USF Libraries
USF Digital Collections

Field performance evaluation of Autosense II at Leesburg mainline toll plaza


Material Information

Field performance evaluation of Autosense II at Leesburg mainline toll plaza
Physical Description:
iii, 16 p. : ill. ; 28 cm.
University of South Florida -- Center for Urban Transportation Research
University of South Florida, Center for Urban Transportation Research
Place of Publication:
Tampa, Fla
Publication Date:


Subjects / Keywords:
Traffic flow -- Equipment and supplies -- Testing -- Florida -- Leesburg   ( lcsh )
Imaging systems -- Testing -- Florida -- Leesburg   ( lcsh )
technical report   ( marcgt )
non-fiction   ( marcgt )


Additional Physical Form:
Also available online.
Statement of Responsibility:
prepared by Center for Urban Transportation Research, University of South Florida, College of Engineering.
General Note:
"February 2000."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001926049
oclc - 45193385
usfldc doi - C01-00424
usfldc handle - c1.424
System ID:

This item is only available as the following downloads:

Full Text


:,::'":: : . . . . : : . /": /'_/ : : . ... .. ' / / "'. . AUT.OSENSE. II at Leesburg Mainline Toll Plaza .... .. . . . -: ' . . . .. . . : ..


Field Performance Evaluation of Autosense JJ At Leesburg Mainline Toll Plaza Prepared by The Center for Urban Transportation Research Univers i ty of South Florida CoUege of Engineering February 2000 The opinions and findings expressed in this report are those of the Center for Urban TransporftJtion Research (CUTR) and the U11iversity of South Florida and not necessarily those of the Florida Department of Transportation Office of1bll OperatiollS (FDOT-OTO) or the project's technology partner (Schwartz Electro-Optics Inc.). This report has been prepared in coopera tion with the FDOT -OTO, the Tlrnpike Data Center in Boca Rato11, and Schwartz Electro-Optics, Inc.


ACKNOWLEDGEMENTS CUTR would like t o recognjze the funding agency for this e v aluation project the Florida De partment of Transportation Office of Toll Operations. In particular J i m Davis and Doug Mar tin who served as FDOT project comanagers, and Brett Massey who provided onsite coordi nation at the Leesburg Mainline Toll P laza. Additionally, Ms. Carrie Latimer, at the Turnpike Data Center in Boca Raton, provided plaza lane transaction reports to CUTR on a daily basis throughout the evaluation period. CUTR sincerely appreciates the opportunity afforded by the FDOT-OTO to conduct th i s independent field evaluation on their behalf. The techno l ogy vendor partner in this evaluation w a s Schwartz E lectro-Opt ics, Inc. (Orl ando, FL), represented by Norman Abramson-Director Bus i ness Development. Michael Pietrzyk, CUTR s ITS Pl:ogram Director, served as principal investigator under thjs contract. The other primary CUTR project staff was Alasdair Cain (Graduate Research Assis tant ) and Susan Horsman (Program Assistant). Without the assistance and effort pro,ided by the aforementioned individuals, thls evaluat ion coni<\ not have been conducted and documented i


EXECUTIVE S UMMARY The Florida Department of Transportation (FDOT) Office of Toll Operations (OTO) retained CUTR to evaluate the functionality, capability and accura cy of Autosense II" (ASm. ASII is a diode-laser-based veh. icle detection and classification sensor manufactured by Schwartz Electro Optics of OrlaJldo. The device is ideally mounted twenty-three feet above the road smface and emits two laser beams at a f ixed angl e s epa ration of ten (10) degrees onto the pavement. As a vehicle passes under the device, the laser beams are broken and the device is able to generate a series of two-dimensional scans of the vehicle. Once the vehicle bas passed fully through the beams, a three dimensional image of the vehicle is developed Th is image is used to classify the vehicle using in-b uilt device algorithms. The evaluation took place at Leesburg Mainline Toll Plaza on the Florida Thrnpike, between May 26 and July 2, 1999. The objective of the evaluatiou was to assess the accuracy of ASH as a vehic le separator Therefore, in one of the tollbooth lan es, the existing device lmdertaking vehic le s eparation was re moved and replaced with AS II. Comparing ground tnth videotape of the lane with the transaction r ecord generated by the plaza computer completed the evaluation. This ground tmth comparison allowed real time evaluation of any errors within the system The capability of ASH to perform vehic le classification was not tested in this evaluation. Over the course of the evaluation period, a total of 28,268 vehicle transactions were assessed. There were a total of twenty three system separation errors during this period. A majority (18 of 23) of the errors occurred when two consecutive two-axle vehicles were r ecorded as a one axle ve. bicle followed by three-aJ<:le vehicle. The likely explanation for th ese 18 errors is a time delay in l ane controller p rocessing of the treadle inputs. This results in a "lower bound" accu racy o f 99.92 percent and an "upper bound" accuracy of99.98 percent. ii


ACKNOWLEDGEMENTS TABLE OF CONTENTS ............................................................................................. EXECUTIVE SUI\B IARY ............................................................................................ ii T C .. ABLE OF ONTENTS .. .. .. ................ .. .. Ul PROJECT BACKGRO UND AND OVER VIE\V .................................... .......... ..... ............. 1 AurosENsE n DE\rrcE DEScRlP'rroN ................................................. 2 Qvervie'V uuouooooooooooouo o u o u o u oo u- 2 Devlce-Spec:Uications .................................................................................. ........... ........ 2 De,ice Operations .......................................................................... ..... ........................ 3 Install ation of ASII at Leesburg Mainline Toll Plaza ... ... ... .. .. .... .. .............. .. .. .. ..... 4 ME'J'HODOLOGY .................................................... .................................................. 4 Description of Source Data ................................................................. ........................ 5 Da.ta formatting and E'9alu.ation Plan .......................................................................... 6 Ground 'I'ruth Comparison ............................................................................................. 9 ())? -JL() General E'rror Apprisal .............................................................................................. 10 Discu-ssion of Individua1 Error Types .. . .. ..................................................................... 12 Sunmtary of F'indillgs .................................................................................................. 14 A.PPEND IC IES Appendix A ASll Product lofonnation A ppendix ll Performance Comparison Between ASll and IDlUS Appendix C Plaza Computer Text File Output of Lone #5, Le .. burg Toll l'laz a Appendix D Separation Error Snnunary LIST OF TAB. L ES Table 1: Upper Bound Error Table for Varying for 95% Confidence IntOI'Yal ....... 2 Table 2: ASD Devil.-e Specifi<."dtions .............................. ............... ................. .. .. ............... ................. ....... 3 Table 3: Clas,o,ificat.ion of Error Types ............................................. ......... ... ......................... ...... ......... 9 Table 4: Summary Results o! ASll Evaluation ....................... _.. .......... ......... ....................... .... ......... 11 LIST OF FIGURES 1gure 1: ASllllevlce ..... ......................................................... _,. ............................. ............. ............... 3 l'1gure 2: Operational ConJlguration of ASU (not to scale) _.. ............. _.. ........................................... 3 F'igure 3: 1\.Sll Vehicle Detection Sequence ............................................................ -............ ...... ...... .. 4 Figure 4: Installation of ASH at Leesburg Mainline Toll Plaia ...................................... _.. ............... S F1gure 5: Schematic of Leesburg Tolll'laza System Operation ....... ........... ..... ........ ....... ...... ............ 5 F1gure 6: ASll.rd Test : l>ata Coll.

Field Performance Evaluation of Autosense ]] At Leesburg Mainline Toll Plaza Prepared b y The Center for Urban Transportation R es earch U ni vers ity of S outh F lo rida C ollege of E n g i n ee r ing F ebruary 2000 The opurions cmd findings expressed in this report are those of the Cen ter for Urban Transpor tation Research ( CUTR) and the University of South Florida and not necessa r ily those of the FlorU!a Department of Transportation Office of Toll Operations (FDOT -OTO) or the proj ect's technology partner (Schwartz Electro-Optics, Inc.). This report has been prepared in coopera Tion w it h the FDOT-OTO, the Turnpike Data Center in Boca Raton, and Schwart z Electro -Optics, I nc


Sample Number of Errors Size I 2 3 4 5 250 100.00% 100.0 0 % 100.00% 29.26% 99.74% 500 100.00% 100 00 % 1 00.00% 92.2a% 99 .87% 750 100.00% 1 00.0 0 % 100.00% 99.92% 1 00 0 100.00% 100 00 % 1 00 .00% 99.99% 9 9.9 4 % 125 0 100.00% 100.00% 100.00% 99 .29% 99.95 % 1500 100.00% 1 00 .00% 100.00% 99 .99% 99.96 % )750 100.00% 100. 00 % 1 00 .00% 100.00% 92. 9 6 % 2000 100.00% 100. 00% 1 00.00% 1 00 .00% 99 97' "' 2250 1 00 .00% 100 .00% 100.00 % 1 00.()(1% 99 97% 2500 100.00% 100 0 0 % 100.00% 100.00% AUTOSENSE II DEVICE DESCRIPTION Overview (AS II) is a 0iode laser-based vehicle dete ction and classification s enso r (YDC). The de vice was d esigned by Schwartz Electro Optics (SEO) and was developed u nder an Intell egent Trans portation Systems Ide as De serv ing Exploratory Analysis ( IHYS-IDEA ) program for the Na tional Ac ademy of Sciences The device relies on an inhere n t laser cha racteristic narrow an gular beam wid th t o prov id e the high resolut ion req ui red for accurat e ve h icle protiling. Th is narrow laser beam width permit s the detect ion of c l osely s p aced veh icles moving at high speed The device produces an overhead imag e of vehicles as t hey pass be low w hich can be used to perform vehicle separation and class i f i cation AS II is part of the Autosense family of sensors which incl udes Autose11se I", Autose11se and Autosense Autosense IIA ha s the additional capability of coun t ing ve h icle a x le s while Autosense Ill" i s capable of classifying vehicles in multiple tr ave l lane s. Device Specifications The specifications of t he device are shown i n T able 2. The device is desi g ned to be robust and require very litt l e maimenaoce The only m ai n tenan c e requi r emen t is keeping t he window clean. as a dirty win dow c an resul t in range errors. The ASH hou sing is nitrogen purged, hermetically sea led prior to shipping, and s ho uld only be opened in a labora tory environment by SEO personnel. SEO staff set all calibration and alignment ad j ustmen ts during final te st and no tleld adjustments are required Initia l installat io n of the dev ic e takes one to two hours and the mean time to repair is 10-15 minutes The de vice is designed to have an unlimited life-sp an, ass u ming that th e device is ret u rned to SEO every five years for routi ne servicing T he dev ic e is shown in Fig u re 1.


Table 2: ASO Device Specifications Mounting J.loight 20 to 25 feec above laue (reeolnmended) 19.lln X 9.6in X Sin Weigbc 2Slb.s Powe. r u H5VAC, 60Hz, 3!SWatu. with hO:'I.tcr: 140W;:,.tt:,; Power Option 230VAC lntcJ:fac:e RS422. Triggct, (R.S232. fiber optic port opti<>.naJ.) u.s. Patents 5.321,490 S,S46,188 Others Pending Device Openttio n The ASll device is shown in Figure 1, and its operational configuration is shown in Figure 2 The ASH device emits two narrow laser beams, at a fixed angle separation of 10 degrees. The first beam has a look down angle of 10 degrees aud the second beam has a look down angle of zero degrees, as shown in Figure 2. At a mounting height of 23 feet, this I 0-degree separation equals a distance of four feet between the two laser beams. The laser beams continuously scan the road at a rate of 720 scans per second. Figure 3 shows the detection messages produced by ASII as a vehicle passes through..the range of the device A transverse height profile of the vehicle is generated each time the vehi c l e i s scanned. Once the end of the vehicle has been detected, ASH generates a three dimensional image of the vehicle by calcula t ing the vehicle speed and combining the sequence of two-dimensional scans. Once the three dimensional image is obtained, an algorithm is used to classify the vehicle. Once thi s i s completed the final output from ASH, the Vehicle Classification Message, i s produced It should be noted that the capability to classify vehicles i s not utilized in this evaluation. In its conf i guration at Leesburg P l aza, ASII is utilized purely as a veh icle separator FirstBesm..._ r 23' For AS/I 1 Figure 1: ASII Device Figure 2: Operational Configurolio11 of ASII (twt to scale) 3


Wodloo qf Tovcl First Beam--+ +--Second Beam output Mes:seqe.s +---+--------Ye:ua ge tl +----------w.e .. uaseln '---------u:e;ssase M Figur e ASI/ Vehicle Detection Sequen ce o f ASH at Leesburg Mainline ToU Plaza Durin g the first week of May, FOOT staff replaced the existing AVC device with AS II in Lane # 5 a t Leesburg Mainline To ll P l a za. The installed ASH device is s h own in Figure 4. FOOT staff angled the device so that the l aser beams cont ac ted t he pavement surface j ust in f ront of the tread l e. On May 18. S E O staff v isited the Lees burg Plaza and re a ligned the ASH device so that the tread l e lay directly between the two l aser beams Figur e 5 ind i cates the configuration of the ''Integrated Real Time Network" at Le esburg Toll P laza. AS II was p art of the Automat ic Vehicle Classi fica t ion elemen t of the network. ASH was responsible for info r ming the QNX Lane Contro ller of the beg inning and end of each vehicle t hat passed thr ou gh the pl aza, so that t he vehicles could be correc tly separ a ted. ASH was also responsible for s ignalin g t h e end o f the transac tion to the system. The QNX Lan e Controller u sed data f rom ASH, the treadle, and the toll collector to determine the number of axles and therefore the required toll, for each vehic l e transacti o n ASH s ability to classify vehicles was not utilized in the evaluation. A new SunPass compon e n t allowing El e ctronic Toll Collec tion, is due to be added to the QNX Lane Controller software early in the year 2000. METHODOLOGY The purpose of the evalua tion was t o ass ess the accuracy of ASH by comparin g ground truth videotape of Lane #5 at Leesb urg Toll P la za with the t ransact ion record generated by the to ll p laza comp uter The bas i c proces s used to obtain the data required for the eval uation is shown in Figure 6.


Threadle Figure 4: JnstalliJ/il)n 1)/ A.SU at l-eesburg Mainline Tl)l/ Plaza Altomatic Vehicle Classification Tol Terminal t r <1 CC[V I maae Qvedav Text File I Figure S : Schemlllic 1)/ Ltesburg Toll Pli>z.a System Operotio11 Description of Source Data As i s shown io Figure 6, there were two sources of data from Leesburg Mainline Toll Plaza which were required for ground truth comparison CC1V Videotape The fixst data source was the Closed Circuit Television (CCTV) videotape, obtained from Leesburg Toll Plaza. CCTV cameras view the to ll booths duriog operational hours and a over lay of rea l time transaction iof ormation is generated on the screen by the CCTV computer. This visual image i s recorded onto videotape at the toll plaza and an example of the type of im age obtained is shown in F igure 7. The plaza c lock is displayed the overlay, allowing accu5


rate real time transaction eval uation. Staff at Leesburg Toll Plaza sent the videotapes to CUTR via 2-day mail. Plaza Computer Texr File The second data source is the Leesburg Plaza c omputer text file. An example of this text file is shown in TableBI Appendix B. Thi s file is the fmaJ trans action record of each day as collated by the plaza computer, and contains an extensive row of transaction data for each vehicle proceeding through the plaza. The Boca Raton Data Collection Center is responsible for collecting and processing transaction information from Leesburg Toll Plaza. Staff at Boca Raton Data Collec tion Center were responsible for identif ying the data required for th e evaluation and sending it to CUTR via e -mail for t11e duration of the evaluation Data Formatting and Evaluation Plan An exampl e of the tr.lnsaction record text files, complete with an explanation of the column headings, is shown in Table B I Appendix B Table B I shows that these text files detail each transaction recorded at Leesburg Plaza However, the files did not contain any independent data relating to the operation or output from the AS II device Therefore in order to assess the accu racy of ASH, the number of "forward a xles" allocated to each vehicle was considered. The text files were imported into Microsoft EXCEL Spreadsheet. formatted to isolate the required data (tile number of forward axles allocated t o each vehicle and time of transaction) and printed out for use in the ground truth comparison. There are four different axle allocations given to each transaction record Appendix B contains definitions of these four axle allocations The registered axl e count is the number of a xles alloca t ed to the vehicle by the toll collector, and the treadle ax l e count is the raw number of axles passing over the treadle. The forward and rev erse axle counts arc the number of axles passing forwards over the treadle and backwards over the treadle respec tively. Staff at Leesburg Toll Plaza advis ed the u s e of forward axles in the evaluation Ground Truth Comparison The evaluation commenced on May 26 1999. FDOTstaff at t he plaza were asked to provide video tape for all weekdays. from 6am to 9 pm, until further notice. Staff at Boca Raton Data Collection Center were asked to send daily output text fil es for the corresponding days. Durin g this stage of the evaluation analysts watched the video tape of the p l aza opera ti on and compared what they saw with the tnmsaction data registered on the CCTV overlay and in the text file. If a vehicle through t11e plaza and the transaction was corre ctly r eco r ded in t erms of transaction time and number of forward axles, the transaction was marked correct An example of a correctly recorded transaction sequence is shown in Figure 7 If the transaction record did not correspond with the ground truth video, the transaction was recorded as an error. Pretest analysis of the videotapes indicated that there were a significant variety of tmnsaction error types that occurred at the plaza. Careful analys i s of the various errors indicated that these errors fell into seven main categories The utili7.ation of these seven ca t egories provided the level of delin eation required to accurately define aU the errors that occurred at the plaza T able 2 details the error codes and their unique defini t ions.


--:---:---:::] ----------.Frame 1: A Pickup t ruck pulling a trai ler approaches the t oll plaza. It i s registe red by the toll collector as a threeaxl e vehicle. as s how n by the CJ-3 code in the "Current Vehicle" section of the overlay Frame2: The t oll colle cto r receives payment. The pickup truck and trailer leave the toll booth and the f ront of the vehicle i s detected by ASTI. The ve h icle passe s over th e tre adle and the three detected are displayed beneath th e "Fwd" and "Net" ax l e counters on the "Current Vehicle" sect i on of the o verla y Frame 3: The ve h icle leaves th e plaza a nd the e n d of the veh i c l e is detected by AS II. The transaction is complete and i s recorde d at 1 4. 55.06 as shown in the "Previous 2 Veh icles" sect io n of the o ve rlay. Figure 7. Exampk ofCo"ect R ecording


Error Type uru This en:or is defined as "Error in transaction timing only ". Tbis euor de finiti on was utilized to describe the situation when the t ime stamp allocated to a transaction was incorrect while all othe r details of the transaction were correct. The transaction time that should be allocated to each vehicle is the time at wbich the vehicle has just passed over the treadle and is leaving the plaza This error normally occurs when the transaction is recorded too late, long after t he vebicle has left the plaza If the transaction is recorded too early, other types of error are caused. Timiug errors are a compot\ent of several of the other error types as is explained below. Therefore this category describes the situation when only the timing aspect of the transaction i s incorrect. Table 3: C/assifi

attrib u table to any vehicle passing through the plaza. This error type ha s to be carefully differ entiatcd from a timing error ( Error Type "T') w hen th e transact i on does represent a veh i cle passing through the plaza. but the transaction is recorded at the incorrect time. Erro r Type "IV" This category describes the s ituation when a vehicle proceeds through the plaza but it i s not recorded in the transac t ion text file. Error Type ' This error was defined as "Wrong axle number aUocated to vehicle, not due to incorrect vehicle separation". T his err o r t ype was differ en t i ated from a separation error by consideting the trans a ct ions pre c eding and following the erro r transaction If the preceding and following vehicles were allocated the correct number of axles in the transaction record, then the error was log ged as an "E" type error. This differen t iates this error type f rom a separa t ion error (S) when add i tiona! axles would be allocated to either the preced i ng or following vehicle. Error Type "S" The only category that applied to the evaluations of ASII, as this category is related to incorrec t vehicle separation. This error type was defined as "Incorrect number of axles allocated to vehicle due to sepa ration error" This category was used when a separation error resul t ed in an incorr ect number of axles allocated to a vehicle. This etTor was characterized by one or more of a vehicle's axles being allocated to a following v ehicle or a nonexistent vehicle. Theref o re one separat ion error resulted in the i ncorrect alloca t ion of ax l es t o t wo consecu t ive vehicles. Since two vehicles were incorrectly recorded due to one separation error, only one error was recorded in these instances. AS II was respons ible for vehicle s eparation within the system and ther efore this was the only error type that could be potentially attributed to ASII. ANALYSIS OF RESULTS Th e ground tru th compari so n phase of the evaluati o n was conducted from May 26 to July 2. During this time a total of 28 268 vehicles were assessed. Table 3 be l ow shows the summa rized results of the e valuation G eneral Error Appraisal This table summarizes all the s ystem en:ors encountered during the evaluation. The "error rate per total vehicles verified" column s h ows that the rate of error occurrence has remained rela tively constant over t h e d u rat i on of the evaluation. Exami n ation of individ ual error categor ies shows t hat the occurrence rate of each error type also remains relat ively constant over the duratio n of the eva l uat ion


Tab! 4: Summary Results of AS/I Evaluatio11 .... l3 "' 1=: l3 ill !fl ill >4 "" 12 !:!! iii 5 d i:i:: ERRORS (j ii: ,_ ;;: !l: 3:i ;;: ii C'i C'i E'O !l! !l! 0 A w 1' L c E s "' S/25199 w 427 4 2 7 0 0 8 0 0 3 0 3 5!27/99 R SOH 1235 0 2 9 0 0 5 2 2 SIW99 F 2038 3273 3 0 61 118 1 15 1 10 618199 M 630 3903 1 1 1 167 0 2 0 2 9 619199 T 2172 6075 4 1 3H ? 0 8 3 6/10/99 w 1198 7273 0 1 5 1 0 l 6111/99 R 2615 9888 5 4 32 1 9 1 2 6/ 14/99 M 2360 12248 7 0 2 0 2 6 2 6/ 1 5/99 T 7 .98 1;104 6 s 0 5 0 1 6 1 2 6116199 w 1 37 1 1 4 4 1 7 6 0 19 3 0 7 1 2 6/18/99 )<' 1610 16027 5 1 22 0 3 1 4 6121/99 M 1 321 17348 3 4 7 0 4 I 6122199 T 848 18 1 96 0 0 0 l 6123199 w 1961 20157 1 0 2 14 2 13 2 2 6124199. R ... 1293 2:WO 4 2 IS 0. 9 1 2 6/25/99 F 802 22 2 52 4 0 17 0 6 2 4 6130199 w 1 934 24186 2 1 27 0 1 2 711/99 R 752 24938 1 2 9 0 2 2 7/2199 } 33;10 28268 27 15 1 0 9 3 65 7 SUM 28,268 8 7 36 43 344 11 165 2' 4 ( average) ERROR RATE(%) 0 3 1 0 1 3 1.5 1.22 0.04 1.58 0.0! 4 (sum) This is true with exception to the high occurrences of the "L'' type error resu l ting i n higher than average error ratios on May 28 and June 8. Aside from these two days, the e rror rate remains between 1 percent and 7 percent Discussion o f Indiv i dua l Erro r Types Each of the seven error types a r e unique and are discussed individually as follows: Error Type "T" Th i s was the most common error category with an incidence rate of 1.53%, and occurred regu larly throughout the evaluation. Timing errors occur frequently at the plaza for two main rea sons. Firstly, the responsibili t y for the transact i on timin g lies with the to ll collec t o r I f th e ;,rrr n Yf?t ::ln' n nnwtrssr 11


transaction is not recorded at the correct time, it is difficult for the system to proces s the trans action correctly This occurs because the s y s tem i s informed by the ASII unit and the treadle that a vehicle is passing, but the system has not been informed of this by the toll collector. The second source of timing et'for is the QNX Lane Controller which under certain circumstances is unable to proces s the transactions when they are recorded by the toll collector. This event is characterized by a star(*) symbol present be s ide the transaction denoting that the timing of the transaction is "on hold" until the QNX Lane Controller is able to proce s s it. T hi s type of problem is the reason for the high incidence of timing errors on May 28 and June 8, and can last for long continuous periods. Error Type W" This error type has a error frequ enc y rate of 0.13% and descr i be s the instances whe n a vehicle proceeds through the plaza but is not recorded by the system The Toll Collector is a major source of this type of el'fQr because during periods of inconsistent transaction recording, ve hicles can be missed completely by the toll collector. It is common for a vehicle to be m i ssed when it pas ses through the plaza and then added to the transaction record at a later time (see Error Type "A"). Error Type "A This error is defined as "Vehicle recorded in t ext file when no vehicle has and has an error rate of 0.31 % This type of error may be attributed to a number of sources Once again, the major source of this error was the inconsistent transaction proce s sing by the toll collector. As mentioned above, vehicle are often mi s sed by the toll collector (W errors) and then compen sated for slightly later wi th an extra vehicle resulting in an "A" error. Type "A" errors are also caused when a vehicle, or part of a vehicle, is missed by the toll collector resulting in treadle axles unaccounted for in the sy s tem. T hese "residual" axles need to be discharged before the next transaction can be processed and the refore they are discharged a vehicle, often with one axle, which has not actua ll y pas s ed. An exam ple of this type of occurrence is shown in Figure 8. Detailed review of this error type has s hown that in the errOl' code (U0-8 9) is present in the transaction record for the m ajority of "A" error transactions. This code is recorded when the toll collector does not complete t he transaction, which adds more evidence to the likelihood of t he toll collector being the source of this error type. Error Type "C" This was a highly unusual occurrence with an el'fOr rate over the evaluation period of 0.04 pecent. The reason that this rarely occuned is because the plaza computer usually rectifies et'fors made by the CCTV computer. This error type represents the unusual e v ent of the CCTV computer recording the tran s action correctly which was then rendered i ncorrect by t he action of the plala computer


Error Type "E" The frequency of this erro r type over the study duration was found to be 0.58 percent. Potential sources of error type are the treadle not counting the axles correctly, the toll collector recording the transaction at the wrong time, and the QNX lane controller not collating t he information correctly As w i th the Type "A" errors, an error code (U0-89) was present ill the transaction report for the majority of these error occurrences. This suggests that this error type was attributab l e to a system e rror Error Type "S" This error type re lated to inconect vehic l e separation and therefore was only error that could be attributed to ASD. Th i s error occurred twentythree (23) t imes over the course of th e evalua t i on giving an e JTOJ rate of0. 08 percent. This illdicated that the accuracy of ASII was no lower than 99.92 percent. Each of the twentythree (23) errors are detailed in TableC-1, Appendix C. In eac h case, the table shows the transaction sequence from the vehicle preceding the i n cor rectly separated vehicles to the vehicle following the iacon-ectl y s eparated vehicles. This was done to sho w that the total number of ax.le s over the transaction per i od was correct and hence the i nconect axl e allocation must be due to a separalio n error and not a Type "E" error From Table 4 it is apparent that the majority of the separation errors have similar characteris tics. Ei g htee n (18) o f the twenty-three (23) err ors occur when two consecutive two-axle ver h icles were recorded as a one-axl e vehic l e followed by a three-axle vehicle. The likely expla nati on for these 18 errors i s a time delay in lan e contro ll er processing of the treadl e Of the other five erro rs, two involve incorrect separat ion o f a five axle truck and two involve i n correct separation of a vehicle pulling a trailer Figures 8 and 9 each show a sequen c e o f video frame images detailing separation e rmr occurrences F i g u re 8 d i s p la y s the sequence of events resulting in the incorr ect s eparation of a five-axle truck. I n analyzing this sequence it is clear that this error was caused by the transaction being recorded before all five axles had passed over the treadle. The remaining four axles that ran o ver the treadle after the vehicle was recorded were discharged from the system as an extra four-axle veh i cle Figure 9 shows the seq u ence of sti.lls for the incorr ect separation of a van towing a trai ler. The van was recorded as a two-axle vehicle b efore the trail er axle had passed over the treadl e F i gures 8 and 9 were included to show sequence of e v ents that lead to a separation error These figure s s how that separation er rors were caused when the vehicle was recorded before it has f ully passed over the treadle. TI1e t e n files from the Da t a Center con s isted of inform atio n from ASII, th. e QNX La n e Controller's int e rpreted output from the ASII uni t, and the toll collect or s manual inp u t o f on l y count (classification) for each v ehicle Due t o the f ully integra ted nature of the ASH configuration, the evalua t ion was unable to pinpoint the source of these separation errors Potential error sources ar e the toll collector, the QNX Lane Controller, _ .....,...,....., ______ ..... .... __ 13 ee:rs JIIi:tCr t at:r b ; t'r sfs, m c


r -----.. . ------... ---II --. -----------l ------___ ___j and ASH. Therefore the accuracy as s essrncm of ASH at 99.92% is a worst case scenari o Summ ary of Findings Over the course of the evaluation period, a t otal of 28,268 vehicle transactions were assess ed. There were a total of twenty three ( 23) s ystem separation errors during this period. Eighteen ( 18 ) of th e twenty-three (23) errors occurred when two consecutive two axle ve hicle s were recorded as a one ax l e ve h icle followed by a three axle vehicle. The text tiles output had thre e potential sou r ces of vehicle s epara t ion errors, th e timing and sequ ence of the toll collector's manual input of axle count (classification) for each ve h i cle, t h e QNX Lane Controller's interpretation of the output from the ASH unit, and ASH. The like l y explana t ion for t he 18 of the 23 errors is a t im e delay in lane contro l ler proces s ing of the treadle inputS. Therefore the "lower bowul" accura c y assessment of ASII is 99.92 percent and a n u ppe r bound acc u racy assessmen t is 99.98 percent


Frame 1: Vehicle App1 oaches Toll Plaza Lane #5 is third from the left Truck ap proaches plaza and i s regist ered as a 5 ax l e vehicle by the toll collector (as shown by the Cl-5 code beneath the "Current Vehicle" h eading). Frame 2: First Axle passes ove r Treadle Payment is received by the toll collector and the tmck begins to moveoutofthe plaza. The truck's first axle passes over the treadle, as is shown in the "Fwd" and "Net" ax l e col u mns for the current vehic l e. Frame 3: 'D:ansact ion is Recorded The transaction is recorded, and is now dis played under the "Previous 2 Vehicles" head ing Only one axle has passed over the treadle and therefore the vehicle is recorded as a one axle vehicle. }'r-ame 4: Remaining Four Axles pass oYe r Treadle The truck's remaining four axles pass ove t the tt eadle as it exits the plaza F'rame 5: Remaining Four Axles Re cor d ed The four axles remain in the "Current Vehicle" sectio n until the next vehicle arrives at the pl aza. At this time the four axles are discharged as another vehicle. Figure 8 -Transaction Sequence for lnc"Orreet Separation of Five-Axle Truck


Frame 1: Van towing traile r approaches toll booth and i s classified by the toll collector a three-a xle vehicle Frame 2: The two axles o f the van pa ss over the treadle and ar e display e d on t he over lay under the "Fwd' and uNet" axle. columns for the current v ehicl e. Frame3: The van is r e co rde d as a two axle v e hicle in the transaction r ecord Shonly afterwards the t railer axle pa ss e s over the treadle a n d is d i splayed in the "Cut rent Vehicle" sectio n Frame4: As the next vehicle leaves the t oll b ooth and p asses over the treadle, t he trailer axle i s discharged f r om the "Current Vehicle" section and r e corded as a one axle veh icle. Figure 9. Incorrect Separation of a Vall Towi ng tl Trailer -16


APPENDIX A ASII Product Information


Company History The company was founded in 1983 as a F l orida Corporation by William Schwartz and Richard Wangler to p er form research, development and production of laser systems. Au iuitial private stock offering was held in 1984 to provide w orking capi t al for the company through the sale of common stock. The initia l products were laser weapM simulation systems, which were pro duced under governmental and commercial contn\cts for militar:y and law euvorcement train i .ng. Some commercia weapoos training products were licensed from Litt on Laser Systems. In 198 5, a Research Division was formed in Concord, MAunder the direction of Dr. Peter Moulton, a laser scientist from MIT's Lincoln Laboratory. Th i s Division was established to conduct research and development on new solid state laser matexials and systems and was successful in becoming largely financially self-supporting through Government resemch and development contracts. In 1987, the Solid State Lasers Divis io n was formed to manu f acture products initially d evel oped by the Research Division. Later the Division concentrated on products for laser medicine and in April of 1998, th e Division was sold to LaserSigbt in Orlando. In 1990, a new Sensors Division was spun off from the original Weapons Simulation activities thus creating two divi sion s, Weapons Simulation and Sensors. Recently, the Sensors Division was divided into Com mercial Sensors, to produce standard commercail systems and Aerospace Systems to conduct reserach on more develop m ental and/or c ustom devices. Th e two Divisions produce a vari et y of laser sensor systems based upon semiconductor and solid state laser rangefinder technology. The Weapons Simulation Division was recently renamed to 11-aining Systems in order to better describe the Divisio n s activies. In July of 1998 t .he Research Division was spun off as a separate company, Q-Peak, Inc. to manufacture diode-pumped solid state lasers and continue research and development of advanced solid state laser systems. Related Autoscn.e Projects Highway 407 (Toronto, Canada) 320 Autosenve II" units have been install ed to dat e to perform vehicle detection/classification and serve as a "trigger" for a video enforcement sysbsystem for this highly automated Open Toll Road. Deliveries of A111osense IF unit s were initiated 5/96. The prime contractor is Raytheon HTMS. Interstate 4 (_1\fetropolitan Orlando Area) Two Autosense ll" units were installed in Autust 1999 as part of a Trafficsense"' system for the University of Centn\l F lorida, School o f Engineering The Autosens e liP u nits are provid ing management data (lane occupancy headway, speed and vehicle c las s i f ication). Two additional units will be installed in 2000 to provide travel time data for Interstate 4 lanes through the dowtttown Orlando area. Traffic sense"' allows University researchers to access data from this site via the Internet.


Florida Thrnpike Beginning in 1998 and to date 84 Auto sense !,. units have been delivered to the Florida Depart ment of Transportation to function as vehicle separators at Turnpike toll lanes and other toll facilities Highway E470 (Denver, Colorado) 16 Autosense IJ> sensors arc performing a post classification function in toll lanes leading to the new airport in Denver Colorado. Installati on of these units be gan in 1998. The prime contrac t or is MFS TransTech. SR-91 (Southern California) 135 Autosense I" sens ors were installed in July, 1994. The prim e contraccor is MFS Trans Tech. Interstate 15 HOV Lane (Southern California) 2 Autosense IJ> units were installed in October, 1997. The prime contractor is SAl C. Bronx-Queens Expressway (New York) A total of 19 Autosense I F sensors installed (June/July, 1999) along the Bronx-Queens Ex pres sway and are supp<>rting traffic management functions with vehicle classification safety in several locations. Marbelitc was the prime contractor. Anmin 1\mnel (South Korea) 12 Autosmse II" units were installed i n 1999 to perform pre-classification function s in toll lanes leading into this highway tunnel. Prime Contractor is Poscon. lhwaryung Thnnel (South Korea) Same as the Anmin Tunnel. Autostrade, Ital y Autostrade Telecomunicazion i S.P.A. began ins tallingAutosense II" units in April, 1998 in the vicinity of Florence, Italy at several locations along the Autostrade to perform vehicle classifi cations functions To date 21 units ha,, e been in s talled.


Test Sites { existing and/or planned for CY 99) Flori .da Thrupike Leesburg Plaza -Autosense II' tested as a vehicle separatOI Illinois Toll Roa d (Sept. 99) M e tropo li tan Chicago area -Autose n se II A'' (axle c oun ter)" wi.ll b e tested in Express "I Pass" lanes New York Thruway South o f Albany Autosense liP be ing t est ed for vehicle classification o n open road as triggers for license p l a te e nfo rc ement cameras. Autosen:.e llA" (axle c ounte r ) will be tested in Sept. 99 for post classif ication function in a toll l a n e Orlando-Orange County Expressway (SeptJOct 99) Alliosense IIA$ ( a xl e co unter) will be tested in a toll lane. Highway 441 (Orlando) Schwartz Electro Optics In c. has a gantry over the two south bound lanes of Highway 441 (Nortb Orange B lossom Train ) i n Orlando Currem ly, five Autosense" units are moun ted in a variety of geometries on the gantry. This facility provides real-world tra ffic for the d eve lop m ent of algorithms employed in the Autosmse sys t ems Trondhim, Norway Q-Free ASA h as gantry over a p ubli c highway on which they have mounted twoAuto sense I P and twoAutosense liP units to perform vehicle classification f u nct ions fo r tolling a ppli cati o n s Rijden Test Project Autosense IP and Ill units ha v e been installed si nc e March, 1998 su p porti ng thi s Dutch Test. The prime c on tra c tors te sting Autoseose units include Phillips/B o sch a n d Alcate l. Unit Cos t Autose11se pricing is sensitive t o quan t i ties orde red Tile price for an Autosense II" unit i n qua n tities of 1 0 0 is $5,500.


Insbllation Time The priociple faclOr in instaHation cost is a f unction of the availabili ty of a s tructure upon which to mount an Autosense'" unit either above a traffic lane or adjacent to it. T he acrual process of attaching a mounting plate to another structure, running communications and power cables and inte r facing to a roadside computer/controller nominally takes one to two hours de pending upon the skill and experience of the in stall ers Maip!enance The M ean Time To Repair (MTTR) for Autosense" units is nominally 15 minutes. This repre sents the tim e to disconnect an Amosense unit from a mounting assembly and inswU a new unit. Software "fixes" can be performed without removing the unit from its installation posi tion. Repair costs for a damaged Autosens e unit is a f unction of the extent of damage.


' APPENDIX B Performance Comparison Between ASII and /ORIS>


, Separation Accuracy 99.92%-99 .98%3 99.89%4 Unit Cost s $5,911' $3,600Lane (separation only) Installation Time One-two Hours 112 Day/Laue Maintenance Cost Cleaniog Cost 6 < 500/year Installation Costs Labor for One-two Hours ApproJt. $1-2,000/Lane i l U n limited lifespan. Mean Time Between Maintenance Che c k required by 25,000 hours Failures SEQ every five years 1 SourceSchwartz ElectroO ptics, Orlando 2. Sour<:e -Peek Traffic Sarasota 3. Tests conducted a t Leesburg Plaza and evaluated by CUTR.. A total of 28.268 vehicle transactions were assessed w ith 23 separation e rrors during this period. The l ikely explanat i on fro 1 8 of the 23 errors i s a delay i n lane controller of the u eadl e inputs. 4. Phase II !eSts conducted at Leesburg Plaza by CUTR. A total of 45,228 vehicles were ver ified w ith 49 ldris errors 5. The p rice for an ASII U nit in quantit i es of 100 is $5,600. The o t her items used i n the i nstallat i on of the ASII are mounting plate and communicationcabcls, the price for these quantities i n 100 is $3 1 1 6. The window through which the laser is projected should be cleaned w it h a cloth 2-3 times per year as required.


APPENDIX C P laza Computer Text Fil e Output of Lane #5, Leesburg Toll Plaza Example of data sent to CUTR from Boca Raton Data Collection Center


Table C-1: Example of text file sent from Bocll RaJon DaJil CoT/ection _MSG SEQ PlAZA l.ANE EMPL LOCAL TRANS RG TR M M REG AVC FREE FIV RV _ID NUM CODE NUM JD YE.>\R DAY TIME NUM AX AX F CARD CODE E CASH REV UO CNT 'f'R TR ---------: ------9000 9 537 003100 05M 8174 1999 183 164330 4056 O"f 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9538 003100 05M 8174 1999 183 164337 4057 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9539 003100 OSM 8174 1999 183 164345 4058 O:Z 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9540 003100 05M 8174 1999 183 164401 4059 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9541 003100 05M 8174 1999 183 164409 4060 02 00 0 000000000000000 0 0200 0000 89 0000 01 1!0 9000 9542 003100 OSM 8174 1999 183 164416 4061 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9543 003100 05M 8174 1999 183 164421 4062 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9544 003100 05M 8174 1999 183 164443 4063 05 OS 1 300116800192016 0 0800 0800 00 0000 05 00 _9000 9545 003100 05M 8174 1999 183 164452 4 064 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9546 003100 OSM 8174 1999 183 164458 4065 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9547 003100 05M 8174 1999 183 164506 4066 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9548 003100 05M 8174 1999 183 164520 4067 02 02 0 000000000000000 0 0200 0200 89 0000 02 00 _9000 9 549 003100 OSM 8174 1999 183 164524 4068 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9550 003100 05M 8174 1999 183 164544 4 069 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9551 003100 05M 8174 1999 183 164553 4070 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 _9000 9552 003100 05M 8174 1999 183 164608 4071 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 _9000 9553 003100 05M 8174 1999 183 164637 4072 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9554 003100 05M. 8174 1999 183 !64655 4073 03 03 0 000000000000000 0 0400 0400 89 0000 03 00 9000 9555 003100 05M 8174 1999 183 164702 4074 Op 03 0 000000000000000 0 0200 0400 00 0000 03 00 _9000 9556 003100 05M 8174 1999 183 164707 4075 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9557 003100 05M 8174 1999 183 164732 4076 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9558 003100 05M 8174 1999 183 164745 4077 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9559 003100 05M 8174 1999 /83 164802 4078 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 _9000 9560 003100 05M 8174 1999 183 164809 4079 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9561 003100 05M 8174 1999 183 164824 4080 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9562 003100 05M 8174 1999 183 164841 4081 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 _9000 9563 003100 05M 8174 1999 183 164900 4082 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 _9000 9564 003100 05M 8174 1999 183 164912 4083 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9565 003100 05M 8174 1999 183 164918 4084 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 9000 9566 003100 05M 8174 1999 183 164926 4085 02 02 0 000000000000000 0 0200 0200 00 0000 02 00 _9000 9567 003100 05M 8174 1999 183 164945 4086 02 02 0 000000000000000 0 0200 0200 00 0000 02 00


Description (lj C(l/.mrm Headings MSO_IP: This is t h e m essage identification code. A 9000 code represents a ''Fa r e transaction SEQNUM: This shows t he se que nce n umber given to the transaction PLAZA CODE: This code denotes t h e l oc ati on of t h e .uansact i on. Code 3100 d enotes a mainline tollbooth LANENUM : This column disp lays 1he lane number. The evalumion.f o cussed on La11t #5. EMPLID : This column. di.spla)'S the unique code of the toll collector ill charge of the u ansactio n YEAR: Year in which the trmrsaction took place DAY: Day on which the transactio11 took plt1ce. Days are measured/rom Jan }. LOCAL TIME : Time when the transactiolf was recorded. The tim e is read as two digits ttacltfor hot4rs, mimues and :oeconds; i.e s s TRANSNUM: Transaction number seque n ce RGAX: Registered Axles. The 1111mber of axl.e.'i allocmed to vehicle as deumnined by the toll collector. TRAX: Treadle Axles. The coum of the number of axles a s derermitled by treudle MP: Metluid of Paymem. t\ pa)'lmmt i s given a o tntr)\ CARD CODE: If a prt!paid t?ccoun t is u.tedfor paymtll(, t he account munber i s gi t:en in this column. ME: if a prepaid accoulll was used differen liatts benw.-en slot reader elltf) and manual entry. REG CASH : Registered cash E i the r the computed toll due based on the toll col/eclor classificati o n o r the a ct ual <'OIIected cash aJ registered by t he toll t:r AVCREV: Computed toll based on the A VC classification of the vehicle UO: Code used to des c ribe unusual occurret1ces aJsocimed with thi:1 particular tr

Appendix D Separation Error S u m mary


ERROR DAT E TIME # 1 5/27199 13.08.38 13.08.52 13.09.04 1 3 .09.16 2 5127/99 15.20.51 15.21.27 15.21.59 15.22.06 3 5/28199 14.25.26 14.25.33 14.25.58 14.26.02 4 619199 09.19.14 09.1 9 .38 09.1 9 .43 09.20.32 5 6/9/99 17.24.57 17.25.45 17.25.46 17.26.11 1 3 .16.24 6 6111/99 1 3 16.32 13.16.50 13.17.20 7 6114199 09.37.16 09.37.53 09.38.53 09.39.19 8 6114199 19.50.02 19.50.15 1 9 .50.20 19.50.30 9 6/15199 07.38.01 07.39. 1 1 07.39.42 07.40.01 07.40 .02 10 6/16/99 10.43.10 10. 43.29 10.43 .55 10.44.04 11 6/18199 14.47.17 14.48.23 14.48.28 14.48 .32 12 6123/99 19.05.24 19.05.25 19.05.31 19.06.00 1 3 6/23,199 20.14.59 20.16.00 20.16.32 20.16.38 Table D l Separation Error Summary FORWARD OBSERVED AXLES AXLES 2 2 1 2 3 2 2 2 2 2 1 2 3 2 2 2 2 2 1 2 3 2 1 2 2 2 1 2 3 2 1 2 5 5 1 2 3 2 2 2 2 2 1 2 3 2 2 2 5 5 I 5 4 0 5 5 2 2 1 2 3 2 2 2 2 2 1 2 2 2 1 2 2 2 2 1 2 3 2 2 2 2 2 2 3 1 2 2 2 2 1 2 3 2 2 2 2 2 1 5 4 2 2 ERROR VEHICLE CODE TYPE Minivan 0.84 P icku& truck . car Van U 0-84 truck ar 2axle truck Car U 0-8 4 Car Van Car Car U0-84 C ar Car Car saxe l truck UO.B4 suv suv Car Car C ar UO.B4 Van Car . 5 axle truck 5 axle truck UO.B4 no vehicle U0 5 axle truck Car suv UOB4 Min ivan M inivan Van U0 Bus P ickup truck U0-89 car Car U0-84 Pickup truck car Car Van Van w ith trailer U0-!14 C ar suv U0-84 Car C a r Car Car U().B4 5axle truck no vehicle vehicle


TableD 1 (Continued) OATt; TJMt; M A Yf "" AXL t4.t 9 59 2 1 4 6124/99 14.21. 04 1 U0-84 14.21. 1 8 3 14.21. 31 2 09.43. 24 2 2 Car 1 5 6125199 09.43. 39 1 2 U0-84 Van 09.43. 53 3 2 Car 09.44. 35 2 2 Car 10.01. 24 2 2 truck t 6 6125199 t0 0 1 47 1 2 U0-84 10.ot. 48 3 2 10.02. 14 2 2 06.t 8 1 3 5 5 5axle truck t 7 7/t/99 06.t 8 35 1 2 U0 suv 06.t8. 51 3 2 Car 06.19. 09 2 2 Car 09.08. 24 2 2 Car 1 8 7/1/99 09.09. 14 1 3 U0-84 Van with trailer 09.09. 14 2 . No vehicle 09.09. 17 2 2 Van 1 9 7 /2199 U0-84 :ar t4 .19. 50 2 2 Car 20 7/2199 t 4 .20. 02 t 2 U0-84 Car 14.20. 08 3 2 Motorcycle 14.20. 1 4 2 2 suv t5 5 t 58 2 2 Car 712199 t5.52. 02 1 2 U0-84 Car 2t t 5 .52. 08 3 2 g:; t5 .52. 1 6 2 2 16.40. 1 4 2 2 suv 7/2199 t6.40. 25 1 2 U0-84 Car 22 16.40. 29 3 2 Car t6.40. 37 2 2 Car U0-84 23 7/2199 3 ar 2 P ickup t ruck


Center for Urban Transportation Research College of Engi neering University of South F l orida 4202 E. Fow ler Avenue, CUT 100 T ampa, Flori da 33620-5375 (813) 974-3120, fax (813) 974 -5168 http://www eutr

xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader cam 2200229Ia 4500
controlfield tag 001 001926049
005 20080327151253.0
008 001018s2000 flua t 000 0 eng d
datafield ind1 7 ind2 024
subfield code a C01-00424
2 local
b F54 2000
0 245
Field performance evaluation of Autosense II at Leesburg mainline toll plaza /
prepared by Center for Urban Transportation Research, University of South Florida, College of Engineering.
[Tampa, Fla. :
University of South Florida, Center for Urban Transportation Research],
iii, 16 p. :
ill. ;
28 cm.
"February 2000."
Also available online.
Traffic flow
z Florida
x Equipment and supplies
Imaging systems
University of South Florida.
Center for Urban Transportation Research.
1 8 773
t Center for Urban Transportation Research Publications [USF].
4 856