively expensive and susceptible to breakage and they require ongoing maintenance Their use in indoor environment or inaccessible areas is therefore problem atical Lithium batteries are the preferred choice for most remote sensing applications because they have the highest specific energy and energy density energy per unit volume of all battery types Even if increases in chip capacity and processor production capabilities have reduced the energy requirement for both computing and communication regarding the future an important goal is to reduce the need for battery changeouts over long periods i e m
24. curve plotted on Fig 4 16 Ph 1 The battery charge level decreases very slowly with the battery life time The av erage TOF Error and the TOF Error standard deviation are not significantly influenced by the battery level Experimental evaluation of the MScMS ultrasound transducers 51 Ph 2 In the final part of the battery life potential difference lower than 2 3 V the dis charge is very quick and the measured potential difference falls to zero rapidly This phase is characterised by a knee in the battery charge level curve In this phase the cor responding TOF Error average value and standard deviation explode Battery Level Vs Discharge Time 2800 N O 2 oO ine O oO N DS 2 oO Q Battery Level mV 2000 0 1000 2000 3000 4000 5000 6000 7000 Discharge Time min Average TOF Vs Discharge Time 20 0 w Average TOF Error us A O O pe o O O O 8 oO oO a oO O Jia 0 0 0 1000 2000 3000 4000 5000 6000 7000 Discharge Time min TOF Standard Deviation Vs Discharge Time TOF Error St Deviation us 0 1000 2000 3000 4000 5000 6000 7000 Discharge Time min Fig 4 16 Battery level a average TOF Error b and TOF Error standard deviation c depend ing on the Cricket devices battery discharge time Each point value is c
25. depending on many different factors related to the use of US transceivers such as temperature humidity air turbulence transducers geometry transducer bandwidth US signal attenuation etc When implementing a thresh olding detection method the major effects are due to the factors related to the US signal attenuation The most important sources of attenuation are Franceschini et al 2008 II MIT C S A LL 2004 e transceivers distance e transceivers misalignment angle e transducer battery charge level With the aim of organically investigating the effect of these factors on TOF measure ments a complete experimental factorial plan is built Fig 4 5 shows a representation scheme of the experimental setup Experimental evaluation of the MScMS ultrasound transducers 39 transmitter and receiver are positioned facing each other transceivers distance are positioned at known distance 1st factor transmitter face is not perfectly aligned with receiver face A misalignment angle 0 with regard to the transmitter face is introduced 2nd factor On the other hand the re ceiver face is perpendicular to the US waves direction of propagation transmitter battery charge level is monitored measuring the battery potential difference 3rd factor Each Cricket is equipped with two AA rechargeable 2700 mAh batteries connected in series Their potential difference is measured by a standard voltmeter The potential difference is not
26. evaluated and compared through computer simulations and experimental validation tests e Development of an ad hoc software pre processor in order to guide the operator in positioning the constellation devices around the working volume according to the di mensional characteristics of the measured object Such a tool would be helpful for de termining a proper alignment of constellation devices and guaranteeing a full coverage of the measuring area e Automatic mapping of indoor environments Different kinds of measurement light acoustic noise pressure temperature acceleration magnetic field humidity etc can be associated to positional measurements It can be obtained by equipping the measur ing probe devices with additional sensor boards Then measuring operations can be automated by mounting the probe on a robotized vehicle e Enhancement and redesign and of the Cricket devices either from the hardware and the software viewpoints The most important issues are the miniaturization of the US trans Conclusions and future directions 107 ducers and the implementation of more refined US ranging methods for example ba sed on phase detection References ABERNATHY W J UTTERBACK J 1978 Patterns of Industrial Innovation Technology Review Vol 80 No 7 pp 40 47 AKCAN H KRIAKOV V BRONNIMANN H DELIS A 2006 GPSFree Node Localization in Mo bile Wireless Sensor Networks In Proceedings of MobiDE 06 Chic
27. following information e times of flight among at least 4 constellation Crickets and the 2 mobile probe Crick ets A and B e a standard of length for referability given by the a priori known distance between the mobile probe Crickets A and B By an automatic optimization we calculate the s value which better satisfies the pre vious constraints with reference to a particular portion of the working volume In this way the s value can be recalculated for each single measurement Working volume discontinuities A requirement of the measuring instruments is to measure uniformly and with no discon tinuities all the points within the working volume Due to its technology MScMS is based on a network of distributed devices communicating through RF and US While RF sensors communication range is almost omni directional and up to 25 m US sensors have a communication range limited by cones of vision with an opening angle of about 170 and a range of no more than 6 8 m see Fig 3 9 Signal strength outside the cones drops to 1 of the maximum value see the radiation pattern in Fig 4 3 Priyantha et al 2000 170 l i ceiling cone of vision of network device D cone of vision of network device D MScMS prototype 31 Fig 3 9 Representation scheme of the US sensors cones of vision It is therefore important to provide a full coverage to the area served by constellation devices by pr
28. ing measurements and has strong effects on the measurements uncertainty MScMS software provides a semi automatic procedure to achieve the constellation localization minimizing the user s effort CMMs do not need such procedure due to the different system technology 5 3 4 Metrological performances Dimensional measurement The technology employed in particular the use of US transceiver is responsible for MScMS s low accuracy compared to CMMs Franceschini et al 2008 II The use of US MScMS and CMMs a structured comparison 61 transducers can be critical for measurement accuracy because of many aspects already discussed in Section 4 3 In order to give an idea of MScMS prototype performances repeatability and repro ducibility tests have been carried out Results are reported in Tab 3 1 Also CMMs performances may change depending on many factors like machine di mensions climatic conditions or probe speed of contact Nevertheless CMMs are some order of magnitude more accurate than MScMS To provide an example of CMMs stan dard performance Tab 5 2 reports the maximum permitted error MPE on distance measurements related to a standard CMM machine DEA 2007 In general the MPE grows up with the dimension of the CMM Tab 5 2 Performance of a standard CMM DEA 2007 Standard CMM performance Stroke x mm Stroke y mm Stroke z mm MPE E for ISO 10360 2 um 500 700 500 from 1 5 L 333 Other kinds of measurem
29. of flight Ultrasound Vector bar International Vocabulary of Basic and General Terms in Metrology Wireless Sensor Network 1 Introduction The field of large scale metrology can be defined as the metrology of large machines and structures that is to say the metrology of objects in which the linear dimensions range from tens to hundreds of meters Puttock 1978 There is an increasing trend for accu rate measurement of length in particular the 3D coordinate metrology at length scales of 5m to 100m has become a routine requirement in industries such as aircraft and ship con struction In this direction there have been significant advances across a broad range of technologies including laser interferometry absolute distance metrology very high den sity CCD Charge Coupled Device cameras and so on Many types of metrological equipments utilizing different kind of technologies opti cal mechanical electromagnetic etc give physical representations of measured objects in a three dimensional Cartesian coordinate system Coordinate Measuring Machines CMMs theodolites tacheometers photogrammetry equipments GPS Global Position ing Systems Laser Trackers are typical instruments to do it Each of these systems is more or less adequate depending on measuring conditions user s experience and skill and other factors like time cost size accuracy portability etc Classical CMMs that make possible performing repeated and
30. sampling rate differences are small it is impossible to receive concurrent data from all transmitters The inevitable difference in data stream ing is in the range of a few hundredths of a second This effect does not create any prob lem for static measurements however it will affect the dynamic measurement Fig 6 7 shows such a scenario in which sensors are moving in time t For any sensor the posi tion at time period ty t is calculated by triangulating data collected in very close but for different instants Moore et al 2004 transmitter 1 transmitter 2 j Q1 0 2 05 Es moving sensor p t Dt p2 t2 p3 t3 ON N 05 0 oa O Co transmitter 3 transmitter A Fig 6 7 If a sensor moves data from transmitters are inevitably received in different instants It can be assumed for the purpose of discussion that the data collection occurs by sensing information received firstly by transmitter 1 secondly by transmitter 2 thirdly by trans 80 Chapter 6 mitter 3 and finally by transmitter 4 At time t a moving sensor s position is read when it is located in position p at time t2 when it is in position pz and so on Even if the differ ence consists of a few tens of a second it produces a location error Therefore the faster the sensor moves the larger the error becomes In this case the experiments for the system metrological performance were performed in static conditions in order to av
31. the geometry of the reference artefact used for the initial setup on measurement performance This should lead to finding an optimal geometry to minimise the error in the transmitters location Also more detailed experiments will be done in order to accurately 106 Chapter 8 characterise the system depending on different types of setup strategies and external conditions Evolution of the wireless sensor networks Finally the attention was focused on the wireless sensor networking technology from a general point of view This technology has been driven by advances in sensing and computation and has been integrated by innovations in communications MScMS and the iGPS are two clear examples of innovative system based on WSN technology Providing reliable wireless connectivity self configurability and scalability while at the same time coping with the limitations imposed by low cost energy supply miniaturization and standardization of sensor nodes the field of distributed sensor networks presents a multitude of challenging research problems For the MScMS future development two critical aspects are the miniaturization of the US transducers and the improvement of the constellation devices power efficiency Future directions Future work on this project includes e Analysis comparison and improvement of different possible techniques for the loca tion of constellation devices Three semi automatic algorithmic procedures will be
32. the working volume For each point we have calculated the standard deviations Ox oy 6 related to the registered Cartesian coordinates x y Z MScMS prototype 27 e Reproducibility test Reproducibility is defined as closeness of the agreement be tween the results of successive measurements of the same measurand carried out under changed conditions of measurement GUM 2004 VIM 2004 This test is similar to the previous one with the only difference that the mobile probe orientation is changed before each measurement with the aim of approaching each single point from a dif ferent direction see Fig 3 6 b measured single point Pee eeccescsccsee a repeatability the mobile probe position and orientation b reproducibility the mobile probe direction is are the same in the different measurements changed before every measurement Fig 3 6 Representation scheme of the practical tests carried out to evaluate MScMS performances The statistical results of these preliminary tests are reported in Tab 3 1 Tab 3 1 Results of the MScMS preliminary tests repeatability reproducibility Ox Oy Oz Ox Oy Jl 3 5 7 3 Let notice that oz value is basically lower than o and oy both for repeatability and re Mean standard deviation mm producibility tests This behaviour is due to the geometric configuration of the constella tion devices in general network devices are mounted on the ceil
33. to drive the user through measurements and online offline elaborations Today MScMS Achilles heel is represented by its low accuracy few centimetres re lated to the measured points position This is mainly due to the use of US transceivers implementation of the threshold signal detection method non punctiform dimension speed of sound dependence on temperature etc As research perspectives all factors af fecting system accuracy will be analysed and improved in detail in order to reduce their effect 4 Experimental evaluation of the MScMS ultrasound transducers 4 1 Introduction Ultrasonic US sensors are used in many application fields In general the main features of ultrasound transducers change depending on the propagation medium solids liquids air One of the most important applications of US transducers is distance measurement where the propagation medium of the acoustic signals is typically air Common applica tions associated with distance measurement are presence detection identification of ob jects measurement of the shape and the orientation of workpieces collision avoidance room surveillance liquid level and flow measurement Delpaut et al 1986 Ultrasonic ranging systems are traditionally low cost compared to other technologies like the optical laser based Unfortunately they are also characterized by low accuracy low reliability due to reflections of the transmitted signals and limited range Manth
34. 0 meters It should be easy to move and install low priced and able to work indoor inside warehouses workshops laboratories Tab 2 1 identifies the MScMS basic requirements Considering them we briefly analyse the most common measuring tools and tech niques Tab 2 2 shows the result of a qualitative comparison among five measuring in struments theodolite tacheometer CMM Laser Tracker photogrammetry system and GPS The last row of the table takes account of MScMS target performances Different considerations rise from Tab 2 2 CMMs in spite of being very accurate measuring instruments are expensive bulky and not easily movable On the other hand theodolites or GPS are smaller and lightweighter but not very flexible in terms of differ ent types of measurements offered Even more GPS systems are less accurate and cannot operate indoor Interferometrical Laser Trackers and digital photogrammetry equipments 8 Chapter 2 are extremely accurate but complex and expensive at the same time Sandwith and Pred more 2001 Points to be measured need to be identified by the use of reflective markers or projected light spots Theodolites tacheometers are typically used in topography but are not suitable to measure complex shaped objects Tab 2 1 Definition and description of MScMS basic requirements Requirement Description Portability Easy to move easy to assemble disassemble lightweight and small sized Fast Instal
35. 13 364 x x Ts 96 4 96 4 8 0 13 364 x x e 96 4 96 4 8 7 13 374 x x o 96 4 96 4 10 0 13 377 x x x X ve falallalalala eae e Ek ey PL EEL LLL Al Aer LLELL RL RRL ELLE ve ve LS e a fbb ska a Ld Fig 4 13 Results of the Minitab best regression tool After performing the linear regression the model obtained 1s TOF Error 61 7 0 113 d 2 64 0 22 8 V 0 000005 d 0 0464 07 0 000791 d 0 0 0259 d V 4 3 48 Chapter 4 According to the factorial plan analysis it can be noticed that the relationship of the re sponse TOF Error with O can be considered to be quadratic while the relationships with d and V can be considered to be linear Furthermore there are weak interactions either between d and 0 and between d and V Regression Analysis TOF error versus d 0 V d 67 d 6 d V The regression equation is TOF Error 61 7 0 113 d 2 64 22 8 V 0 000005 da 0 0464 0 0 000791 d 0 0259 qV Predictor Coef SE Coef T P Const 61 71 25 42 2 43 0 016 D 0 11333 0 01086 10 44 0 000 0 2 60429 0 1231 21 46 0 000 V 22 819 9 959 2 29 0 022 d 0 00000469 0 00000088 5 34 0 000 0 0 046412 0 001683 27 57 0 000 d 0 0 00079060 0 00002801 28 22 0 000 av 0 025872 0 003962 6 53 0 000 S 13 3608 R Sq 96 4 R Sg adj 96 4 Analysis of Variance source DE SS MS E P Regression 7 2479914 354273 1984 60 0 000 Residual Error 517 92290 17
36. 3 a PC to store and elaborate data MIT C S A LL 2004 Crickets and mobile probe exploit ultrasound US transceivers in order to evalu ate mutual distances The constellation devices act as reference points essential for the location of the probe Each US device has a communication range limited by a cone of transmission within an opening angle of about 170 and a maximum distance of no more than 8 m The mo bile probe location in the working volume is obtained by a trilateration Consequently the probe can be located only if it communicates with at least 4 constellation devices at once Akcan et al 2006 The system makes it possible to calculate the position in terms of spatial coordinates of the object points touched by the probe Acquired data are then available for differ ent types of elaboration to determine the geometric features of the measured objects dis tances curves or surfaces One of the most critical aspects in the system set up is the constellation devices posi tioning Constellation devices operate as reference points or beacons for the mobile probe In principle Crickets can be positioned without restriction all around the measured object However the number and position of constellation devices are strongly related to the dimensions and shape both of the measuring volume and the measured object It is important to assure a full coverage of the space served by constellation devices by a proper alignme
37. 6x10 m considering a plant layout e the system is setup using the mobile probe as a reference bar The 1GPS performance has been initially estimated through three tests 1 Repeatability test In this test a point within the working volume was measured re peatedly about 150 times to benefit from the high sampling rate of the instrument During these measurements the probe was left in a fixed position The test was re peated for 30 different points in different areas of the working volume For each point coordinate the residuals between the single measurements and their average value were calculated Then for each Cartesian coordinate x y z all the residuals from all the 30 points were put together The residuals show a normally distributed pattern The repeatability indicator is given by the standard deviations Ox Oy Oz re lated to each Cartesian coordinate residual see Tab 6 1 2 Reproducibility test This test was similar to the previous one with the only differ ence being that the probe was replaced before each single point measurement Hence each point was approached from a different direction using different orienta tions of the probe Reproducibility gives a preliminary indication of the system s ac curacy whereas repeatability gives a preliminary indication of the target system s accuracy This is based on compensating the most important causes of systematic er rors Tab 6 1 shows the standard deviations rela
38. 9 Total 524 2572205 Source DF Seq SS D 1 1376051 0 1 799368 V 1 13900 q 1 5096 0 1 135709 d 0 1 142179 d V 1 7613 Fig 4 14 ANOVA applied to the regression output Residual Plots for TOF error Normal Probability Plot Versus Fits 99 99 99 S 50 T a 10 1 0 01 50 25 0 25 50 0 100 200 300 Residual Fitted Value Fig 4 15 Residual plots related to the regression model response Experimental evaluation of the MScMS ultrasound transducers 49 The regression output is quantitatively examined by an ANOVA see the table on Fig 4 14 Analysing the results it can be sentenced that all the terms in Eq 4 2 are signifi cant Examining the residual plot Fig 4 15 we can notice that residuals behaviour seems to be random even if the dispersion is not constant This is consistent with the fact that the TOF Error distribution is not homoscedastic 4 4 2 Interpretation of the results Summarising we can say that TOF Errors can be influenced by three factors related to the US signal attenuation In particular we have found that e the most important factors interaction is between d and transducer distance d and misalignment angle 0 have great effect e transceivers battery charge level V has a small effect e the most important interaction is due to factors d and 0 The experimental confirmation that these three factors are sources of US signal at tenuation is given by the TOF overestimation Each
39. AISANO D MASTROGIACOMO L 2008 II Mobile Spatial co ordinate Measuring System MScMS Introduction to the system Forthcoming in Interna tional Journal of Production Research DOI 10 1080 00207540701881852 FRANCESCHINI F GALETTO M SETTINERI L 2002 On Line Diagnostic Tools for CMM Per formance The International Journal of Advanced Manufacturing Technology Vol 19 No 2 pp 125 130 GOTSMAN C AND KOREN Y 2004 Distributed Graph Layout for Sensor Networks In Proceed ings of Graph Drawing 12th International Symposium New York Vol 3383 2005 pp 273 284 GRAHAM ROWE D 2005 Compact fuel cells may oust feeble batteries New Scientist Ist Octo ber 2005 References 111 GUM 2004 Guide to the Expression of Uncertainty in Measurement International Organiza tion for Standardization Geneva Switzerland GUSTAFSSON F GUNNARSSON F 2003 Positioning using Time Difference of Arrival measure ments Proceedings of the IEEE International Conference on Acoustics Speech and Signal Processing ICASSP 2003 Hong Kong Vol 6 pp 553 556 HEDGES T H TAKAGI H PRATT T SOBEL M J 2003 Position measurement system and method using cone math calibration US Patent 6 535 282 B2 of March 18th 2003 IBM RESEARCH 2005 Wireless sensor networking Innovation Matters IEC 1131 part 3 1993 Programmable controllers Part 3 Programming languages Standard IEC 1131 3 August 1993 IEEE P1451
40. Engineers makes sensor devices independent of the protocol used on the network IEEE P1451 1 1999 Considering this point of view sensor devices should be smart and plug and play The idea is similar to that of writing a word processing program that must be able to print under an operating system such as Windows The application deals 100 Chapter 7 with the printer at a high level of abstraction and by loading the appropriate printer driver the application can print on different kinds of printers with no changes to the ap plication itself 2 Protocols standardization With the maturation of networking technology you can choose any one of the many different protocols developed so far to build a networked sensor application The choice of which one to use is not dictated so much by the techni cal features of the protocol as by other considerations such as the protocols compatibility with a particular network technology or the availability of an application or software package with that technology IBM Research 2005 There are different possible ap proaches for standardization of WSN protocols A first standardization approach is suggested by IEC international Electrotechnical Commission which attempts to provide a standard set of programming languages for WSN applications IEC 1131 part 3 1993 IEC Committee hopes that the use of such standard languages will make program code portable from one device to a
41. MILEHAM A OWEN G 2007 A Review of Two Multistation Large Volume Measuring Instrument MScMS AND iGPS Digital Enterprise Technology DET 2007 Bath UK pp 418 425 MAISANO D JAMSHIDI J FRANCESCHINI F MAROPOULOS P MASTROGIACOMO L MILEHAM A OWEN G 2008 Indoor GPS system functionality and initial performance evaluation Forthcoming in International Journal of Manufacturing Research DOI 10 1504 IJMR 2008 019214 MAHAJAN A FIGUEROA F 1999 An automatic self installation and calibration method for a 3D position sensing system using ultrasonics Robotics and Autonomous Systems No 28 pp 281 294 MANTHEY W KROEMER N MAGORI V 1991 Ultrasonic transducers and transducer arrays for applications in air Measurement Science and Technology Vol 3 No 3 pp 249 261 MARTIN J M JIMENEZ A R SECO F CALDERON L PONS J L CERES R 2002 Estimating the 3D position from time delay data of US waves experimental analysis and new processing algorithm Sensors and Actuators Vol 101 Issue 3 pp 311 321 METRIS 2007 http www metris com large_volume_ tracking positioning METRONOR 2007 http www metronor com MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LAB 2004 Cricket v2 User Manual http cricket csail mit edu v2man html MONTGOMERY D C 2008 Design and Analysis of Experiments 7th Edition Wiley New York MOORE D LEONARD J RUS D TELLER S S 2004 Robust di
42. S sssssseeeeeeeecosesesecoeseoscesecocoosooeeceeseseesesesesesessesesssese 1 T MOCO ee AEA A E EE ERN l 1 2 Typic lfeatures of sensor NetWOrkS oisirdrnnasiin ae a 1 7 3 Growth potential and future advancement ccccccseesssecceccceeeeeeeeececeeeeaaeeeeeees 1 dek Pinakcon deran ason O sac aameeieuaousaeians 1 8 Conclusions and future directions scsscscoeseccesesesocecoceccccococccccocesoeccscecesesssesessessesseee 1 ECE 5 iy 11 6 Sen a eg a aaa 1 2D 3D CCD CMM CNC CSAIL DISPEA DoE EF GPS GUM ICT IEC IEEE iGPS IP LED MIT MPE MScMS PC PVDF RF RH T TDoA TOF US V bar VIM WSN List of acronyms and abbreviations two dimensional three dimensional Charge Coupled Device Coordinate Measuring Machine Computer Numerical Control Computer Science and Artificial Intelligence Lab Dipartimento di Sistemi di Produzione ed Economia dell Azienda Design of Experiments Error Function Global Position System Guide to the expression of Uncertainty in Measurement Information and Communications Technology International Electrotechnical Commission Institute of Electrical and Electronics Engineers indoor GPS Internet Protocol Light Emitting Diode Massachusetts Institute of Technology Maximum Permitted Error Mobile Spatial coordinate Measuring System Personal Computer Polyvinylidene fluoride Radio Frequency Relative Humidity Temperature Time Difference of Arrival Time
43. Torino on a new system prototype for dimensional measurement called Mobile Spatial coordinate Measuring Sys tem MScMS Franceschini et al 2008 II MScMS determines dimensional features of large size objects and has been designed to overcome some limits shown by other wide spread measuring sets used nowadays like Coordinate Measuring Machines CMMs theodolites tacheometers photogrammetry equipments GPS based systems Laser Track ers Bosch 1995 Pozzi 2002 Basing on a distributed sensor networks structure MScMS can accomplish rapid di mensional measurements in a wide range of indoor operating environments It consists of distributed wireless devices communicating with each other through radiofrequency RF and ultrasound US transceivers This frame makes the system easy to handle and to move and gives the possibility of placing its components freely around the workpiece The wireless devices known as Crickets are developed by the Massachusetts Insti tute of Technology MIT Being quite small light and potentially cheap if mass pro duced they fit to obtain a wide range of different network configurations Priyantha et al 2000 Balakrishnan et al 2003 These features make the new system suitable for particular types of measurement which can not be carried out for example by conventional CMMs Typical is the case of large size objects which are unable to be transferred to the measuring system area be ca
44. a direct measurement of the battery charge level but since they are correlated it is an useful indicator of it Franceschini et al 2007 I Transmitter Receiver V 3 factor 0 2 factor d 1 factor Fig 4 5 Experimental setup TOF is measured changing these 3 factors at different levels Seven levels for O transmitter rotations from 0 to 60 in 10 intervals For larger an gles transmitter and receivers do not easily communicate due to the strong decrease in the US signal strength see Fig 6 Three levels for d short medium and long distance between transceivers These dis tances have been measured using 3 reference bars accurately calibrated using a stan dard Coordinate Measuring Machine accuracy lower than a hundredth of mm Furu tani and Kamahora 2001 Five levels for V from 2 3V to 2 7 V in 0 1 V intervals Tab 4 1 provides a summary of the combinations for the three factor levels There are 7 3 5 105 different combinations to be carried out According to the facto rial plans good practice measurements are randomized Montgomery 2008 For each of these combinations 50 measurements of the TOF are performed All the experiments 40 Chapter 4 have been replicated 5 times The total number of combinations analysed is 105 5 525 with 50 measurements per combination Tab 4 1 List of the experiments on the Cricket s US transducers Factors 1 Transceive
45. accurate Taylor et al 2005 Franceschini et al 2008 I Patwari et al 2005 Sottile and Spirito 2005 Mahajan and Figueroa 1999 Two techniques for the location of constellation devices were designed 1st approach First technique consists in touching using the mobile probe different reference points within measuring area It is good to select points that are easily reachable and easy to be manually located in a reference coordinate system For example points laying on objects with a simple and known geometry like parallelepiped vertexes Spatial coordinates x yi Zi of the distributed constellation devices are the unknown parameters of the problem Location of each device is performed using a trilateration To identify a new device it is necessary knowing distances from at least 4 reference points Chen et al 2003 Fig 2 9 a represents the procedure to determine distances from some reference points and a con stellation Cricket The probe tip is placed next to the point P2 with the aim of calculating the distance from Cricket B4 point D The following distances are known e AD and BD from constellation Cricket B4 and devices A and B e AB and P2B from devices A and B mounted on the mobile probe and from the de vice B and the probe tip P2 To calculate distance PD we can use Carnot Theorem see Fig 2 9 b Applying this theorem to triangle ABD we obtain the following equation 19 Principle functioni
46. ago Illinois USA ANSI ASME B89 4 19 2006 2006 Performance Evaluation of Laser Based Spherical Coordinate Measurement Systems ANSI TEEE STD 176 1987 1988 Standard on Piezoelectricity American National Standards In stitute New York ARC SECOND 2004 Product Literature http arcsecond com BALAKRISHNAN H BALIGA R CURTIS D GORACZKO M MIU A PRIYANTHA N B SMITH A STEELE K TELLER S WANG K 2003 Lessons from Developing and Deploying the Cricket Indoor Location System Preprint BERNERS A C WEBSTER J G WORRINGHAM C J STELMACH G E 1995 An ultrasonic time of flight system for hand movement measurement Physiological Measurement Vol 16 pp 203 211 BOHN D A 1988 Environmental Effects on the Speed of Sound Journal of the Audio Engineer ing Society Vol 36 No 4 Bosca J A 1995 Coordinate Measuring Machines and Systems Edited by Marcel Dekker Inc ISBN 0 8247 958 1 4 Box G E P HUNTER W G HUNTER J S 1978 Statistics for Experimenters Wiley New York CAPKUN S HAMDI M HUBAUX J P 2001 GPS free positioning in mobile ad hoc networks In Proceedings of the 34th Hawaii International Conference on System Sciences CHAI KEONG MAHONEN P UUSITALO M 2005 Standardization Efforts amp Future Research Is sues for Wireless Sensors amp Mobile Ad Hoc Networks IEICE Trans Commun Vol E88 B No 9 pp 3500 3507 CHEE YEE CHONG KUMAR S P
47. al Conference on Mechatronics Istanbul Turkey KARL H WILLIG A 2003 A short survey of wireless sensor networks Technical Report of Technical University Berlin Telecommunication Networks Group Wwww tkn tu berlin de publications papers TechReport_03_018 pdf KAZYS R MAZEIKA L SLITERIS R VOLEISIS A 2007 Online Profiling of Nonplanar Objects by High Resolution Air Coupled Ultrasonic Distance Measurements EEE Transactions on Instrumentation and Measurement Vol 56 No 5 pp 1825 1830 112 References KOUMPIS K HANNA L ANDERSSON M JOHANSSON M 2005 Wireless Industrial Control and Monitoring beyond Cable Replacement In Proceedings PROFIBUS International Conference Coombe Abbey Warwickshire UK LAMANCUSA J FIGUEROA J F 1990 Ranging errors caused by angular misalignment between ultrasonic transducer pairs Journal of Acoustic Society of America Vol 87 No 3 pp 1327 1335 LEVENE H 1960 Contributions to Probability and Statistics Stanford University Press LILIENTHAL A J DUCKETT T 2004 Building Gas Concentration Gridmaps with a Mobile Ro bot Robotics and Autonomous Systems Vol 48 No 1 pp 3 16 MAGORI V WALKER H 1987 Ultrasonic Presence Sensors with Wide Range and High Local Resolution IEEE Transactions on Ultrasonics Ferroelectrics and Frequency Control Vol 34 No 2 pp 201 211 MAISANO D JAMSHIDI J FRANCESCHINI F MAROPOULOS P MASTROGIACOMO L
48. alculated over 100 individ ual measurements As a result in order to avoid a wrong estimate of TOF it is important to replace the batteries before they reach the quick discharge phase This purpose can be automati cally succeeded by controlling the Crickets battery level through a firmware utility Shnayder et al 2004 32 Chapter 4 Test of repeatability of the US transducers The US transceivers repeatability is tested positioning three different couples of Cricket transceivers at the same known distance 3633 mm with their faces perfectly aligned For each couple of devices 100 different individual TOF Error measurements are taken Data are analysed by a one factor Analysis of Variance ANOVA to test the null hypothesis that there is no difference in the TOF Errors mean values measured by different couples of transceivers the examined factor Test of repeatability using different Cricket s US transducers 120 0 e 1st couple of transceivers a 2nd couple of transceivers __ 3rd couple of transceivers 115 0 110 0 105 0 100 0 TOF Error us ol oO Co ee ol oO 0 0 e Measurements 60 65 70 75 80 85 90 95 100 oO o1 Oo _ Sz N N o1 oO oO ao o1 gt Fig 4 17 Plot of TOF Error from different Cricket s US transducers As
49. ansmission to reception determines the distance to the target e time of flight a separate transmitter is pointed towards the receiver Instead of relying on reflections this system detects the direct transmission of the signal from transmitter to receiver After measuring the TOF the sensors distance can be calculated knowing the speed of sound value Cricket devices being equipped with either a US transmitter and a receiver implement the TOF technique A complex problem when using US transducers is the choice of the characteristic pa rameters typically resonant frequency and bandwidth For distance measurement with relatively high precision few millimetres transducers with a wide bandwidth are Experimental evaluation of the MScMS ultrasound transducers 35 needed Bandwidth is a measure of how rapidly a signal reaches the steady state A signal at the receiver obtained from transducers with a small bandwidth climbs slowly from its beginning to its peak in time domain causing a relatively large transient time at the receiver This behaviour is shown in Fig 7 Cheng Chang 2007 Tong et al 2004 A second factor affecting measurement accuracy is the transducer resonant frequency With increasing frequency and thus reducing wavelength a better resolution is achiev able Unfortunately both the transducer bandwidth and resonant frequency are directly correlated with the US attenuation and consequently they limit
50. cation software The purpose of this software is to drive the user through meas urements and to make results display efficient Functions provided are similar to those typically implemented by CMM software packages MScMS likewise CMMs makes it possible to determine the shape geometry of objects circumferences cylinders plans cones spheres etc on the basis of a set of measured surface points gathered from the mobile probe using classical optimization algorithms Overmars 1997 More in detail the software is organized into three application modules to assist the user in the following operations Initialization This is a guided procedure to switch on wireless devices Crickets and Bluetooth adapter and open the PC connection for data reception from the mobile probe Semi automatic localization of the constellation This procedure is described in Sec tion 2 4 4 MScMS prototype 25 Measurements Execution of different kinds of measurement single points meas urements distance measurements curves and surfaces evaluation see Fig 3 4 and Fig 3 5 Fig 3 3 MScMS software menu Fig 3 3 Fig 3 4 and Fig 3 5 show some displays of the MScMS software Single Points Measurements oi lt lt 0 9 0 ie pe mm k a mat x 1000 y 2000 z 1000 mm Fig 3 4 Display for the measurement of single points 26 Chapter 3 Measurements are taken like this when the probe trigger is pu
51. ced and generally require a relatively large time for installation and start up Recently the 1GPS performance was studied and tested during a three months research activity carried out at the University of Bath UK attending the project LVMA Large Volume Metrology Assembly http www bath ac uk A detailed description of this system and a comparison with MScMS is presented in Chap 6 1 4 Organization of the dissertation The remainder of this dissertation contains a detailed description of the principle func tioning and the implementation of MScMS Then the system performance is evaluated and compared with two other existing systems for large scale dimensional measurements the CMMs and the 1GPS More specifically the thesis 1s structured like this e Chap 2 presents the MScMS design features and modus operandi In particular the at tention is focused on the system principle functioning and the hardware software archi tecture e Chap 3 describes the first MScMS prototype presenting a preliminary experimental evaluation of its metrological performance Also this chapter identifies the system critical aspects and possible improvements e Chap 4 concentrates the attention on the main features of the US transceivers equip ping the system They are deeply analysed by means of a structured experimental plan e Chap 5 provides a structured comparison between MScMS and the classical CMMs e Chap 6 discusses the iGPS technologica
52. ces of errors in TOF evaluation are investigated in a structured way by means of an experimental factorial plan The results of this analysis can be useful to identify the major MScMS sources of inaccuracy and to determine how the error in TOF evaluation changes in the different points within the Cricket transmitters cones of vision see Fig 3 9 The chapter is organised in four sections Section 4 2 describes the main features of piexo electric US transceivers like those equipping MScMS Section 4 3 provides a detailed description of the factorial plan analysing the effects and the possible interactions of the sources of attenuation Section 4 4 presents and discusses the results of the factorial plan Finally the conclusions and future direction of this research are given in Section 4 5 4 2 Piezo electric US transducers In modern ultrasonic distance measurement systems for industrial applications piezo electric transducers clearly dominate Typical advantages are their compact rugged me chanical design high efficiency great range of operation temperature and relatively low cost Airborne ultrasound systems have been developed for many types of distance meas urement using two possible techniques Berners et al 1995 e pulse echo a transducer emits a burst of US which bounces off any object in the path of the beam The transducer then acts as a receiver for the reflected signal A meas urement of the time delay from tr
53. ck of this method is the average 98 Chapter 7 delay per packet Undoubtedly it would directly increase with the minimum amount of data collected by each node before transmission Whether or not this is acceptable de pends on the application b The second strategy focuses on the network paths followed during the data rout ing phase for transmitting information among nodes Since every network node can only communicate to its immediate neighbours data packets travel through the mesh of connections in a peer to peer manner see Fig 7 4 Typically the routing paradigm used refers to shortest path or minimum hop Generally energy consumption is not uniformly distributed because some network nodes e g peripheral nodes hardly communicate while others e g central nodes are congested and tend to drain their energy very quickly It should be noted that nodes which die sooner limit the lifetime of the entire network Innovative routing protocols try to provide a more uniform resource utilization shaping the traffic flow depending on battery reserve The innovative concept consists in allowing distribution of the mes sage traffic across several message channels so that traffic flows over less congested ones For example when a node detects that its energy reserve has dropped below a certain threshold it discourages other nodes from sending data to it The goal is to choose routes comprising a minimum number of nodes with su
54. cking traffic monitoring systems environmental monitoring logistics industrial diagnostics warehousing quality control and so on In scientific research there is a pressing interest around WSNs be cause they are greatly innovative with regard to obtain information from the environment investigated Patwari et al 2005 MScMS and the iGPS are two clear examples of inno vative system based on WSN technology The aim of this chapter is to analyse the development of WSNs from a general point of view trying to identify the most significant and innovative features regarding the future Discussing on opportunities improvements and development of WSNs many question arise e How WSN technology will expand e Will future network devices and applications be standardized e Will communication protocols change e How the problem of power consumption will be approached e How WSN will interface with the standard protocols e g the Internet Protocol This brief dissertation tries to identify and analyse some of the crucial aspects for the future reviewing the significant literature on the subject Section 7 2 provides a brief de scription of the general features modus operandi and requirements of WSNs Section 7 3 92 Chapter 7 provides a general discussion on potential and future development of WSN technology Then the attention focuses on two practical aspects which probably wi
55. components wireless constellation devices that should be easily moved and arranged around the measuring area depending on the exigency MScMS is scalable or modular since the number of constellation devices can be in creased depending on the measurement volume to be covered without compromising network communication and slowing down measurement activities 58 Chapter 5 On the contrary CMMs are rigid and bulky systems in which the dimensions range can reach tens of meters There is a great variety of CMMs their working volume size can go up to hundreds of cubic meters As discussed in the following sections performances and costs are strongly influenced by CMMs dimensions Phillips et al 2000 Working volume geometry MScMS may work in a non convex working volume that is to say a volume which does not contain the entire line segment joining any pair of its points e g points A and B in Fig 5 2 MScMS due to its distributed nature easily fit different types of indoor work ing environments even with inside obstacles Fig 5 2 Representation scheme of the concept of non convex working volume plant view Considering CMMs there are not discontinuities in the measuring volume since all the points within this area can be reached by the electromechanical probe Although there are CMMs with large working volumes 1 e horizontal arm and gantry CMMs the presence of obstacles in the proximity of the measured object can be
56. dition to the azimuth and elevation angles from the transmitter to the sensor more information is needed to perform a sensor position calculation which is the relative posi tion and orientation of the transmitters 76 Chapter 6 Transmitters make a constellation of reference points that are located through a system setup process The relative position and orientation of the transmitters are determined us ing an advanced algorithm which is known as bundle adjustment Hedges et al 2003 Chen et al 2003 An additional component of setup is to determine the system scale which is the absolute distance between two known points such as the length of a reference bar iGPS provides a relatively rapid and semiautomated localisation procedure requiring relatively few manual measurements Akcan et al 2006 Once the setup has been completed the measurements can be performed using a port able handheld measurement probe known as a V bar This probe is equipped with two sensors Fig 6 2 and Fig 6 5 that should be carried by an operator in order to measure the coordinates of the points touched by the probe tip To be stable and insensitive to thermal expansion the portable probe is mainly made of composite material For our ini tial experiments the V bar was used as a reference length the inter sensors distance 1s about 202 mm in the transmitters localization procedure As discussed in Section 6 4 the accuracy of transmitters location is i
57. e at the resonance fre quency producing acoustic ultrasound impulses ANSI IEEE Std 176 1987 1988 On the other hand receivers transform the vibration produced by ultrasonic waves in electric impulses A detailed characterization of these transducers is presented in Chap 4 2 3 2 Evaluation of distances between Cricket devices Crickets devices continuously communicate each other in order to evaluate mutual dis tances Devices communication range is typically 6 8 meters in absence of interposed obstacles The technique implemented by each pair of Crickets to estimate mutual distance is known as Time Difference of Arrival TDoA It is based on the comparison between the propagation time of two signals with different speed RF and US in this case Savvides et al 2001 TDoA technique is described as follows a At random time intervals included between 150 and 350 milliseconds each device transmits a RF query packet to other devices within its communication range checking Principle functioning and MScMS architecture 11 if neighbouring Crickets are ready to receive a US signal Fig 2 2 a Priyantha et al 2000 b Ready devices reply sending a RF acknowledgement authorizing next signals transmis sion Fig 2 2 b c Querying Cricket is now authorized to concurrently send a RF and US signal Fig 2 2 c transmitting device receiving device a Query RF NWN RF Antenna for RF transmission LE _ 1
58. e communication between the PC and the control system and for selecting the mobile probe type Calibration verification and system positioning e Calibration It is defined as operation establishing the relation between quantity val ues provided by measurement standards and the corresponding indications of a measur ing system carried out under specified conditions and including evaluation of meas urement uncertainty SO 1993 In general calibration defines a rule which converts the values output by the instrument s sensors to values that can be related to the appro priate standard units Importantly these calibrated values should be associated to cor responding uncertainties which reliably take into account the uncertainties of all the quantities that have an influence For MScMS calibration is an operation that can be performed every time the system is started up This in order to test system integrity and to set those parameters on which measurements depend temperature humidity etc This operation does not need a so 60 Chapter 5 phisticated instrumentation and it is carried out by measuring a standard reference arte fact with a priori known geometry Obviously this calibration procedure is not valid for CMMs because of their different technology and in particular their rigid structure CMMs calibration can not be ac complished directly by the user but requires a more complex procedure defined by in ternat
59. e constellation devices freely around the workpiece He should take care of using a proper number of constellation devices and setting their orientation in order to cover the measuring area After this a semi automatic localization procedure can be performed to locate the constellation devices This procedure consists in measuring an artefact with known geometry in different posi tions within the working volume On the other hand the set up procedure for CMMs is much more complex and requires highly skilled technicians and complex instruments like interferometric laser tracers Measuring For both the two systems the measuring phase is rather user friendly Regarding MScMS the system makes it possible to modify the measuring volume depending on the exigency e g when the workpiece is moved or replaced with a different one simply adding or moving some of the constellation devices Of course every time the position of one or more constellation devices is changed the set up phase should be performed again On the contrary CMMs are rigid systems in which the working volume size is fixed MScMS and CMMs a structured comparison 69 5 4 Final considerations This chapter compared MScMS to CMMs the most commonly used devices for objects dimensional measurements MScMS and classical CMMs are similar considering measurement activities however due to their different technological features they have many differences for example
60. e meas uring machine Semi automatic proce dure for the probe quali fication idem idem idem Warning signals idem The software structure is modular see the representation scheme in Fig 5 4 Each module is associated to a specific activity System start up dimensional measurements results displaying Modules are linked together by different operational paths Each path represents a sequence of screenshots The great advantage of a modular structure is that it can be progressively extended according to the measuring system en hancement Fig 5 5 to 5 7 show some screenshots of MScMS user interface MScMS and CMMs a structured comparison 65 1 Introduction page 2 System start Lip 3 Main page 4 Probe 5 System shut qualification down Fe ER E T Error 8 Wireless 9 Network 10 Measuring 11 System J aan i j 12 single 13 Distances 14 Intersections 15 Angles sua j shapes Fig 5 4 MScMS software architecture Mobile Spatial coordinate Measuring System i ons Calibration Wireteas comm Fig 5 5 The MScMS s main menu screenshot 66 Chapter 5 Fig 5 7 a Choice of the measuring activities b Single shape measurement 5 3 7 Flexibility Kinds of measurement Considering flexibility as the ability of performing different types of measurement MScMS is more flexible than classical CMMs As described above MScMS offer
61. ed longer reference bars normally generate better results in the above mentioned setup process iGPS performance evaluation and comparison with MScMS 81 Zakrzewski 2003 This can be intuitively explained by the simplified representation scheme shown in Fig 6 8 Let us suppose that transmitter T x7 yr is located in a plane 2D The two angles and Q2 from the sensors at the extremities of the reference bar and the bar length are known assuming the uncertainty is null Since there is an un avoidable uncertainty related to the bar length information it will produce uncertainty in the location of the transmitter T Considering the short bar A in Fig 6 8 the uncertainty on the position of T is given by segment S Using a longer bar B the location uncer tainty decreases see segment Sg in Fig 6 8 This example shows that the longer the ref erence bar the lower the uncertainty on the T location However the use of too long reference bars is not practical and may produce other er rors which may inversely influence transmitters location accuracy e g flexing or ther mal expansion of the bar error related to the angles uncertainty Environmental factors iGPS like most measuring instruments is sensitive to several environmental factors in particular temperature light and vibrations It is well known that laser signals are sensi tive to changes in air conditions especially in terms of temperature which can exhib
62. ed and the nominal position Cartesian coor dinates are calculated Then the standard deviations related to the coordinates Ox Oy O are calculated The residuals have been verified to be normally distributed Based on these results the 1GPS uncertainty referring to a 20 interval can be roughly estimated to be less than 1 mm portable probe Fig 6 9 National Physics Laboratory artefact Cross el al 1998 used for iGPS experiments Results of these preliminary tests are reported in Tab 6 1 Considering the different testing conditions these results are reasonably consistent with the results of some tests carried out by 1GPS constructors ARC Second 2004 In general z value is lower than ox and oy for repeatability reproducibility and accuracy 84 Chapter 6 tests This is due to the geometric configuration of the constellation devices as transmit ters are mounted on tripods which are set more or less at the same height Therefore they can be considered as approximately placed on a horizontal plane XY perpendicular to the vertical Z axis Patwari et al 2005 Tab 6 1 Results of the iGPS preliminary tests performed in the specific testing conditions de scribed in Section 6 5 accuracy w a e e da s a d 6 6 Systems comparison Mean st deviation mm 0 211 In this section MScMS and iGPS are compared The summary of the comparison and the results is given in Tab 6 2 Tab 6 2
63. ed by the US transducers equipping MScMS an exploratory analysis of the most important sources of error has been performed The result is that the most important effects are due to the US signal attenuation which is directly caused by the implementation of the thresholding signal detection method Figueroa and Lamancusa 1992 The three major sources of attenuation are 1 transceivers distance 2 transceivers misalignment angle 3 transducer battery charge level These factors have been analysed through an organic experimental factorial plan According to the results transceivers distance and misalignment angle are the most significant Unfortunately the actual metrological performance of MScMS is strongly limited by the measuring errors derived by the use of US transceivers Regarding the future the system s accuracy could be improved implementing more refined US ranging methods for example based on phase detection with fixed frequency signals and with frequency modulated signals Manthey et al 1991 Tong et al 2001 The main drawback is that these detection methods are more expensive because they require complex hardware software Another possible solution to the error derived by the transmitter misalignment is the use of omni directional ultrasonic transducer like the cylindrical polyvinylidene fluoride PVDF film transducers Toda 2002 But here again Cricket devices should be partially redesigned either from the hardware and
64. ensor devices can be deployed to perform industrial control and monitoring for instance control of the air conditions of pollution tempera ture and pressure in different areas of the factory or for emergency responses in case of incidents Doss and Chandra 2005 Pan et al 2006 Koumpis et al 2005 Oh et al 2006 Factory logistics and warehousing In an industrial warehouse mobile forklifts generally move along corridors in order to reach the shelves where goods are stored Forklifts and shelves can be equipped with ultrasound transceivers that communicate with each other with the purpose of evaluating mutual distances Intel Corporation 2005 This type of wireless sensor network can be utilized to calculate the position of the forklifts for e Indoor Navigation Mobile forklifts equipped with wireless transceiver are automati cally guided towards their destination Wang and Xi 2006 Introduction 5 e Traffic Monitoring The physical traffic can be monitored in order to identify the most congested areas or to improve goods distribution Capkun et al 2001 Large scale dimensional measuring Besides the MScMS two innovative measuring sys tems for large scale dimensional measurements are the 3rd Tech Hi Ball and Metris iGPS Welch et al 2001 Rooks 2004 Metris 2007 These systems all based on optical technologies and recently industrialised are lightweight and very accurate but they are relatively high pri
65. ent probe technologies CMMs consist of three ba sic components see Fig 5 1 e the machine body three carriages move the probe along the X Y and Z Cartesian co ordinate axes e a measuring probe to touch the surface points of a workpiece e a control and computing system to calculate the Cartesian coordinates of the points and evaluate the shape features of the workpiece s surface 56 Chapter 5 machine body iji yori em ee control and a ss tl computing system ie doe measuring probe measured object Dr Fig 5 1 A typical Coordinate Measuring Machine CMM DEA 2007 CMMs are widely used in many industrial sectors to perform product control The rea son why they are so widespread is their reliability and accuracy Curtis and Farago 1994 CMMs software makes it possible to perform complex types of measurement sur face construction intersections projections In spite of their diffusion these machines can not measure every kind of object With a few exceptions gantry or horizontal harm CMMs which are expensive and not portable CMMs can not measure large size ob jects due to their limited measuring volume 5 3 Comparison criteria The MScMS prototype has been designed to be portable with the aim of measuring large size objects and minimizing manual activities MScMS and CMMs will be compared ac cording to the set of criteria requirements listed in Tab 5 1 In the following subsections t
66. ents While CMMs have been designed with the purpose of performing only dimensional measurement MScMS can carry out other kinds of measurement More precisely Cricket devices may be equipped with additional sensor boards This gives the possibility to MScMS associate single position measurements with other kinds of measurement such as light intensity temperature acceleration magnetic field pressure humidity or noise pollution Accuracy of these kinds of measurement depends on embedded sensors utilized Crossbow Technology 2008 5 3 5 Measurements system diagnostics On line measurements diagnostics As said before MScMS is sensible to external factors such as environmental conditions temperature humidity presence of obstacles among distributed devices MScMS soft ware provides some diagnostic tools to control the measurements activities and assist in the detection of abnormal functioning Firstly it gives the opportunity of watching the data exchanged among the wireless devices making it possible to discover abnormal functioning of the system components Secondly it allows a graphic display of the probe s range of vision that is to say the set of constellation devices it can communicate 62 Chapter 5 with see Fig 5 3 This helps the operator to check whether the probe is in the optimal position to perform a specific measurement i e if it communicates with at least 4 con stellation devices Furthermore we implemented
67. er in order to manage sensor hardware and software functionality reli able and user friendly standard operating systems should be designed and developed 7 3 Growth potential and future advancement Modern research on sensor networks started around 1980 for military purposes The de velopment of technology has been driven by advances in sensing computation commu nications and more in general by the great expansion of ICT Information and Com munications Technology see Fig 7 2 Current WSNs can exploit technologies not available 25 years ago and perform functions that were not even dreamed of at that time Sensors processors and communication devices are all getting much smaller and cheaper and WSN technology has enormous potential in terms of delivering new benefits to society Intel 2006 WSNs can be used in many fields ranging from environmental monitoring to industrial sensing as well as traditional military applications At the pre sent time several companies and manufacturers are studying the potential of WSNs in order to differentiate conventional products and to be disruptive to competitors Neil 94 Chapter 7 2005 For example predictive maintenance service of conventional industrial equip ments is a good opportunity to do just that and it would be quite easily implemented with the emerging sensor based wireless networking technology which simplify the way manufacturers gather information In this and o
68. erentiate the transmitters Additionally the transmitter speed is con tinuously tracked and used to convert the timing intervals into angles As shown in Fig 6 4 the two fanned beams radiated from the rotating head of each transmitter are tilted with respect to the rotation axis the vertical axis of the transmitter nominally at 30 and 30 This angular method is used to calculate the elevation angle by iGPS performance evaluation and comparison with MScMS 75 e knowing the angles of the fanned beams b with respect to vertical as shown in Fig 6 4 e determining the difference in timing between the arrival of laser 1 and laser 2 to the Sensor e knowing the speed of rotation of the transmitter which is continually tracked laser 2 TnS e laser 1 30 30 r LED strobe laser 1 laser 2 Aol os timing measurement to timing measurement calculate the azimuth angle At to calculate 0 Fig 6 4 Representation scheme of the transmitter s fanned beams Metris 2007 The measurement of azimuth angle requires a horizontal index which is created by firing an omnidirectional LED strobe at a fixed direction in the rotation of the transmit ter s head Referencing the timing diagram at the bottom of Fig 6 4 the azimuth angle is determined by e knowing the angles of the beams e making a timing measurement between the strobe and the laser pulses e knowing the speed of rotation of the transmitter In ad
69. estimates its position by performing the iterative minimization of an Error Function EF defined as Principle functioning and MScMS architecture 15 ee M G f n EF 2 4 being M measured distance between the 1 th node and the unknown device P G calculated distance between the estimated position of P u v w and the known position of the i th device C Xj yi Zi n number of constellation devices C 1 l n within device P communication range C2 X2 y25 Z2 L Cs C4 Ms f Xb Y6 Z6 Cs X4 Y4 Z4 g Xs Y85 Zs i Ma 7 7 4 7 7 7 s ees ra ae Cs r D 7 ae a M Xs Y5 Zs C X7 Y75 Z7 P Z u v w ene Se M3 s X3 Y3 Z3 N bS Ea e C X1 yis Z1 P communication range Fig 2 7 Location of a generic device P Each of the two Cricket mounted on the mobile probe locates its own position using the known locations of at least four constellation Crickets and the measured distance from them All information needed for the location is sent to a PC for a centralized com puting 2 4 2 Location of points touched by the probe tip The probe tip V lies on the same line of devices A and B see Fig 2 5 This line can be univocally determined knowing coordinates of points A Xa ya Za and B Xg Yp Zp and their distance d A V The pa
70. et al 2000 The consequence of the use of amplitude threshold detection is the occurrence of errors in TOF evaluation The implementation of the threshold detection method at the receivers 1s a source of inaccuracy As represented in Fig 3 10 and Fig 4 4 the signal transient time at the receiver strongly influences the ranging precision This may cause relatively large errors in the TOF evaluation one ore more US time periods Actually since the speed of sound is about 340 m s one US time period corresponds to a distance of about 8 5 mm Considering that the threshold can be exceeded even 4 period late distance overestimation can be up to 3 4 cm 32 Chapter 3 signal amplitude 5 _ full amplitude signal ETO A fi fi fi fi fm with decreased amplitude amplitude threshold set at the receiver time from the arrival of the first US signal at the receiver T 25 us Fig 3 10 Representation scheme of the error produced by the use of amplitude threshold detection method The signal transient time at the receiver strongly influences the ranging precision 3 4 Final considerations MScMS first prototype is adaptable to different working environments and does not re quire long installation or start up times Before performing measurements constellation devices freely distributed around the measuring area automatically locate themselves in few minutes System is supported by an ad hoc software created in Matlab
71. etworking re search trying to identify and analyse the potential and the crucial aspects for future Some experts opinion is that sensor networks could potentially become a disruptive technology when the miniaturization power consumption standardization and cost prob lems are solved WSNs are still a young research field much activity is still on going to solve many open issues before their will be ready for an important practical deployment 8 Conclusions and future directions This final chapter describes the primary contributions of this thesis and the possible future research developments MScMS Prototype A preliminary prototype of MScMS was built and tested with the purpose of verifying system feasibility and to evaluate its performances The system is adaptable to different working environments and does not require long installation or start up times Before performing measurements constellation devices freely distributed around the measuring area locate themselves by means of a semi automatic procedure System is supported by an ad hoc software created in Matlab to drive user through measurements and online offline elaborations Actually measurements consist in e touching the desired points on the measurand surface by using a mobile probe e pulling the probe trigger for performing the measurement and sending the information via Bluetooth to the PC e calculating the Cartesian coordinates x y z o
72. expected TOF Error does not significantly change depending on the different Cricket devices used Fig 4 17 shows the plot of the TOF Error measured 100 times in the same conditions using 3 different couples of Cricket transceivers One way ANOVA 1st couple 2nd couple 3rd couple of Cricket transceivers Source DEF SS MS F P Factor 2 26 6 13 3 0 33 0 720 Error 297 12019 9 40 5 Total 299 12046 6 Individual 95 CIs For Mean Based on Pooled StDev Level N Mean olDey E E EE TRER eel eee RE Sa a cei Lest couple 100 100 72 ba 26 Cean enra anie E net oe alan a and couple 100 100 04 6 Se ect eye eee ey e ET 3rd couple 100 100 61 6 67 Gaeta en ee Se overlapping region gt v pHini pHini IIAU 100 200 100 80 T0160 Pooled StDev 6 36 Fig 4 18 Results of an ANOVA to test the Cricket s US transducers repeatability Experimental evaluation of the MScMS ultrasound transducers 53 As shown measurements obtained using different devices generally overlap Conse quently it can be said that the use of different US transducers devices does not influence the TOF Error This qualitative impression is confirmed by the results of the ANOVA in Fig 4 18 4 5 Final notes and future work The chapter analysed the most important sources of error related to the TOF measurements performed by the US transducers which MScMS is equipped with Measurement error may change depending on many different factors however
73. ey et al 1991 US sensors provide high accuracy only in certain working contexts Excellent performances can be achieved when measuring for example short fixed distances and controlling envi ronmental conditions temperature and humidity The most common technique for dis tance evaluation is by measuring the time of flight TOF of the US signal either from a transmitter to a receiver or using a single transceiver which transmits the US signal and receives the corresponding reflected signal Other aspects influencing the performance of ultrasonic sensors are the type of transducers and the signal detection method used i e thresholding envelope peak phase detection discussed in Section 4 2 For this reason different types of transducers can be employed depending on the specific application Most of commercially available air ultrasonic transducers are ceramic based and operate at 40 kHz Transducers that operate at higher frequencies such as at 200 kHz are more limited and more expensive Toda Dahl 2006 This chapter focuses on the US transducers used by the Mobile Spatial coordinate Measuring System MScMS The characterization of the MScMS US transceiver is performed by means of several experiments organically designed through a factorial 34 Chapter 4 plan and performed in different measuring conditions Particular emphasis is given to the effect of the US signal attenuation on the TOF estimation Also the major sour
74. f the points by specific algorithms and eventually identifying the geometrical features of the measurand surface The prototype actual performance was estimated by two practical tests repeatability and reproducibility Regarding the repeatability test the average standard deviations 6 Oy O2 related to the point Cartesian coordinates are around 5 mm Regarding the repro ducibility test they are around 7 mm This low metrological performance is the actual Achilles heel of MScMS This is mainly due to the use of ultrasound transceivers non punctiform dimension speed of sound dependence on environmental factors use of the threshold detection method for detecting the US signal etc As research perspectives all factors affecting system accuracy should be analysed in detail with the aim of compen sating them or reducing their effect Furthermore even if this topic is not fully discussed in this thesis MScMS offers the possibility of simultaneously performing different kinds of measurement light acous 104 Chapter 8 tic noise pressure temperature acceleration magnetic field and humidity gas concentra tion associating them to the position measurement These kinds of measurement which can not be achieved with a classical CMM are useful for the mapping of indoor environ ments Fischer et al 2001 Lilienthal and Duckett 2004 Safigianni et al 2005 MScMS ultrasound transducers With regard to the TOF measur
75. f transmitters and sensors can be increased depending on the measurement environment Such characteristics how ever do not compromise the network communication or slow down the setup activities and measurements ARC Second 2004 Before starting measurements the location of transmitters has to be determined This phase should be fast and automated as much as possible to prevent any conflict with the system adaptability to different working environments During measurements for each sensor the position x y z 1s calculated Each trans mitter presents two measurement values to each sensor the horizontal azimuth and the vertical elevation 8 angles see Fig 6 3 Sensors can calculate their position when ever they locate in the line of sight of two or more transmitters The principle used is tri angulation Niculescu and Nath 2003 transmitter i rotating head 0 elevation we ee ee ee e i O S e OS eee cece bee weer Ce ecccccesecccceere Fig 6 3 Azimuth and elevation 0 angles from a transmitter to a sensor Here follows a description of how sensors measure angles from the transmitters Each transmitter generates two rotating infrared laser beams and an infrared LED strobe These optical signals are converted into timing pulses through the use of a photo detector The rotation speed of the spinning head in each transmitter is deliberately set to a different value in order to diff
76. fficient remaining power Q final receiver patt AI g r a S s 1 gt F Pa Sob o Q o Nodes sensor devices J n se 1 hS 4 re a ey NS Edges connections between 2 nodes ay aa n iY DNE F a Q enans x Communication Hops SE a i y E SS 7 in Sa 2 L h r original sender Fig 7 4 Schematic representation of data transmission among network nodes Standardization Most of the research in WSN field has been more technology push driven than trying to specify clear requirements from applications Today there are different networking de Future wireless sensor networks 99 vices and protocols supported by research bodies which attempt to define standard com binations of technology and functionality in order to allow devices to interoperate with each other WSN devices are often chosen to fulfil some specific projects or partnership requirements In this scenario there is no strong standardization organization that is spe cifically aiming to harmonize interfaces and interoperability functionalities for WSN sys tems Romer et al 2002 The industrial exploitation and research challenge is to find out enough commonalities to build more generic platforms architectures and standards providing common ground for the future instead of collecting different approaches Although unlimited technology push is often required to develop disruptive technologies it is also i
77. functioning and MScMS architecture 9 1 a constellation of wireless devices distributed around the measuring area 2 a mobile probe to register the coordinates of the object touched points 3 a PC to store data sent via Bluetooth by the mobile probe and an ad hoc application software The mobile probe is equipped with two wireless devices identical to those making up the constellation These devices known as Crickets are developed by Massachusetts In stitute of Technology and Crossbow Technology They utilize two US transceivers in or der to communicate and evaluate mutual distances MIT C S A LL 2004 Crossbow Technology 2008 The system makes it possible to calculate the position in terms of spatial coordinates of the object points touched by the probe More precisely when a trigger mounted on the mobile probe is pulled the current coordinates of the probe tip are calculated and sent to a PC via Bluetooth Acquired data are then available for different types of elaboration determination of distances curves or surfaces of measured objects Constellation devices Crickets operate as reference points or beacons for the mobile probe The spatial location of the constellation devices follows a semi automatic proce dure described in Subsection 2 4 4 Constellation devices are distributed without con straint around the object to measure In the following subsections we describe the MScMS hardware
78. he previous criteria are individually analysed in order to perform specific comparisons between MScMS and classical CMMs MScMS and CMMs a structured comparison 57 Tab 5 1 Comparison criteria 5 3 1 Portability 5 3 2 Working volume Size Geometry 5 3 3 Set up Installation Start up Calibration verification and system positioning 5 3 4 Metrological performances Dimensional measurement Other kinds of measurements 5 3 5 Measurements system diagnostics On line Off line 5 3 6 Ease of use Automation Software user interface 5 3 7 Flexibility Kind of measurement Geometric relation Concurrent measurements 5 3 8 Cost Purchasing Maintenance Set up phase Measuring phase 5 3 9 System management 5 3 1 Portability MScMS is composed by distributed and lightweight wireless devices which are easily portable and installable in the area around the measured object They can be fixed to the ceiling or mounted on standard supports and tripods see Fig 3 1 While the MScMS components can be moved to different operating environments tra ditional CMMs are embedded in a precise working area Once installed CMMs have to be permanently used there To be moved they need to be disassembled re assembled re installed and re started up spending a lot of time and with much effort 5 3 2 Working volume Working volume size The big difference from traditional CMMs is that MScMS structure is not rigidly con nected It is made of separate
79. icket firmware Each Cricket device performs two types of operations a time of flight measurement of US signals transmitted received from other devices At random time intervals included between 150 and 350 milliseconds each device tries to synchronize itself with neighbours in order to exchange US signals Synchronization information is transmitted through RF packets b when a Cricket receives a new distance from a neighbour or directly measured stores and sends it to its neighbours by a RF packet containing a new list of inter node distances 18 Chapter 2 Firmware coordinates the communication among Cricket devices making them able to cooperate and share information about inter node distances When the user pulls the mo bile probe trigger all information is sent via Bluetooth to a PC for elaborations 2 4 4 Semi automatic location of the constellation Location of Cricket devices should be fast and automated as much as possible This op eration if manually performed is tedious and conflicting with system adaptability to different working places As a consequence in order to minimize human moderation a method for a semi automatic localization has been implemented It is important to remark that accuracy in the localization of constellation nodes is fundamental for accuracy in the next mobile probe location The more Crickets position are affected by uncertainty the less the following measurements will be
80. ing or at the top of the measuring area for this reason they can be considered as approximately placed on a plane XY perpendicular to the vertical Z axis see Fig 3 1 Since we have experimentally verified that the distribution of the point coordinates can be considered to be normal both for repeatability and reproducibility data the variability range considering a 99 73 confidence level is given by 30 Montgomery 2008 Reproducibility range is an index of the instrument actual accuracy whereas repeat ability variation range is an index of the target instrument accuracy supposing to com pensate the most important causes of systematic errors 28 Chapter 3 The most critical aspects of the whole measuring system are due to US sensors In par ticular 1 Dimensions of US transceivers 2 Different types of noise affecting US signals 3 Speed of sound dependence on environmental conditions 4 Working volume discontinuities 5 Use of amplitude threshold detection at receivers These aspects are individually discussed in the following subsections Dimensions of US transceivers A source of uncertainty in US time of flight measurements is due to non punctiform US sensors The volume of each piezo electric crystal is about 1 cm As shown in Fig 3 7 it is difficult to determine the exact point of departure arrival of a US signal exchanged be tween a pair of Crickets These points are placed on the US sensors surface
81. ing the case of interest we obtained A 0 52 It was demon strated through the Levene s test that the transformed response variance is now stabilized Of course a problem is that it may be uncomfortable working with the transformed re sponse y in the transformed scale since it can result in a nonsensical value over the factor space of interest To construct a model in terms of the original response the oppo 1 site change of variable y is performed To have a first idea of the single examined factors effect on the TOF Error we use the Main Effects Plot see Fig 4 7 The points in the plot are the means of the response vari able at the various levels of each factor for each level of the examined factor the mean is calculated averaging all the responses obtained changing the remaining two factors A reference line is drawn at the grand mean of the response data This kind of plot is useful for comparing magnitudes of main effects Main Effects Plot for Corrected TOF Error 0 degrees corrected TOF Error 1160 2034 3671 0 10 20 30 40 50 60 Mean 16 14 12 10 corrected TOF Error 2 3 2 4 2 5 2 6 27 Fig 4 7 Main effect plot for means related to the three examined factors O misalignment angle d transceivers distance V batteries potential difference Experimental evaluation of the MScMS ultrasound transducers 43 The qualitative result is that misalignment angle and transmitter
82. ion devices can be increased depending on the desired measurement environment Such characteris tics however do not compromise the network communication and do not slow down the activities such as constellation location and measurements System diagnostics MScMS is sensible to external factors such as the environmental conditions of the meas uring area temperature humidity presence of obstacles among distributed devices As mentioned above Subsection 5 3 5 MScMS software provides some diagnostic tools to control the activities and assist in the detection of abnormal functioning Wrong distance measurements like the ones due to US reflection diffraction or other measuring acci dents among Cricket devices can be indirectly detected and rejected To this purpose we have provided an effective diagnostic test able to discriminate with a little uncertainty good from wrong distance measurements This test is based on the analysis of the residu als related to the error function EF optimized during the trilateration process see the description in Section 3 3 Franceschini et al 2007 II Moore et al 2004 To filter bad measurements due to external factors as light temperature or vibrations iGPS also provides other types of diagnostic controls The reliability of measurements dramatically increases by using multiple fixed sensors which are placed at a priori known positions With these sensors the system can perform an automatic ini
83. ional standards ISO 10360 2001 In particular CMMs calibration consists in a sequence of manual activities that must be carried out once or twice a year and re quires highly qualified operators and complex instruments like laser interferometers e Verification It is defined as confirmation through examination of a given item and provision of objective evidence that it fulfils specified requirements ISO 1993 An other activity to make MScMS suitable for the measurement is the system verification It should be periodically performed to verify and adjust the measuring scale adopted for example the ultrasound speed changes with air temperature and humidity This operation is performed by the use of a standard reference artefact ISO 10360 2001 CMMs verification is done using some standard reference artefacts or repeatedly measuring the same points to evaluate eventual measurements drifts Different ap proaches have been proposed in this direction Franceschini and Galetto 2007 When ever a CMM does not fulfil specified requirements highly qualified operators have to intervene e System positioning It is defined as operation establishing the initial position of the constellation devices Every time MScMS is installed a crucial activity is the constel lation devices positioning In order to locate the mobile probe MScMS has to know the position of constellation devices This step needs to be completed before perform
84. ions due to the different factors and their interac tions Results of an ANOVA can be considered reliable as long as the following assump tions are met 1 response variable is normally distributed 2 data are independent 3 AA Chapter 4 variances of populations are equal After applying the Box Cox response transformation all these assumptions are satisfied Analysing the ANOVA results Fisher s test it can be sentenced that all three factors are significant and their interactions as well With regard to the effect of the single fac tors the most important are d and 0 while the effect of V is minor small F value This is consistent with the Main Effect Plot of Fig 4 7 With regard to the factors interactions they are all statistically significant small p values but very weak The strongest is the one between d and 0 General Linear Model corrected TOF error versus d 0 V Factor Type Levels Values d fixed 3 1160 2034 3671 0 fixed 7 0 10 20 30 40 50 60 V fixed 5 gee 24y 2s Bobs Ze Analysis of Variance for corrected TOF error Source DF Seq SS Adj SS Adj MS P d 2 4071 35 4071 35 2035 67 0 000 single factors effect o 6 2368 87 2368 87 394 81 0 000 V 4 44 22 44 22 11 06 0 000 Hiasi Son 2 w LT 221 50 1G eke 0 000 V 8 18 09 18 09 2 26 0 000 couples of factors O V 24 30 15 30 15 1 26 0 000 Error 468 46 36 46 36 0 10 524 6700 59 Total Fig 4 9 ANOVA applied to the transformed response of
85. it both temporal and spatial variations within large working volumes Light typically has a go no go effect that is to say if sensors are exposed to light the laser beams can be obscured and consequently measurements cannot be performed at all To avoid this problem for the experiments in this chapter the lights in the laboratory are kept at mini mum especially in the area near to the sensors and transmitters Vibrations are another source of error that can produce little movements of the measured workpiece or the measuring equipment This effect can be large and it should be considered when analys ing the results To filter bad points from the measurement due to external factors such as light tem perature or vibrations the iGPS software provides several diagnostic controls The reli ability of measurements increases significantly by using auxiliary sensors which are placed in fixed positions at a priori known distances With these sensors the system can correct the initial setup in real time by compensating the changes in the environmental conditions of the measuring field and determining whether the system is conforming to the desired tolerance Kang and Tesar 2004 82 Chapter 6 6 5 Experimental work for iGPS preliminary performance analysis Explorative tests are performed to evaluate the iGPS metrological performance in the fol lowing conditions e use of 4 transmitters e measuring area of about 60 m
86. item generates a reduction in the US signal amplitude due to the implementation of the thresholding signal detection method E a ah E et eas BN a ey A ae ce I envelope of a full nal 7 amplitude received signal me s envelope of a half Htt amplitude received signal detection threshold Receiver electric output signal mV D ZO 40 BO 0 100 120 140 160 1860 200 220 240 260 2580 300 Time from the arrival of the first US signal at the receiver us Fig 4 19 Considering the same variability AV in the receiver voltage signal the corresponding uncertainty in the time of flight changes The more attenuated the signal the larger the time of flight variability Another interesting result is that the standard deviation related to TOF Error is depend ent on the US signal attenuation This behaviour is a consequence of the thresholding de 50 Chapter 4 tection method Since each transmitter is characterized by a proper natural variability due to power and control supply air conditions and so on the envelope of the US signal at the receiver will be included within an uncertainty bandwidth in grey in Fig 4 19 Con sidering signals with different amplitudes and assuming the uncertainty bandwidth to be the same the larger the transient slope the lower the TOF uncertainty U1 and U2 in Fig 4 19 Obviously this behaviour is directly caused by the use of the thresholding detection method and it is a
87. l features and principle and provides a com parison with the MScMS e Chap 7 presents a short general analysis of the development of WSNs This can be in teresting considering that MScMS and other innovative measuring systems are based on distributed WSNSs 6 Chapter 1 e Finally Chap 8 summarizes the thesis contributions and mentions possible future di rections for improving the MScMS performance 2 Principle functioning and MScMS architecture 2 1 Introduction The purpose of this chapter is to describe the MScMS hardware software firmware archi tecture and functionalities Before introducing MScMS in Section 2 2 we provide a structured description of re quirements and functionalities that a generic system for large scale dimensional meas urements should meet At the same time we present a taxonomy of the most common techniques and metrological equipments for dimensional measuring Major advantages and drawbacks are highlighted The attention is subsequently focused on the MScMS de sign analysing in detail the following aspects hardware and software configuration dis cussion of the location algorithms implemented by MScMS description of the semi automatic procedure for the spatial location of the MScMS constellation devices 2 2 System requirements and comparison with other measuring techniques MScMS has been designed to perform dimensional measurements of medium large size objects with dimensions up to 30 6
88. lation Before being ready to work system installation start up or calibration and Start Up should be fast and easy to perform Low Price Low costs of production installation and maintenance Metrological Appropriate metrological performances in terms of stability repeatability Performances reproducibility and accuracy ISO 5725 1986 Working Vol The area covered by the instrument should be wide enough to perform ume measurements of large size objects dimensions up to 30 60 meters System should be user friendly An intuitive software interface should guide Ey he the user through measurements Work Indoor System should be able to work indoor inside warehouses workshops or laboratories Flexibility System should be able to perform different measurement typologies i e de termination of point coordinates distances curves surfaces etc Tab 2 2 Measuring systems comparison qualitative performance evaluation REQUIREMENTS ____BEQUIREMENTS o en AANG Installation Metrological Working e sare rein Sag om paea Wt eee a in THEODOLITE LARGE MEDI cM a MScMS Purpose HIGH LOW MEDIUM In conclusion none of the previous measuring systems fulfil all previous requirements MScMS is a system based on the WSN technology able to make a trade off among these requirements 2 3 MScMS hardware equipment MScMS is made up of three basic parts Franceschini et al 2008 IH Principle
89. lest and the most widely used and applies to any type of short duration signal By this method implemented by Cricket devices the receiver electric output signal is compared with a threshold level 65 mV for Cricket devices such that arrival of the wave is acknowledged when the signal reaches this level This method de pends on the amplitude of the pulse received the larger the signal amplitude the smaller the time taken by the signal before reaching the threshold Considering the example in Fig 4 4 when the signal has a full amplitude the detection threshold is first exceeded by the second peak of the US waveform When the waveform is attenuated by a factor of 0 5 half amplitude signal the detection threshold is first exceeded by the third peak of the US waveform If the channel attenuation is quite significant it may cause the threshold to be exceeded a few periods late instead of just one period late Considering that for a 40kHz US a period is 25 us this error will approximately be in integer multiples of 25us Since the speed of sound is around 340 m s a one period error corresponds to a 25 340 1000 8 5 mm distance overestimation In practice since the threshold can be ex ceeded even 4 period late distance overestimation can be up to 3 4 cm Envelope peak detection It is a modification of thresholding which may be called adjust able thresholding This method acknowledges arrival of the signal when a maximum am plitude is de
90. ll be critical for future advancements e power consumption e standardization Final considerations are reported in Section 7 4 7 2 Typical features of sensor networks In order to prepare the field for the dissertation about WSNs future sceneries this section summarizes their basic features requirements and modus operandi Sensor networks typically consist of a large number of nodes densely distributed Each sensor node communicates with other nodes within its communication range Silently and wirelessly each sensor collects data for instance position estimates monitoring of light temperature or other environmental factors The collected data are relayed to its neighbouring devices and then to a specified destination where they are processed A wireless network is typically modelled as a graph where each node represents a physical device Two nodes are linked by an edge if and only if they can directly com municate or rather if they are connected see Fig 7 1 Sensory data when gathered from all the devices and analyzed by more traditional computers paint a comprehensive high resolution picture of the surroundings in real time a ae e bee Lg ae OON i 7 o 7 ae Nodes sensor devices bos ob fo sx LL o Edges connections between 2 nodes Mie I L k ra i pn r ie 7 w eee ee o KA E toh ci Nodes communication ranges Fig 7 1 Schematic representation of a wireless
91. lled the application software calculates Cartesian coordinates of the point touched by the probe tip If meas urement is correctly taken an acoustic signal is emitted Measure results are displayed us ing numeric and graphical representations Fig 3 3 shows some screenshots of the soft ware main menu and sub menus Circumference Function to determine a circumference on a horizontal plane points minimum Aiur of Peart a in j Crcumference Parameters ays Center x 1122 mm y 40 mm eal Radium 219mm ies Ox 13 mm zii Oy 13 mm LES Dhira iets ti oni 2 2 recone Fz a Fig 3 5 Display for the measurement of a circle 3 3 MScMS actual performance critical aspects and possible improvements A preliminary prototype of MScMS has been set up and tested with the purpose of veri fying system feasibility and to evaluate its performance The prototype actual perform ance has been estimated carrying out two practical tests e Repeatability test Repeatability is defined as closeness of the agreement between the results of successive measurements of the same measurand carried out under the same conditions of measurement GUM 2004 VIM 2004 In this test a single point within the working volume is measured repeating the measurement about 50 times leaving the mobile probe in a fixed position see Fig 3 6 a The test is repeated meas uring at least 20 different points in different areas of
92. matically performed by the central PC 3 MScMS prototype 3 1 Introduction The first part of the chapter describes the features of the first MScMS prototype devel oped at the Industrial Metrology and Quality Laboratory of DISPEA Politecnico di Torino Then the results of practical tests to evaluate the system metrological perform ance are presented Finally MScMS critical aspects and possible improvements are dis cussed 3 2 Description of the first MScMS prototype The first prototype of MScMS is made by the following elements e Cricket constellation 22 Cricket devices have been freely distributed around a measur ing area covering a volume of about 60 m To make their positioning easy we used different supports such as booms articulated arms and tripods see Fig 3 1 Fig 3 1 Practical application of MScMS 24 Chapter 3 e Mobile probe It is made by a rigid structure containing the following elements two Cricket devices a tip to touch the points of measured objects Tip V and Cricket devices A and B are aligned and spaced as indicated d A B 450 mm e d A V 540 mm see Fig 3 2 a Bluetooth transceiver connected with one of the two Cricket devices by a RS232 serial port Fig 3 2 Mobile probe prototype e Personal computer An ad hoc application software runs on a standard PC To receive data sent by the probe the PC is equipped with a Bluetooth transceiver e Appli
93. ms of nodes each node memorizes the distances from its direct neighbours contained in the communication range see Fig 2 4 z distances received by device B8 distances discarded by B8 distances stored by B8 and sent to its neighbours B8 communication range Fig 2 4 Distance information handled by a single device B8 The shadow highlights the B8 communication range Principle functioning and MScMS architecture 13 2 3 4 The mobile probe The mobile probe is equipped with two Cricket devices aligned with the tip and has a Bluetooth transmitter for sending data to the PC see Fig 2 5 orthogonal view _ perspective view Cricket devices Bluetooth adapter to PC probe tip touching measured object fixed distances trigger Fig 2 5 Schematic representation of the mobile probe The probe s Crickets locate themselves using the distance information from the con stellation Crickets The principle is described in Subsection 2 4 1 System has been designed to be deployed over small or wide areas depending on the dimension of the measured objects The measuring area can be covered varying the number of constellation Crickets 2 4 MScMS software architecture This section describes software firmware features of MScMS for implementing the fol lowing operations e location of Crickets mounted on the mobile probe e location of points touched by the probe e communication and data sharing am
94. munication among network nodes Traditional routing protocols have not been designed for such exigency Researchers are working on novel light weight messaging protocols that do not rely on full TCP IP con nectivity and are capable of operating directly over low power wireless protocols Since communication is significantly more energy expensive than computation this purpose seems very reasonable In other terms because of energy and bandwidth constraints WSNs pose additional technical challenges in network control and routing collaborative information processing querying and tasking Chee Yee Chong 2003 Energy efficient routing protocols are based on two strategies which are not conflicting but rather they can be implemented in conjunction in order to increase energetic autonomy of network devices a broadcasting economization b uniform spreading of the network traffic Schurgers 2001 a The first strategy suggests to combine fuse data generated by different sensors in order to reduce the number of packets sent among nodes In practical terms neighbouring nodes are grouped in local clusters Each cluster broadcasts to others only when it collects a certain amount of data from the nodes which includes This process makes it possible to reduce significantly the network traffic and save energy Additionally it enhances communication capability because data can be compacted as they contain partly the same information The drawba
95. n the following of the chapter growht performance WSNs at the R amp D investment present time industrial revenues time Fig 7 3 Estimated s curve for WSNs technology In general the life cycle of innovation for a generic product or technology can be de scribed using the s curve The s curve maps growth of revenue performance or produc tivity against investment or time see Fig 7 3 In the early stage of a particular innova tion growth is relatively slow as the new product technology establishes itself At some point customers begin to demand and the product technology growth increases exponen 96 Chapter 7 tially New incremental innovations or changes to the product technology allow growth to continue Towards the end of its life cycle growth slows and may even begin to decline Focusing on WSNs life cycle we estimate it to be in a start up phase because WSN products are at this time entering the market very slowly see Fig 7 3 Much more ef fort should be taken to make WSNs technology popular and widely implemented As wireless sensor networks are still a young research field much activity is still on going to solve many open issues before WSNs will be ready for practical deployment Karl and Willig 2003 In next paragraphs we discuss two aspects which certainly will be critical for future advancements They respectively are power consumption and proto cols and components standardization Power
96. ne by designing several experiments that can be performed with the two systems under similar controlled labora tory conditions Regarding the iGPS a future research will deal with detailed analysis of the effects of the reference bar length used for the initial setup on measurement performance This should lead to finding an optimal length of reference bar for bundle adjustment to mini mise the error in the transmitters location Also more detailed experiments will be done iGPS performance evaluation and comparison with MScMS 89 in order to accurately characterise the system depending on different types of setup strategies and external conditions 7 Future wireless sensor networks 7 1 Introduction A wireless sensor network WSN consists of a large number of nodes with a dense dis tribution equipped with sensor devices and transceivers When networked together these devices can provide high resolution knowledge about sensed phenomena Due to dramatic advances in integrated circuits and radio technologies networks of wireless sensors are more and more utilized for a variety of applications While outdoor applications are widespread today for example consider the Global Positioning System GPS several indoor applications can benefit from knowledge of location or other physical conditions of the environment investigated Such applications span a wide range including human and robotic navigation people and objects tra
97. networks IEEE Signal processing Magazine Vol 22 No 4 pp 54 69 PEPPERL FUCHS 2005 Internal Report on Factory Automation http www pepperl fuchs com PRIYANTHA N B BALAKRISHNAN H DEMAINE H D TELLER S 2005 Mobile Assisted Local ization in Wireless Sensor Networks Proceedings of the 24th Annual Joint Conference of the IEEE Communications Society on Computer Communications INFOCOM 2005 Vol 1 Mi ami Florida March 13 17 2005 pp 172 183 PHILLIPS S D SAWYERA D BORCHARDTA B WARDA D BEUTELB D E 2000 A novel arti fact for testing large coordinate measuring machines Journal of the International Societies for Precision Engineering and Nanotechnology No 25 pp 29 34 Pozzi F 2002 Comparison of 3D Measurement Techniques in Cultural Heritage Application User Point of View Proceedings of the Ist International Symposium on 3D Data Processing Visualization and Transmission EEE Computer Society 19 21 June Padova Italy pp 762 765 PRIYANTHA N B CHAKRABORTY A BALAKRISHNAN H 2000 The Cricket Location Support system Proceedings of the 6th ACM MOBICOM Boston MA PUTTOCK M J 1978 Large Scale Metrology Ann CIRP Vol 21 No 1 pp 351 356 ROMER K KASTEN O MATTERN F 2002 Middleware Challenges for Wireless Sensor Net works Mobile Computing and Communications Review Vol 6 No 2 pp 59 61 ROOKS B 2004 A vision of the future at TEAM Sensor Review Vol 24
98. nevitable that some standardization and cases of industrial use will help on stabilizing field bootstrapping in dustrial exploitation and to attract early adopters Toh et al 2005 In the following discussion we briefly present two research issues and challenges for standardization They are 1 components standardization and 2 protocols standardiza tion I Components standardization There are different possible approaches to standardize WSNs components A first approach consists in creating a catalogue of standard parts that can be used Im plementers must ensure that their products conform to specifications The main drawback is that such a catalogue will often be out of date and incomplete In addition the specifi cations attempt to regulate the architecture of network devices could result counterpro ductive because it could discourage the development of innovative and unconventional architectures Crater 1992 Another approach for standardization is based on object oriented design principles The idea is that each sensor is independent of the microprocessor to which it is attached by specifying a digital interface and digital data sheet stored on the sensor This would al low any sensor to be connected to any network connected device Object oriented design principles ensure sensors interoperability leaving open details of implementation A third similar approach suggested by IEEE Institute of Electrical and Electronics
99. nfluenced by the reference bar length Regarding the future this aspect will be studied in detail through a structured experimental plan Fig 6 5 iGPS portable hand held measurement probe V bar In summary the measurement procedure is made up of three main steps e Spatial location of each sensor is achieved using a triangulation technique To uniquely determine the relative location of a point in a 3D space at least two transmitters are needed Chen et al 2003 Akcan et al 2006 All information needed for the location is sent to a PC for computing e As shown in Fig 6 5 the probe tip V lies on the line that connects sensors A and B similarly to the MScMS probe Fig 3 2 Therefore the location of the point touched iGPS performance evaluation and comparison with MScMS 77 by the probe tip can be calculated using the coordinates of points A Xa Ya Za and B Xg Yg Zg and the geometrical features of the probe distances dy_a and da_p e Similar to CMMs and Laser Trackers it is possible to determine or create new shapes and geometries of objects using the relevant software The geometries include cylin ders planes circumferences cones spheres and any other standard features This is achieved based on a set of measured points from the part surface Such points are col lected using the portable probe and processed using the classical optimization algo rithms Overmars 1997 6 4 Factors affecting measureme
100. ng and MScMS architecture AD AB BD 2 AB BD cos a 2 11 from which AB BD AD cos a 2 12 2 AB BD 2 13 applying again Carnot theorem to triangle P BD P D P B BD 2 P B BD cos a 2 14 Combining Eq 2 12 with Eq 2 13 we obtain 2 2 2 P D IPB BD ce a AB Eq 2 14 makes it possible to calculate the distance from the reference point P2 to the D constellation device B4 point D OO ls P X3 Y3 Z3 K b P Xo y2 Z2 P X4 Y4 Z4 a P Kis Yl Z1 a Fig 2 9 Location of constellation device B4 utilising distances from the reference points P4 Po P3 P4 The described procedure is repeated for all reference points i e P P4 in Fig 2 9 Once all required distances have been taken a trilateration technique can be applied in order to localize each constellation Cricket The acquisition procedure is driven by an ad hoc software routine Calculations are automatically performed by the central PC 20 Chapter 2 2nd approach Second approach is an extension of the first Previous localization approach is not ade quate for constellations with a large number of Crickets since each device needs knowing distances from at least 4 reference points For that reason we have implemented a semi automatic localization technique which also uses the information on the mutual distances among co
101. ng some indications on the TOF Error expected value de Experimental evaluation of the MScMS ultrasound transducers 47 pending on d O and V In order to evaluate the factors interaction but not to complicate too much the analysis we chose a 2 order polynomial model such as TOF Error C Co d C3 0 Cy V Cs d Ce 07 CV Cg d 0 Co d V Ci9 98 V 4 2 With the support of the Minitab best regression tool we constructed a model which best fits experimental results see Fig 4 13 All the terms in Eq 4 2 are considered to be significant except CV quadratic effect of factor V and Cio 0 V interaction between factors 8 and V Best Subsets Regression TOF error versus d 0 V d 0 V7 d 0 d V 0 V Response is TOF error messa as ep ele fete ele favo adj a aa a rete eee a a sie 51 7 6407 8 aee x 2 30 1 90 0 907 8 22 117 x 2 es s 89 4 994 3 22 77 x 2 _ es 4 89 4 1002 7 22 840 x 3 92 2 92 1 607 0 19 649 x a forz 91 1 749 6 20 857 x 3 ao 90 9 778 3 21 093 a 95 7 95 6 107 1 14 632 x Da 95 4 95 3 149 4 15 121 x x a 95 1 95 1 165 4 15 525 Ts 96 2 96 1 36 4 13 764 x x Ts 95 9 95 9 73 8 14 224 x x x s f o 95 9 73 8 14 224 x x e _ 96 4 96 3 10 0 13 416 x x x 6 96 2 96 2 33 2 is 711 x x x _ BELA ELA LEE 7 96 4 96 4 6 7 13 361 x x Te 96 4 96 4 8 0
102. nother inde pendently from hardware features of network devices Another standardization strategy is to extend the existing Internet Protocol IP to WSNs In other terms the aim is to connect WSNs to the existing Internet Any network wishing to be connected to the Internet needs to address the question of how it will inter face with the standard protocols like the Internet Protocol The characteristics of WSNs differentiate them from traditional IP based networks chief among these are WSNs large scale unattended systems consisting of resource constrained nodes that are best suited to application specific data centric routing These fundamental differences rule out the pos sibility of all IP sensor networks and recommend the use of application level gateways or overlay IP networks as the best approach for integration between WSNs and the Internet 7 4 Final considerations The development of sensor networking technology has been driven by advances in sens ing and computation and these technologies have been integrated by innovations in communications Providing reliable wireless connectivity self configurability and scal ability while at the same time coping with the limitations imposed by low cost energy supply miniaturization and standardization of sensor nodes presents a multitude of chal lenging research problems Future wireless sensor networks 101 This chapter provided various perspectives related to wireless sensor n
103. ns variables are calculated using the corresponding 50 individual TOF Error measurements This behaviour is well shown on Fig 4 6 where for each of the 525 factorial plan combi nations the average TOF Error and the respective standard deviation calculated using the corresponding 50 individual measurements are plotted It can be noticed that the larger the average TOF Error value the larger the individual measurements dispersion The non constancy of the TOF Error variance is also tested through the Levene s statisti cal test Since the assumption of homogeneity of TOF Error variances is violated the Analysis of Variance ANOVA cannot be properly applied in order to verify if factors have a sig nificant effect on the response TOF Error and if there are factor interactions Montgom ery 2008 The usual approach to dealing with nonconstant variance is to apply a vari ance stabilizing transformation In this approach the conclusions of the analysis of variance will apply to the transformed populations The most common transformation is the exponential y y where A is the parameter of the transformation Box and Cox pro posed an optimization method for determining the transformation parameter Box et al 42 Chapter 4 1978 Once a value of A is selected by the Box Cox method the experimenter can ana lyse the data using y as the transformed response it will be identified hereafter as cor rected TOF Error Consider
104. nstellation Crickets This technique is based on two steps e As described for the first approach the mobile probe is used to touch 4 reference points in order to locate 5 constellation Crickets ite PC D a Pa D B1 B2 D B1 B3 D B1 B4 distances utilized in the D p2 B3 semi automatic location D po B4 B1 B5 constellation Crickets of the constellation BSBA B3 B5 A B probe Crickets Dain Fig 2 10 Constellation location using the mobile probe as a ear e Subsequently the mobile probe is used as a ear to receive the mutual distances of all the constellation Crickets including the 5 which have been located Signal gathered are sent to the PC see Fig 2 10 This information combined with the information on the 5 located Crickets is used to locate the whole constellation by means of an incremental algorithm Moore et al 2004 This algorithm starts with a set of 5 no des with known coordinates Other nodes in the network determine their own coordi nates using distances from them As an unknown node obtains an acceptable position estimate it may serve as a new reference point This process can be incrementally ap plied until all nodes in the network obtain their specific coordinates Principle functioning and MScMS architecture 21 The procedure is driven by an ad hoc software routine Time required for self localization is about 1 2 minutes Calculations are auto
105. nt During the tests performed many factors affecting the quality of measurement were iden tified and analysed The most significant factors include e number of transmitters e movement of the sensors during measurement e location of transmitters setup e environmental factors These will be individually analysed in the following paragraphs Number of transmitters The number of transmitters is strictly related to their communication range and the meas urement volume Since the communication range of each transmitter is around 30m the transmitters density within the measuring volume does not have to be high For this ex periment four transmitters are used which cover a relatively large working area about 300m considering a plant layout The influence of the number of transmitters seen by a sensor on its position error is analysed using exploratory tests combined with simulation These tests are useful to ob tain preliminary indications In the future this effect will be studied in more detail by means of a structured DoE Design of Experiments Actually 30 points with a priori known positions are measured averaging 150 repeated measurements per point while the number of transmitters for the desired points is deliberately changed from 2 to 4 transmitters Coordinates position errors residuals have been determined considering the difference between the a priori known coordinates position and the co
106. nt of US transmitters The spatial location of the constellation devices fol lows a semi automatic procedure The accuracy in the location of constellation devices is fundamental for the accuracy in the next mobile probe location Patwary et al 2005 Introduction 3 1 2 The new paradigm of the distributed measuring systems For the purpose of discussion the large scale dimensional measurement systems can be classified into centralized and distributed systems In the case of centralized instruments measurements may independently arise by a single stand alone unit which is a centralized complete system i e a CMM a laser scanner or a Laser Tracker while distributed in struments are made of two or more distributed units 1 e MScMS or other innovative sys tems like the indoor GPS described in Chap 6 Metris 2007 Distributed measurement systems introduce a new paradigm in the field of large scale metrology Due to their nature they are portable and can be easily transferred around the area where the measurand is Compared to centralized systems distributed systems may cover larger measuring areas with no need for repositioning the instrumentation devices around the measured object Kang et Tesar 2004 MScMS can be classified as a modular distributed measuring system for large volume objects Even if at present ttime MScMS is still a prototype and needs to be further devel oped the system enables factory wide location of multiple
107. o sys tems Number of constellation devices For both MScMS and iGPS the number of the constellation devices depends on their communication range and the measurement volume In the case of MScMS the experi mental results showed that a coverage of an indoor working volume about 4 cubic meters large is achievable using at least one network device per square meter depending on the workshop layout Comparatively since the communication range of the transmitters of the iGPS is widely larger the transmitters density within the measuring volume is dra matically lower Sample rate With reference to the point collection frequency MScMS and iGPS are very dissimilar This difference depends on the speed of the exchanged signals between constellation de vices and probe devices The speed of US signals is about 340 m s while laser signals are considerably faster 300 000 km s Consequently MScMS sampling rate which is about 2 points per second is much lower than iGPS that is about 50 points per second Localisation of the constellation devices MScMS and iGPS give the opportunity of arranging constellation devices in different ways depending on the exigencies Every time the systems are moved that is when the position of the constellation devices is changed a localisation should be performed Ob viously this step needs to be completed before performing measurements and has strong effects on the measurements accuracy For this p
108. objects applicable in manu facturing and assembly Mainly it can be used by aerospace manufacturers but can also be adopted by automotive and industrial manufacturers both for positioning and tracking applications Since MScMS main components are a number of wireless devices distrib uted around the measuring area this not rigidly connected frame makes the system easy to handle and to move and gives the possibility of placing its components freely around the workpiece adapting to the environment and not requiring particular facilities As a consequence MScMS is suitable for particular types of measurement which can not be carried out by traditional frame instruments like conventional CMMs because they are bulky and cannot be comfortably moved The introduction of distributed measuring systems will probably have important effects on simplifying the current measuring practices within large scale industrial metrology Maisano et al 2007 This tendency is confirmed by other recent distributed measuring systems based on laser and optical technology the indoor GPS iGPS the Portable CMM and the Hi Ball ARC Second 2004 Metris 2007 Metronor 2007 Welch et al 2001 All these systems even they use different technologies are composed of a number of sensors arranged around the measuring area which can be viewed by a sensor probe measuring the object surface 4 Chapter 1 1 3 Literature review Dramatic advances in integrated
109. oid errors caused by the movements of the sensors Transmitters location setup iGPS gives the opportunity of arranging transmitters in different ways depending on the desired measuring area and the workpiece geometry Every time the position of the transmitters is changed a setup should be performed Obviously this step needs to be completed before performing measurements and its accuracy has strong effects on the ac curacy of the measurements results Patwari et al 2005 For this iGPS software provides a semi automated setup procedure that requires a few initial measurements that can be done manually or automatically for example by a robot During the setup procedure the system scale is determined by placing two sensors at known distance within the measur ing area in at least 8 different positions and orientations To that purpose a reference bar of a priori known length can be used Sa position uncertainty using the bar Ay D T Xt t transmitter to be located N N N N N N N N N I 2 4 4 4 N N 7 P N I X a uncertainty related to uncertainty related to Pia the length of the bar the length of the bar Us o A 9B 7X6 A 2 B N N N N N N bar A length bar B length Fig 6 8 Intuitive representation of the effect of the reference bar length on the transmitters local isation error When reference bars with different lengths but similar uncertainties are us
110. ong Cricket devices semi automatic location of constellation devices Fig 2 6 represents the first three operations All operations are better described in the following subsections 14 Chapter 2 X4 Y4 Z4 X2 Y2 Z2 X3 Y3 As distances from device B distances from device A 0 V point touched by the probe Xi yis Z1 Xs Y55 Zs measured X Fig 2 6 Location of points touched by the probe 2 4 1 Location of Crickets mounted on the mobile probe Spatial location of each Cricket probe is performed using a trilateration technique Trilat eration uses the known locations of beacon reference points To uniquely determine the relative location of a point on a 3D space at least 4 reference points are generally needed Chen et al 2003 Sandwith and Predmore 2001 Akcan et al 2006 In general a trilateration problem can be formulated as follows Given a set of n nodes constellation devices with known coordinates Xi Yi Zi being i 1 n and a set of meas ured distances M a system of equations can be solved to calculate the unknown position of a generic point P u v w see Fig 2 7 Ku Hyv z wy M x u y wy Hz w mi 2 3 x u y v Z w M n If the trilateration problem is over defined 4 or more reference points it can be solved using a least mean squares approach Savvides et al 2001 Martin et al 2002 Each unknown node generically P
111. onths or years It will reduce maintenance and operating costs resulting in a higher return on investment and a most efficient use of sensor network resources In order to meet this target there are two different research directions Future wireless sensor networks 97 1 New generation battery types With new applications seemingly cropping up on a daily basis and sensing devices becoming increasingly feature rich and power hungry design engineers are sending a wakeup call to battery manufacturers for innovative solutions such as hybrid lithium battery technologies or even more hydrogen fuel cells Jacobs 2004 Recently great attention is focused on miniature fuel cell because they can store a lot more energy than other standard cells making it possible to supply portable devices for long Graham Rowe 2005 In a few years networked sensors and actuators will outnumber traditional electronic appliances They will enable a plethora of new services and applications in industrial automation asset management environmental monitoring medical and transportation business and in a variety of safety and security scenarios In these conditions sensor and actuators or low power devices requiring only intermittent connectivity should be able to operate on batteries for months or years 2 Energy efficient routing protocols Another approach to reduce power consumption is to develop energy efficient routing protocols for com
112. oper alignment of the US transmitters towards the measuring area To increase the working volume coverage it is necessary to increase the number of constellation devices In general the best solution is mounting the constellation devices on the ceiling or at the top of the measuring area as shown in Fig 3 1 On the basis of practical tests we determined that the coverage of a indoor working volume about 4 meters high can be achieved using about one constellation device per Square meter considering a plant layout Use of amplitude threshold detection at receivers To evaluate time of flight TOF receivers can detect signals with amplitude equal or greater than a threshold value Since US transceivers operate at 40 kHz frequency the time period of a complete wave cycle is 1 40 000 s 25 us US waves are saw tooth shaped with a linear rise see Fig 3 10 Considering fresh US signals at the transmitter their amplitude may decrease depend ing on two basic factors e distance attenuation signal amplitude decreases depending on the distance covered e transmitter orientation since US transmitters are not omni directional signal ampli tude changes depending on their orientation In particular the maximum signal strength is related to the direction perpendicular to the transducer surface at the axis of the cone of vision while signal amplitude drops to 1 of the maximum value at 40 away from it see Fig 3 3 Priyantha
113. ordinates posi 78 Chapter 6 tion of the points calculated by triangulation Then the coordinates position errors re lated to all the 30 points are put together showing a normally distributed pattern In the simulation experiment the effect of the number of transmitters is studied vary ing the transmitters number from 2 to 8 The result showed to have a very large difference in performance between 2 and 3 transmitters Passing from 3 to 4 transmitters the im provement in the accuracy is still large For 5 or more transmitters improvement showed to be negligible This behaviour is shown in Fig 6 6 in which the standard deviations Ox Oy O related to the coordinates position errors are plotted based on the number of transmitters from 2 to 8 We can notice that the position error standard deviations re lated to 2 and 3 transmitters are much larger than the ones related to 4 or more transmit ters in fact considering the vertical axis they plot out of scale In these tests the posi tion of the 30 different measured points is assumed not to affect the coordinate position Crrors o mm position error o VS transmitters number 2 00 1 80 1 60 1 40 1 20 00 0 80 0 60 0 40 0 20 0 00 Fig 6 6 Influence of the transmitters number on the position error For example during the measurement by four transmitters if the path between a transmit
114. p Med Hoc Net 2006 June 14 17 Lipari Italy TAYLOR C RAHIMI A BACHRACH J SHROBE H 2005 Simultaneous Localization Calibra tion and Tracking in an ad Hoc Sensor Network Tech Rep Computer Science and Artificial Intelligence Laboratory of MIT https dspace mit edu handle 1721 1 30541 TODA M 2002 Cylindrical PVDF Film Transmitters and Receivers for Air Ultrasound EEE transactions on ultrasonics ferroelectrics and frequency control Vol 49 No 5 pp 626 634 TODA M DAHL J 2005 PVDF corrugated transducer for ultrasonic ranging sensor Sensors and Actuators Vol 134 pp 427 435 TOH C K 2004 Research challenges for deployable amp usable ad hoc mobile wireless networks Proc Japan Ad Hoc Wireless Networks Symposium TOH C K MAHONEN P UUSITALO M 2005 Standardization Efforts amp Future Research Issues for Wireless Sensors amp Mobile Ad Hoc Networks IEICE Trans Commun Vol E88 B No 9 pp 3500 3507 TONG C C FIGUEROA J F BARBIERI E 2001 A Method for Short or Long Range Time of Flight Measurements Using Phase Detection With an Analog Circuit IEEE Transactions on Instrumentations and Measurement Vol 50 No 5 pp 1324 1328 VALERY N 1999 A Survey of Innovation in Industry The Economist February 20 1999 p 7 VIM 2004 International Vocabulary of Basic and General Terms in Metrology International Organization for Standardization Geneva Switzerland
115. prob lematic since they may collide with the moving carriages Considering this aspect MScMSS is more flexible than CMMs MScMS and CMMs a structured comparison 59 5 3 3 Set up Installation MScMS gives the opportunity of arranging constellation devices in different ways de pending on the application requirements Every time the system is installed a localization should be performed This step needs to be completed before performing measurements and has strong effects on the measurements accuracy MScMS software provides a semi automatic procedure to achieve the constellation localization minimizing the user s effort see Subsection 2 4 4 Patwari et al 2005 Franceschini et al 2008 I It makes it possi ble to calculate the position of the wireless devices arranged around the measuring area and to establish a Cartesian coordinate reference system Nagpal et al 2003 CMMs installation requires a great effort the system made of different components has to be carried and assembled into the working place by highly skilled technicians Start up MScMS should be started up in order to activate the communication between the PC and the system and for selecting the mobile probe type Probe qualification makes it possible to know the probe geometrical characteristics necessary to determine the coordinates of the points touched by the probe s tip Franceschini et al 2008 II Also CMMs should be started up for activating th
116. rametric equation of this line is 16 Chapter 2 x x x X t y y yg y t 2 5 Z Z Z Z t The distance d A V can be expressed as 2 6 Coordinates of point V Xy yy Zy are univocally determined solving a system of 4 equations in 4 unknown values X y Z and t Xy X X X Ty Yv Z Ya yp y ty 2d Zy Z Zy Z ty 2 7 Replacing terms xy yy Zy in the fourth equation d A V xa xa x xa ty ya ya ama 4 lee 2 8 Then t 2 9 The denominator of Eq 2 9 is the distance d A B between the two Cricket devices in stalled on the mobile probe In conclusion coordinates of the point V can be calculated as Xy x X x __ fe alle a 2 10 E ap ee ee Zy Z Z5 Z Eq 2 10 univocally locates the point V using spatial coordinates of Crickets A and B Distances d A B and d A V are a priori known as they depend on the probe geometry Principle functioning and MScMS architecture 17 The previous model is based on the assumption that US sensors A and B and probe tip V are punctiform geometric elements In practice the model is inevitably approxi mated because sensors A and B have non punctiform dimensions see Fig 2 5 To mini mize point P position uncertainty the following condition should be approached d B V lt lt d A V Zakrzewski 2003 2 4 3 Cricket firmware Firmware is es
117. rs distance d 2 Transmitter misalignment 3 Battery level V angle 0 Short d 1160 mm VIIN e V 2 6 V 5 Medium d 2034 mm Vasy 4 VWie24ay Long d 3671 mm V5 2 3V all the possible 7 3 5 105 different combinations are carried out in random order for each combination TOF measurements are repeated 50 times and the average value is taken all the 105 combinations above are replicated 5 times Consequently the total number of combinations is 529 The response variable considered in the factorial plan is the TOF error defined as fol lows TOF Error Measured TOF Expected TOF 4 1 being e Measured TOF TOF measured by the couple of Cricket devices e Expected TOF d s where d is the transceivers known distance and s is the speed of sound in the experimental conditions For example with a temperature T 24 C and a relative humidity RH 27 s is about 346 m s TOF Error is used as an indicator of the inaccuracy in TOF evaluation Franceschini et al 2007 I The experiments are performed in a controlled environment T 24 C and RH 27 to prevent outlier distance measurements due to reflected ultrasonic signals or to variations in the environmental conditions 4 4 Analysis of the experimental results Subsection 4 4 1 shows and discusses the results of the factorial plan Subsection 4 4 2 summarizes them providing theoretical interpretations of some important aspects Sub
118. s and may vary depending on their relative position 1 2 cm ultrasound points of departure arrival Fig 3 7 Points of departure arrival of US exchanged between 2 Crickets Regarding the future Cricket devices will be modified in order to minimize this prob lem for example by miniaturizing the US sensors Different types of noise affecting US signal During measurements the user should not obstruct US signal propagation Two possible drawbacks may occur MScMS prototype 29 e transmitted US signal does not reach the receiver because it is completely shielded by an obstacle e transmitted US signal diffracts and goes round the interposed obstacle reaching the re ceiver In this case path covered by US is longer than the real distance between trans mitter and receiver see Fig 3 8 Fig 3 8 US diffraction The second case is more complicated to manage than the first In general it is not easy to notice possible path deflections Probe can be prone to other types of noise like exter nal sources of US For example US produced by metal objects jingling However wrong distance measurements like the ones described can be indirectly detected and rejected To that purpose an effective diagnostic tool is the Error Function EF see Eq 2 4 Franceschini et al 2002 Franceschini et al 2007 II This function evaluated during the localization of both the mobile probe devices A and B is an index of the bias
119. s distance have an im portant effect while the effect of battery charge level is minor In order to qualitatively judging the presence of interactions among the three factors an Interaction Plot is constructed in Fig 4 8 This plot represents the means for each level of a factor with the level of a second factor held constant considering two factors for each combination of their levels the mean is calculated averaging the responses obtained changing the remaining factor Interaction between two levels is present when the re sponse at a factor level depends upon the level s of other factors Parallel lines in an in teractions plot indicate no interaction The greater the departure of the lines from the par allel state the higher the degree of interaction Montgomery 2008 Interaction Plot for Corrected TOF Error Data Means 0 10 20 30 40 50 60 2 3 2 4 2 5 2 6 2 7 corrected TOF Error 0 degrees O degree O z E ut o O k O D D z V Volts Fig 4 8 Interaction plot for Corrected TOF Error considering the three factors d 0 V The qualitative result is that misalignment angle and transmitters distance have an im portant effect while the effect of battery level is minor but not irrelevant Both the factors effect end their interactions are quantitatively examined by performing an Analysis of Variance ANOVA see Fig 4 9 In the ANOVA the variance related to the response is partitioned into contribut
120. s the possibility of simultaneously performing different measurements light acoustic noise MScMS and CMMs a structured comparison 67 pressure temperature acceleration magnetic field and humidity associating them to the position measurement These kinds of measurement which can not be achieved with a classical CMM can be useful for the mapping of indoor environments Fischer et al 2001 Lilienthal and Duckett 2004 Safigianni et al 2005 Geometric relations The software functions offered by MScMS are very similar to those offered by classical CMMs e single shape measurement block 12 in Fig 5 4 In this case the measured workpiece s feature corresponds to a precise geometric shape circle plan cylinder etc e relationships among different shapes The measured feature arises from a relationship between two or more different parts of the object s shape like distances intersections or angles between curves surfaces blocks 13 15 in Fig 5 4 Concurrent measurements A significant peculiarity of MScMS is given by the flexibility of the Cricket devices They are light small and cheap and have an embedded processor to perform easy compu tations For this distributed computational capacity MScMS can simultaneously support two or more probes in order to execute concurrent measurements It is so possible to per form simultaneous measurements on a single object or even on different objects improv ing the
121. s the symptom that CMMs performance is dete riorating As a consequence whenever a stop occurs the operator has to investigate about its reason Although being performed during the measurement cycle these diagnostics can not be considered as on line since they are performed after measurements 5 3 6 Ease of use Automation MScMS and traditional CMMs are equipped with software packages which automate data processing Due to its technology MScMS operates only manually the user brings the mobile probe to the object in order to touch a set of points on its surface This is an im portant difference from CMMs which are typically controlled by CNC CMMs software makes it possible to create routines to automatically perform the same measurements on identical objects This implies a large reduction of time and costs when the number of identical objects to be measured is large By means of a self learning tool the user can also choose to manually measure the first object allowing the system to learn the meas urement patch to be repeated Unfortunately the MScMS software does not provide the same facility due to the manual nature of measurements Software user interface Both devices CMMs and MScMS provide a software user interface Their functions are based on a similar structure with the aim of guiding the user through the various activi ties Tab 5 3 summarizes the results of a comparison between the MScMS and CMMs soft ware user in
122. sensor network Future wireless sensor networks 93 WSNs can be utilized for a large number of purposes Various requirements influence the design of sensor network A scalable network makes it easy to expand and contract its resources nodes inde pendently from the performances It is often advised to focus network design on hardware scalability rather than on capacity It is typically cheaper to add a new node in order to improve network performance than to improve the capacity that each single node can handle The potential size of future sensor networks will pose a great challenge with re gard to the system scalability The combination of small size low cost and wireless networking functionality makes sensor network technology exceptionally attractive As prices become more accessible scientists will be able to deploy many sensors simultaneously with better proximity to the physical phenomena being monitored and more detailed tracking leading to ubiquitous computing Romer et al 2002 Another important requirement for WSNs is self configurability If sensor networks are to be widely deployed setting them or extracting meaningful data must be simple For example in many applications knowing the physical location of network nodes is essen tial To reach this purpose since manual methods are tedious especially for large scale sensor networks many self localization methods have been recently studied and imple mented Moreov
123. sential to organize RF and US communication among Cricket devices Firmware is written in NesC language and works under the operating system TinyOS NesC is derived from C and it is currently utilized to program MICA Mote devices pro duced by Crossbow Technologies which Crickets are derived from NesC is object oriented and event based Programs are organized in independent modules They interre late themselves by means of reciprocal queries replies MIT C S A LL 2004 Moore et al 2004 Fig 2 8 shows a schematic flow chart of Cricket firmware i i 7 1 i v i Y Y Y i v y event RF data reception event reception of the go event US signal event probe trigger pull new distances J ahead for US transmission reception data elaboration trasmission of US signal event time out for a new request of US transmission event reception via RF of authorization about US transmission RF request of f7777 handshaking RF authorization for US transmission transmission to PC via Bluetooth updating and data forwarding towards request of US transmission via RF and waiting updating and data forwarding towards neighbours via RF neighbours via RF RF channel handling US channel handling events for the modules activation C main modules C sub modules for communication Fig 2 8 A schematic flow chart of the Cr
124. sign of an ultrasonic ranging system Journal of Acoustic Society of America Vol 91 No 1 pp 486 494 FISCHER M L PRICE P N THATCHER T L SCHWALBE C A CRAIG M J WOOD E E SEXTRO R G AND GADGIL A J 2001 Rapid Measurements and Mapping of Tracer Gas Concentrations in a Large Indoor Space Atmospheric Environment No 35 pp 2837 2844 FURUTANI R KAMAHORAa K 2001 Development of a new artefact for the calibration of large scale instruments Measurement Vol 30 No 2 pp 139 143 FRANCESCHINI F GALETTO M 2007 A taxonomy of model based redundancy methods for CMM on line performance verification International Journal of Technology Management 2007 Vol 37 No 1 2 pp 104 124 FRANCESCHINI F GALETTO M MAISANO D 2007 I Management by Measurement Designing Key Indicators and Performance Measurement Systems Springer Verlag Berlin DOI 10 1007 978 3 540 73212 9 FRANCESCHINI F GALETTO M MAISANO D MASTROGIACOMO L 2007 I On line diagnostics tools in the Mobile Spatial coordinate Measuring System MScMS Forthcoming in Precision Engineering DOI 10 1016 j precisioneng 2008 11 003 FRANCESCHINI F GALETTO M MAISANO D MASTROGIACOMO L 2008 I A Review of Local ization Algorithms for Distributed Wireless Sensor Networks in Manufacturing Forthcoming in International Journal of Computer Integrated Manufacturing DOI 10 1080 0951 1920601182217 FRANCESCHINI F GALETTO M M
125. source of inaccuracy in TOF estimation Cricket s accuracy could be improved if the receiver could exactly calculate when it received the start of the pulse by implementing a more refined US detection method 4 4 3 Additional experiments The two following paragraphs present two additional experiments aimed at deepening the analysis carried out by the factorial plan They respectively are 1 complete battery discharge cycle to investigate in detail the relationship between the battery level and the error in the TOF evaluation 2 analysis of the repeatability of Cricket devices in the TOF measurements Analysis of the Cricket devices battery discharge Factorial plan results showed that the battery level has a small effect on TOF Error However abnormal TOF Error measurements were noticed during the last part of the Cricket devices battery life This test aims at studying the relationship between the battery charge level and the error in the TOF evaluation It consists in measuring TOF Error at more than a hundred different transmitter battery levels from a full charge to a complete battery discharge Transmitter and receiver are positioned at the known distance of 1582 mm with their faces perfectly aligned 0 0 Here are presented the results of the analysis of TOF Error and the respective standard deviation for different battery levels during a complete battery discharge cycle Two characteristic phases can be identified in the
126. stributed network localization with noisy range measurements Proceedings of SenSys 2004 Baltimore MD pp 50 61 NAGPAL R SHROBE H AND BACHRACH J 2003 Organizing a Global Coordinate System from Local Information on an Ad Hoc Sensor Network Proceedings of the International Workshop on Information Processing in Sensor Networks IPSN 2003 Palo Alto CA NEIL S 2005 Disruptive Technologies Sensor Networks Managing Automation Online Maga zine http www managingautomation com maonline magazine read jspx References 113 NICULESCU D AND NATH B 2003 Ad hoc positioning system APS using AOA In Proceed ings of IEEE Annual Joint Conference IEEE Computer and Communications Societies INFOCOM 03 pp 1734 1743 OH S CHEN P MANZO M SASTRY S 2006 Instrumenting Wireless Sensor Networks for Real Time Surveillance in Proc of the International Conference on Robotics and Automation Orlando Florida OVERMARS M H 1997 Designing the Computational Geometry Algorithms Library CGAL Proceedings of WACG 96 Lecture Notes in Computer Science Springer Verlag PAN M TSAI C TSENG Y 2006 Emergency Guiding and Monitoring Applications in Indoor 3D Environments by Wireless Sensor Networks Tech Rep http www cs berkeley edu kamin pubs icra06_mttvideo pdf PATWARI N ASH J KYPEROUNTAS S HERO III A MOSES R CORREAL N 2005 Locating the Nodes Cooperative localization in wireless sensor
127. system presetting start up measurement execution etc In our opinion they can easily coexist since each system has some peculiar technological features that make it suitable for specific uses The lower accuracy of MScMS makes it difficult to compete with CMMs for measuring small size objects However MScMS becomes competitive in the dimensional evaluation of large size workpieces where is often required to move the ma chine to the place where the object is Furthermore MScMS offers the possibility of si multaneously performing different measurements light acoustic noise pressure tem perature acceleration magnetic field and humidity gas concentration associating them to the position measurement Fischer et al 2001 Lilienthal and Duckett 2004 Safi gianni et al 2005 These kinds of measurement which can not be achieved with a clas sical CMM can be useful for the mapping of indoor environments 6 iGPS performance evaluation and comparison with MScMS 6 1 Introduction This chapter describes the indoor GPS iGPS an innovative measuring system to per form dimensional measurements on large scale object The system based on laser tech nology has many common aspects with MScMS The two systems are portable and easy to install and have components with small dimensions that are distributed around the measuring area For both the systems measurements are taken touching few points on the objects surface with a probe tip
128. system sample rate As the MScMS constellation is scalable and can assume dif ferent topologies different operators can measure different objects in different parts of the network CMMs are not able to simultaneously perform more than one measurement at a time 5 3 8 Cost Purchasing Cost is a point in favour of MScMS Its components Cricket devices supports and booms adapters have an individual cost of the order of some tens of euros As a con sequence the system overall cost is in the order of some thousands of euros On the other hand the cost of classical CMMs even the most economical and simple is one or two order of magnitude higher 68 Chapter 5 Maintenance The MScMS system does not need a really complicate maintenance Maintenance costs are low since the system does not require the intervention of highly qualified operators Activities of calibration and verification can be easily carried out by the user CMM maintenance is a much more complicated activity because it needs well pre pared operators to maintain the system Typically maintenance contracts cost about three thousand euros per year for a single CMM 5 3 9 System management From a system management point of view the two measuring systems mayor implica tions concern two phases set up and measuring Set up Before performing measurements both the 2 systems need to be set up Regarding MScMS the operator has the possibility of placing th
129. t be completely suitable for dynamic measurements However by predicting the direction of movement and by us ing error compensation methods this limitation may be resolved and iGPS could poten tially be utilised for slow dynamic measurements MScMS and iGPS are compared in order to highlight the pros and cons of each sys tem based on the experimental results and available information from the literature In measurement activities MScMS and classical 1GPS are similar however they present many differences due to their different technological features The technological differ ences affect several factors within the systems including system presetting start up and measurement execution It can be concluded that these systems can easily coexist since each system is suitable for specific applications due to their technological features The metrological performance of iGPS is superior compared to MScMS however the overall cost of MScMS is more attractive in applications that do not require a higher level of ac curacy Both of these systems are lightweight easily adaptable to different working envi ronments and can be rapidly installed and used Prior to performing measurements con stellation devices are freely distributed around the area of work and semi automatically located in a few minutes Future work includes detailed experiments in order to more accurately characterise the advantages and weaknesses of the two systems This will be do
130. t have a phase cancelling effect This because the distance from an arbitrary point on the transmitter surface to the receiver becomes al most constant and the difference is much smaller than the wavelength Toda 2002 On the other hand if the transmitter is misaligned with the receiver the US signal amplitude will be attenuated because of the disruptive interference of the different US signals from the transmitter different surface points This effect is represented by the simplified scheme in Fig 4 2 This scheme considers the interaction of the waves from two points on the transducer surface the same principle can be extended to all the surface points 36 Chapter 4 ra i vt 1 a J Nwielal Come _ E N F a Na T f a w j k al t Py F i S l i T Plastic Binse Lt Piezo Disk oni h 100 al 40 4 Connector A Connector H B a b Fig 4 1 a internal construction of a Murata MA40S4R piezo electric ultrasonic transmit ter receiver The dimensions of the piezo material causes the disk to resonate at a precise fre quency around 40kHz b representation of the transmitter bandwidth by means of a frequency response plot Receiver Transmitter faces aligned Transmitter full amplitude Receiver misaligned a b Fig 4 2 US signal strength dependence on the transmitter angle 0 The simplified scheme repre sents the interaction of the waves from 2 points on the transducer surface
131. tal effects like temperature gradients vibrations or direct light Also the quality of the initial system setup is a fundamental aspect It is also shown that with the existing technology iGPS may not be completely suitable for dynamic measurements However by predicting the direction of movement and by using error compensation methods this limitation may be resolved and iGPS could potentially be utilised for slow dynamic measurements Furthermore MScMS and iGPS are compared in order to highlight the pros and cons of each system based on the experimental results In measurement activities MScMS and classical iGPS are similar At the same time they present many differences due to their different technological features Both of these systems are lightweight easily adaptable to different working environments and can be rapidly installed and used Prior to performing measurements constellation devices are freely distributed around the working area and located performing a semi automatic procedure The 1GPS performance in terms of repeatability reproducibility and accuracy has been studied by initial experiments According to the results 1GPS repeatability and reproducibility are approximately two orders of magnitude better than the MScMS However the overall cost of MScMS is more attractive in applications that do not require a higher level of accuracy Regarding the iGPS a future research will deal with detailed analysis of the effects of
132. tected Therefore it does not depend upon the absolute magnitude of the pulse but only upon its shape As a consequence it is more accurate and robust than sim ple magnitude thresholding where the acknowledge time can easily jump by one period Phase detection Other more refined ranging methods are based on phase detection with fixed frequency signals and with frequency modulated signals These methods however requires complex hardware and software They use a digital signal processor to process 38 Chapter 4 the phase measurements and overcome the inherent range limitation of one wavelength Manthey et al 1991 Tong et al 2001 Transient time at the receiver porte a ee ee ee fe 3 detection threshold 1 J Receiver electric output signal mV i aoe half amplitude signal full amplitude signal a 20 40 BU o0 100 120 1440 160 160 200 220 240 260 2860 SU Time from the arrival of the first US signal at the receiver us Fig 4 4 Schematic representation of thresholding detection A minimum number of cycles are necessary to bring the receiver to steady state conditions transient time at the receiver Johansson et al 2005 The error in the distance measurement is dependent on the received US signal ampli tude because the time taken for the received signal to reach the threshold is dependent on it 4 3 Factors affecting US transceivers MScMS measurement accuracy may change
133. ted to each Cartesian co ordinate As expected the standard deviations are higher than the repeatability tests 3 Accuracy test Accuracy of measurement is the closeness of the agreement between the result of a measurement and the value of the measurand GUM 2004 VIM 2004 This test was performed using a calibrated reference artefact with known di mensions Cross et al 1998 The reference artefact consisted of two one meter bars assembled to create a two meter long reference bar The reference bar was made of composite materials with different isostatic supports on which the mobile probe can be placed during measurement see Fig 6 9 The nominal dimensions of the artefact iGPS performance evaluation and comparison with MScMS 83 points nominal position and nominal distances between points are calibrated using a laser interferometer and a CMM which are at least two orders of magnitude more accurate than the iGPS These distance measurements are repeated by placing the ar tefact in 30 different positions and orientations within the measuring area To repro duce a common measuring strategy each point position is calculated by averaging 150 single position measurements The standard deviation related to the distance re siduals Gpisr in Tab 6 1 that is to say the differences between nominal distances and distance measured with iGPS is also calculated Moreover for each point coor dinate the residuals between the measur
134. ter and a desired sensor is accidentally blocked and the sensor can only see three of the transmitters the measurement quality will drop This can happen when the line of sight between a sensor and one or more transmitters is obstructed by the operator or the workpiece body Consequently the transmitters should be arranged around the measuring area in suitable positions to gain maximum coverage e g near the ceiling to reduce the risk of obstructions Regarding the future some trials will be carried out in order to iGPS performance evaluation and comparison with MScMS 79 study the best way of positioning the transmitters depending on the measured object and the measuring area Sensors movement during measurement iGPS can be used to perform either static or dynamic measurements In particular during aircraft assembly operations it can be useful to perform dynamic measurements How ever the system performs best in static measurement This is due to the positioning method used The position of each sensor can be calculated by triangulation using the two angles and 0 from each transmitter Transmitters sampling rate depends on the angu lar speed of their rotating heads As explained above the spinning speed is unique for each transmitter to be differentiated Assuming the rotation speed is around 3000 rev min each transmitter will be able to communicate with sensors about 3000 60 50 times per second Even though the transmitters
135. terfaces As for CMMs MScMS software has been developed to help operators by e leading them through the start up and measuring activities e providing tools and functions which simplify their work e displaying the results in a clear and complete way 64 Chapter 5 Tab 5 3 Comparison between the MScMS and CMMs software packages Stage System Startup System presetting Dimensional measurement Activities System initiali zation Probe qualifica tion Constellation localization Choice of the measuring activ ity Selection of the shape or rela tionship to measure Measurement execution Audio visual signals Output display Software tools MScMS Semi automatic procedure to open the Bluetooth connection Manual definition of the probe s geometrical features Semi automatic procedure guided by visual instructions Display and memorization of the lo calized constellation layout Single shape measurement Relationships among different shapes distances intersections or angles Selection of the shape or relation ship to measure Measurement setting and execution Warning signals Display of the probe s communica tion range and network connectivity Numerical and graphical display of the measured points 2D and 3D charts Numerical and graphical display of the object s features Measurements System diagnostics CMMs Semi automatic proce dure to start up th
136. the detection range In other terms considering the same US signal amplitude the radiated signal amplitude at a given distance from the transmitter becomes smaller if its bandwidth and resonant fre quency increase Tong et al 2005 Kazys et al 2007 For this reason the selection of ultrasonic frequency and bandwidth is a compromise between accuracy and detection range The piezo electric transducer adopted by Cricket devices is a low cost general purpose model Murata MA4OSAR see Fig 4 1 a with a relative wide bandwidth see Fig 4 1 b in which the centre frequency is about 40kHz This working frequency is a trade off between accuracy considering the single distances it is around 1 2 centimetres and de tection range up to 6 8 meters Balakrishnan et al 2003 The acoustic strength of the radiation from a flat transducer with piston motion like the Crickets US transducers is generally angle dependent because of the phase differ ence of waves from each point on the surface Actually the acoustic radiation is the inte gral sum of the waves from all points on the transmitter surface and the propagation path difference from each point to a reference observation point has a phase cancellation effect which leads to signal attenuation Lamancusa Figueroa 1990 However if the receiver is directly facing the transmitter at sufficient distance from it the acoustic radiation from each point of the transducer surface does no
137. the effect of the two most significant factors O and d on the TOF Error is evident while the effect of the battery charge level V is very small compared to the previous two TOF Error is always positive because of the TOF overestimation due to the signal attenuation which is proportional to d O and V The effect of the transceivers distance on the TOF Error is also well shown in Fig 4 12 plotting the TOF Error versus the transceivers distance for different misalignment angles aca TOF Error vs Transceivers Distance d e 00 degrees ns 10 degrees 300 0 4 20 degrees gt x 30 degrees 250 0 40 degrees e 50 degrees 60 degrees 200 0 150 0 TOF Error us 100 0 1000 1500 2000 2500 3000 3500 4000 Transceivers Distance d mm Fig 4 12 TOF Error depending on the transceivers distance d The plotted curves are related to different transmitter misalignment angles 0 The effect of the signal attenuation TOF overesti mation increases with the transceivers distance The most interesting considerations related to the factorial plan experiments are dis cussed and interpreted in Subsection 4 4 4 Linear regression model Considering the results of the factorial plan we constructed a linear regression model rep resenting the relationship among TOF Error and the three factors d O and V Such a model can be useful for providi
138. the factorial plan Surface Plot of TOF Error vs d and 0 300 200 TOF Error um 100 ee 4000 3000 d mm 2000 1000 40 20 0 degrees Fig 4 10 Surface plot to represent the effect of the interaction of factors d and O on the TOF Error Experimental evaluation of the MScMS ultrasound transducers 45 The effect of this interaction on the TOF Error is represented by the surface plot in Fig 4 10 As shown the composition of large misalignment angles 0 and large distances d produces TOF Errors which are larger than the these obtained adding the effects of the single factors taken separately Another representation of the experimental outputs is given by Fig 4 11 where the av erage TOF Error and the corresponding standard deviation calculated for each combina tion of factors using the 50 repeated measurements are plotted depending on V and 9 for each of the 3 transceivers distances Average TOF Vs Misaligment Angle d 1160 mm Average TOF Error us TOF Error Standard Deviation us 0 10 20 30 40 50 60 70 Misalignment Angle degrees TOF Standard Deviation Vs Misaligment Angle d 1160 mm 10 20 30 40 Misalignment Angle degrees 50 60 70 Average TOF Vs Misaligment Angle d 2034 mm Average TOF Error us TOF Error Standard Deviation us
139. the software viewpoints Other techniques for compensating the measurement error are not easy to be implemented because of the difficulty in simultaneously controlling all the factors producing the US signal attenuation Comparison between MScMS and CMMs MScMS was compared to the classical CMMs the most commonly used equipments for objects dimensional measurements Considering measurement activities MScMS and classical CMMs are similar However due to their different technological features they have many differences for example system presetting start up measurement Conclusions and future directions 105 execution etc The lower accuracy of MScMS makes it difficult to compete with CMMs when it comes to measuring small size objects On the other hand MScMS becomes competitive in the dimensional evaluation of large size workpieces where is convenient to move the machine to the place where the object is and where the required level of accuracy is not very high iGPS performance and comparison with MScMS The main issues and factors affecting the quality of 1GPS measurement were reviewed The introduction of the 1GPS and other measuring systems based on distributed components may have important effects on simplifying the current measuring practices within large scale industrial metrology For iGPS it is shown that the result of measurements improves by increasing the number of transmitters and also controlling the environmen
140. ther cases WSNs can be seen as the key ingredient to allow the proliferation of new technologic solutions which enhance the per formances of existing products or processes Water steam Electricity Chemicals Petrochemicals Digital networks power Rail Intemal combustion Electronics Software Textiles Steel engine Aviation New media iron First Wave Second Wave Third Wave Fourt ve Fifth Wave Fig 7 2 The waves of innovation Valery 1999 On the other hand the industrialization process of WSN components is certainly in the early introduction phase and as a consequence develops very slowly Several commer cial companies such as Ember Crossbow and Sensoria are now building and deploy ing very different types of small sensor nodes Sirbu et al 2006 As noted by Abernathy and Utterback near the beginning of a new product market and before the emergence of a dominant design there is a great deal of product variety Abernathy and Utterback 1978 At the turn of the 20th century automobiles came with internal combustion engines or steam engines three four or five wheels front steering or rear steering and many other configurations before the dominant design of a four wheel vehicle with internal combustion front wheel steering and rear drive wheels emerged At the turn of the 21st century we are in a similar place with sensor networks Different vendors produce incompatible products of proprietary design The
141. tial setup to con tinually correct the measurement field and determine whether the system is conforming to the desired tolerance Kang and Tesar 2004 Cost Cost is a point in favour of MScMS since its main components including Cricket de vices supports and booms adapters etc have an individual cost of a few tens of euros This reduces the overall cost of the system On the other hand the cost of iGPS even for the most economical and simple configuration is around 200 000 6 7 Summary and final considerations The main issues and factors affecting the results of iGPS measurement are reviewed The outline system performance in terms of repeatability reproducibility and accuracy was 88 Chapter 6 studied by initial experiments According to the results that are obtained by averaging 150 readings for each point s measurement the accuracy results are within 0 2 mm This is achieved over a two meter length however for real large scale metrology similar experi ment should be repeated for larger size lengths for instance 10 to 20m The result of measurement improves by increasing the number of transmitters even if for 5 or more transmitters improvement showed to be negligible and also controlling the environmental effects like temperature gradients vibrations or direct light Also the quality of the initial system setup is a fundamental aspect It is also shown that with the existing technology 1GPS may no
142. ts for example the 72 Chapter 6 MScMS system uses a network of devices In general distributed measurement systems due to their topology and the light weight of each of their units are portable and can be easily transferred to the measurand The other distributed contact measuring instruments shown in Fig 6 1 are the Metris iGPS and the 3 Tech Hi Ball Hi Ball is a system composed of a number of infrared LEDs arranged around the measuring area which can be viewed by an optical sensor probe measuring the object surface The probe is able to locate itself measuring the angles from the LEDs and performing a triangulation Welch et al 2001 LARGE VOLUME MEASURING INSTRUMENTS CONTACT NON CONTACT CENTRALIZED DISTRIBUTED CENTRALIZED DISTRIBUTED CMM Indoor GPS Theodolite Photogrammetry e g Laser Tracker MScMS Tacheometer Laser Radar Hi Ball Optycal probe CMM Fig 6 1 Classification of large volume measuring instruments iGPS is a modular large volume tracking system enabling factory wide localisation of multiple objects with metrologic accuracy applicable in manufacturing and assembly The system components of iGPS are a number of transmitters a control centre sensors and receivers Kang and Tesar 2004 The distributed nature of the system eases the han dling and provides scalability for the coverage of the measuring area For this reason iGPS is more suitable for particular types of measurement which can not be carried out
143. urement and its portable probe Measurements are taken by touching the required points on the object s surface with a probe that is equipped with double sensors Points are defined on a Cartesian coordinate system the coordinates are then processed by specific algorithms in order to determine geometric features Such measured features are then used to extract the desired dimen sional information such as feature positions and angles between two features ARC Sec ond 2004 There are several standards for conventional dimensional metrology systems ISO 10360 2001 ANSI ASME 2006 However currently there are no international stan dards or best practice guide for the application of iGPS For this actually the system measuring performance is strongly dependent upon the system configuration arrange ment of the transmitters and setup 6 3 iIGPS technology and operating features Typically system components of 1GPS are two or more transmitters a control centre and a number of wireless sensors Transmitters operate as reference points with known position continually generating three signals two infrared laser fanned beams rotating in the head of the transmitter and 74 Chapter 6 an infrared LED strobe Maisano et al 2008 Arc Second 2004 Sensors are passive elements which can be placed on the surface of the object to be measured to receive the transmitters signals iGPS is a scalable or modular system since the number o
144. urpose MScMS and iGPS provide two different semi automated localisation procedures both requiring few manual measure ments System calibration check Another activity to make MScMS suitable for the measurement is the system calibration check It is well known that the speed of sound changes with air conditions in terms of temperature and humidity which can exhibit both temporal and spatial variations within large working volumes As a consequence the speed of sound should be often measured 86 Chapter 6 and updated in the calculations To real time verify its value an optimisation procedure is implemented A similar procedure is applied into iGPS using a reference bar What became clear from the tests is that iGPS s absolute uncertainty is directly related to the quality of the scale bar measurement and its initial calibration The procedure can be fully automated using two fixed sensors which are tied to the extremities of an interferometric scale bar The implementation of auto calibration minimises downtime and corrects for environ mental conditions in the measurement field continuously and in real time Metrological performances Results of preliminary repeatability and reproducibility tests to evaluate the performances of MScMS and iGPS are shown in Fig 3 1 These tests are described respectively in Sec tions 3 3 and 6 5 Tab 6 3 Comparison between the MScMS and the iGPS metrological performances The specific testing
145. use of their dimensions or other logistical constraints and thus require the measuring system to be moved to them In the dissertation the system is described exhaustively and characterized through prac tical experiments Then the system is compared to classical CMMs and the indoor GPS iGPS an innovative laser based system for large scale metrology Finally future direc tions of this research are given Contents List Of acronyms and abbreviations ssssccscccccrssssssscscccccccssssssssscccsscessssssssccssssseees Vil Di Tntrod chomasnonanisaonosoinin eo asda esa sae sean adeeb eae 1 1 1 The Mobile Spatial coordinate Measuring System MScMS ccceeseeeeeeees 1 1 2 The new paradigm of the distributed measuring systems cccccceeeeeeeeeeeeeeeees 1 Lo EMER AUT TeVe Wee o a ae eepeanteneanceaatientensnaGee saueeatads 1 1 4 Organization of the dissertation ccccccceececccccecceceeeeeeecaaaseesseseeeseeeeeeeeeeeeeeeeeaas 1 2 Principle functioning and MScMS architecture ccsssssssccccsssssssscssccccsseeeees 1 De ToduC HO Meere E S sarin oeeeesecastea vane 1 2 2 System requirements and comparison with other measuring techniques 1 2 5 MSCMS hatdwirecgui Pen ai a 1 Zoek C ACKECdEVICES a E E E E a ee 1 2 3 2 Evaluation of distances between Cricket devices sssssssenessssssssseeersssssse 1 2a CLICKELS COMMMUMICAIOM rea EE E E E 1 ZO ME MODRC PLODE aaia e
146. y have made very dif ferent choices in the design space according to their respective competences target mar ket or limitations of the available technology Standards which define dominant de signs will gradually begin to emerge Future wireless sensor networks 95 However we have to state that regarding the present WSN applications are still in infancy even if there has been a tremendous amount of work done towards solving re search problems Toh 2004 It is quite remarkable how little we have real mass market wireless ad hoc products available taking into account the massive amount of research done Oversimplifying and somewhat exaggerating networking research has been more strongly technology push related and apart of few special cases such as military net works there is only a limited number of well recognized and accepted application cases Chai Keong et al 2005 It is not easy to forecast the future of WSNs due to the great abundance of industrial areas of interest but at the same time the lack of tangible strategies to transfer the best new ideas onto the market by demonstrating benefits to both users and the company that provides those benefits Weiss 2002 Some experts opinion is that sensor networks could potentially become a disruptive technology when the miniaturization power con sumption standardization and cost problems are solved The last two of these issues are discussed i

MScMS (Mobile Spatial coordinate Measuring System)

Contents

Download Pdf Manuals

Related Search

Related Contents