1354 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 61, NO. 5, MAY 2012Fig. 2. Pictorial representation of the relative change in inductance for the inductive loops A, B, and C. The continuous curve in the plot shows the relativechange in inductance when a large metallic object (such as a bus) moves from left to right at a vertical distance of about 60 cm from the plane of the loop. Thedotted curve shows the change in inductance when a small metallic object (e.g., bicycle) moves from left to right. In this case, the object is moving at a distanceof about 8 cm above the plane of the inductive loop.Fig. 3. New inductive loop suitable to sense small as well as large vehicles.The present work proposes a new and simple inductive loopsensor structure  that senses both large- and small-sizedvehicles and differentiates the large one from the small. Thesensor provides a unique output signature for each type ofvehicle. The sensor output information is such that it is not onlypossible to detect and classify the vehicles in an unstructuredtrafﬁc system but also possible to compute individual vehi-cle’s speed and occupancy time. Details of the new inductiveloop structure, principle of measurement, prototype systemdeveloped, experimental set-up, and results of ﬁeld tests arediscussed in the following sections of the paper.II. NEW INDUCTIVE LOOP SENSORFig. 2 shows the schematic diagrams of three possible in-ductive loops for vehicle detection. The coil structure witha large area indicated as “Loop-A” in Fig. 2 is the one inuse for lane-based homogeneous trafﬁc and is well suited todetect large vehicles such as car or bus or truck. When such alarge vehicle goes over the loop, the change in the inductanceLP of the loop will be as indicated by the curve (continuousline) drawn on top of the Loop-A section in Fig. 2. However,if a small vehicle such as a bicycle goes over this loop, thechange in the inductance of the loop will be small, and itis detectable only when the object is directly above the coilposition, as indicated by the curve drawn with a dotted linein Fig. 2, above Loop-A. In all other positions, the resultantchange in the inductance of Loop-A will be negligible. Onthe other hand, for a coil structure as indicated in “Loop-B,”with a smaller cross-sectional area compared with Loop-A,the relative change in the coil inductance (shown by the curvedrawn in dotted line) will be appreciable when a small vehicleapproaches, close to the plane (vertically) of the loop. However,the relative change in inductance (indicated by the continuousline, on top of Loop-B) will be small for a large vehicle (bus ortruck) moving over Loop-B. Thus, Loop-A is mainly sensitiveto large vehicles, and Loop-B is sensitive to small vehicles thatgo very close (i.e., vertical distance from the loop plane) to theloop.The proposed loop, “Loop-C” in Fig. 2, is formed using acontinuous conductor wound as illustrated (shown more clearlyin Fig. 3) to form an outer loop and an inner loop. The loopsare formed such that a current ﬂow through the coil produces amagnetic ﬂux in the outer loop to be in line and aiding the ﬂuxproduced by the inner loop, at the center of the loop. When alarge vehicle such as a bus moves over Loop-C, the loop willgive a relative change in inductance similar to Loop-A, andwhen a small object such as a bicycle goes over the loop, it givesa relative change in inductance similar to Loop-B. Thus, a largevehicle (e.g., bus) and a smaller vehicle close to the loop (e.g.,
SHEIK MOHAMMED ALI et al.: VEHICLE DETECTION SYSTEM FOR HETEROGENEOUS AND LANE-LESS TRAFFIC 1355Fig. 4. Multiple inductive loop system. A system with ﬁve loops per lane isillustrated.bicycle) can be reliably detected with the proposed Loop-Cshown in Fig. 2. A detailed 3-D view of the proposed loop isshown in Fig. 3. The loop coil can be placed below or on thesurface of the road. The loop is connected to the measurementsystem as indicated in Fig. 3. Capacitances C1, C2, and CMalong with inductance LP of the loop coil form a resonantcircuit. This circuit passes the signal at the resonant frequencyto the measurement system and attenuates all the unwantedfrequency components that may be picked up by the loop. Themeasurement system and the loop with inductance LP are con-nected together using a twisted-pair cable. The voltage acrosscapacitor CM is given to a Data Acquisition System (DAS).The data acquired by the DAS is sent to a computer and asuitable algorithm, implemented in the computer using a virtualinstrumentation environment, detects the type and speed ofthe vehicles, and counts the number of vehicles being sensed.A. Multiple Loop Detector SystemFor detecting vehicles that ﬂow in an unorganized fashionwith no-lane discipline, multiple loops having the structure asindicated in Loop-C (refer to Figs. 2 and 3) are placed on theroad side by side, as indicated in Fig. 4. Unlike the presentsystem of having one loop per lane, many loops as practicallypossible are placed in a lane. Consider, there are N such loopsplaced covering the road width, resulting in N-inductances.Fig. 5 shows the equivalent circuit representation of the pro-posed multiple inductive loop system. L1, L2, . . . , LN are theinductances of individual loops. Since L1, L2, . . . , LN are ofthe structure Loop-C, each of the coil is sensitive to both small(e.g., a bicycle) and large vehicles (e.g., a bus). Thus, use ofN loops as indicated in Fig. 5 facilitates the sensing of Nindividual bicycles at a time. If a large vehicle such as a busgoes over the loops, more than one loop will have noticeablechange in inductance simultaneously, thus enabling the sensingof a large vehicle as well. If we consider that the wheels of a busgo over the outer loops L1 and L3, the change in inductancefor L1 and L3 will produce similar signature for the voltagesVC1 and VC3, which will be different from the signature of themiddle loop L2. These signatures and the magnitude of changein inductance are used to detect the presence of a moving busaccurately. The signature of the signals for the aforementionedcondition will be different from that of, for example, threebicycles running in parallel. Hence, such conditions also canbe distinguished.Fig. 5. Equivalent circuit of multiple loop system. Voltage signalsVC1, VC2 . . . , VCN are continuously recorded using a data acquisition unit.The signal conditioning part (introducing resonance) of themeasurement system used for the multiple loop system issimilar to the one employed for a single loop system, as shownin Fig. 3. Individual inductance values of the loops, in a vacantcondition, will be nearly equal to each other as the dimensionsand number of turns used to construct each loop are identical.The voltages VC1, VC2, . . . , VCN are given to the DAS. Thesevoltages change as a function of the change in inductanceof the corresponding loop when a vehicle goes over it. Thesignal acquired by the DAS is ﬁltered, and RMS values ofeach signal are computed and displayed. In the multiple loopsystem, the number of channels used to acquire the data is equalto the number of loops used. In order to avoid the problemsowing to the cross-sensitivity between the loops, at a time, onlythe loops with odd or even numbers are energized and theirresponses measured. A minimum distance of few centimetersis kept between the adjacent loops for ease of separation duringinstallation. Switches S1, S3, S5, etc. are controlled by signalVS1, whereas S2, S4, S6, etc. are regulated using control signalVS2. The loops with switches in position 1 will be at resonanceand work normal and give the expected output. For the circuitin Fig. 5, the switches can be set to position 0 or 1 using theswitch control signals shown in Fig. 6. If required, each loopcan be energized and measured individually by operating thecorresponding switch, and this can be preformed in a sequentialmanner for the loops at a very high rate so that measurementfrom each loop is completed in time to capture the data for ve-hicles moving fast. In such a case, the switch positions (0 or 1)
1356 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 61, NO. 5, MAY 2012Fig. 6. Switch control signals. Switches with odd numbers are controlledusing VS1, whereas that with even numbers are regulated using VS2.Fig. 7. Equivalent circuit of multiple loop system with improved switchingarrangement.of S1, S2, . . . , SN are controlled using individual control volt-ages VS1, VS2, . . . , VSN , respectively.A high quality factor (Q) resonant circuit is preferred foreach loop in Fig. 5 for the best performance. However, thepractical switches S1 to SN have its own “ON” resistances,and hence, the Q of the circuit and the sensitivity gets limited.This issue can be overcome with a modiﬁed circuit in Fig. 5 asgiven in Fig. 7. In this circuit, when a loop, say ith loop, needsto be deactivated (i.e., based on odd or even loop number orin a sequential basis, as explained in the above paragraph), anadditional low impedance ZB (e.g., a resistor RB) is connectedin parallel to the corresponding CM , by setting the ith switch Sito position 0. In this condition, the loop “i” will be detuned fromresonance, resulting in a negligible change in the correspondingloop output signal VCi when a vehicle approaches. The loopsFig. 8. Snap shot of the experimental set-up. One of the inductive loops isshown in the inset. The loops are stitched to the carpet material and placed inthe roadways for testing. The loops touch the road surface, and the carpet is ontop of the loops.with switches in position 1 will be at resonance and operatenormal and give the expected output. For the circuit in Fig. 7also, the switches can be set to position 0 or 1 using the switchcontrol signal shown in Fig. 6.A prototype system of the proposed scheme has been builtand tested. Details of the prototype developed and test resultsare discussed in the next section.III. EXPERIMENTAL SET-UP AND RESULTSIn order to test the practicality of the proposed scheme,a prototype multiple loop detection system with six loops(N = 6) has been designed, built, and tested. A snap shot ofthe experimental set-up, developed and employed in the ﬁeldtesting, is shown in Fig. 8. Each loop has ﬁve turns and anominal inductance of 100 µH. The dimensions of the singleloop are given in Fig. 9. The signal source VS was realized usinga variable frequency sinusoidal source with 10 V peak-to-peakamplitude followed by a power operational ampliﬁer using theIC OPA541. The capacitance values used are: C1 = 0.056 µFand CM = 0.068 µF. This results in a resonance frequency ofabout 65 kHz, and the frequency of excitation for the systemwas chosen to be the same.The switches were realized using ICs MAX4053. The outputvoltages VC1, VC2, . . . , VC6 were acquired using a 16-bit DASUSB 6216, manufactured by National Instruments . Thesignals were sampled at a rate of 400 kSa/s. A test has beenconducted to obtain a sensitivity map of the new loop. In orderto perform this, a cylindrical conductive (mild-steel) test objectwith a diameter of 10 cm and a height of 2 cm was selected.Then, a sheet of paper with suitable size was taken, and alength of 100 cm and a width of 60 cm were marked. This wasthen divided into 60 individual boxes by drawing lines for each10 cm lengthwise and widthwise in the paper. This paper wasthen kept on the loop-1 with both of its centers coincidedand the line in the paper in lengthwise was in parallel to thedirection of loop length. The object was kept in each box (oneat a time) marked in the paper at a height of 18 cm, andcorresponding readings were recorded. The recorded data hasbeen normalized with the maximum output observed among
SHEIK MOHAMMED ALI et al.: VEHICLE DETECTION SYSTEM FOR HETEROGENEOUS AND LANE-LESS TRAFFIC 1357Fig. 9. Dimensions of the outer (width, D1 = 55 cm and length, D2 =80 cm) and inner loops (length, D3 = 20 cm). The number of turns is equalfor both the loops. The inner and outer loops are in series.Fig. 10. Sensitivity map of the new inductive loop. Experiment is performedusing a cylindrical metallic object with a diameter of 10 cm and a height of2 cm. The object was placed at a height of about 18 cm from the plane of theloop.those positions (boxes) and plotted to obtain the sensitivity mapof the loop, as given in Fig. 10. It can be seen from the map(refer to Fig. 10) that the loop has good sensitivity to even asmall object not only at the outer periphery (as can be expectedfor a loop such as Loop-A in Fig. 2) but also in the inner regionsdue to the special structure of the loop with the inner loop.The developed system was then tested in two stages. In the ﬁrststage, a single loop (L1) was selected and output correspondingto various types of vehicles was recorded and compared. Therecorded data for a bicycle, motor cycle, car, and a bus aregiven in Fig. 11. Vehicles moving at various speeds were alsotested, and the signatures were compared. In Fig. 11, it is shownFig. 11. Results from single loop experiment. This shows the output signal(signature) observed for different types of vehicles.that each type of vehicle gives a unique signature output, whichmakes detection as well as segregation easy. Fig. 11 also showsthat the loop detects small vehicles such as bicycle as well aslarge vehicles (bus), as expected.Similar experiments were carried out with the multiple loopdetector, and the outputs from all the six loops were recordedand analyzed simultaneously. Fig. 12 shows the output when amotor cycle and a bicycle run in parallel. The motor cycle goesover loop-1 while the bicycle goes over loop-3 and no vehicleover other loops. Fig. 13 shows the results obtained for a busthat was covering the loops L1 to L3. Fig. 14 shows the testresults obtained from all the six loops simultaneously while var-ious types of vehicles such as bus, car, motor cycle, scooter, etc.were passing through the road in a heterogeneous and no-lanemanner. The signatures produced by each vehicle are shown inFig. 14. A snap shot of the front panel of the virtual instrumentdeveloped to perform the measurements, to process the data,and display the results such as type and number of the vehiclesdetected is shown in Fig. 15. The speed of the vehicle can beaccurately obtained by using double-loop detector system .The speed of the vehicle can be also computed from the datalength for which a detectable change in output is observed, asthe length of the vehicle (as the system detects the type) isknown. From the output signal, the calculated speed and thelength of the vehicle, the time of occupancy for that vehicleis ascertained. It is found from the above experiments that allimportant types of vehicles that go through the roadways canbe detected and distinguished.
1358 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 61, NO. 5, MAY 2012Fig. 12. Results from the multiple loop detector. In this case, a motor bikeand a bicycle were running in parallel. The motorbike went over loop-1 (L1)while the bicycle went over loop-3 (L3) and no vehicle over other loops, e.g.,loop-2 (L2).Fig. 13. Signature of bus obtained from the multiple loop vehicledetector.Fig. 14. Results from the multiple loop detector with six loops (N = 6)recorded when various types of vehicles were simultaneously moving on theroad.Fig. 15. Front panel of the virtual instrument developed to acquire data(output signals from each loop), process it for detection and counting ofvehicles.IV. DISCUSSIONThis section discusses the possible conditions of faulty de-tection by the multiple inductive loop system. Let us analyzea situation where two or more vehicles of the same type (ordifferent type) move in parallel at a close distance. In thedeveloped system, the width of the individual (single) loopis selected in such a way that the possibility of two near-by
SHEIK MOHAMMED ALI et al.: VEHICLE DETECTION SYSTEM FOR HETEROGENEOUS AND LANE-LESS TRAFFIC 1359Fig. 16. Output signals observed for a bicycle moving at 2.5 km/h and10 km/h. The signatures for both the cases are almost identical.vehicles sharing same loop is very low. A small vehicle suchas a bicycle covers only one of the loops, and the detection andidentiﬁcation are easy. In the case of a large vehicle, such asbus, it will cover more than one loop (three in the developedscheme), and we get signatures from these three loops at sametime. The signature information from these three loops togetherhelps to detect and identify the bus. These signatures (from thethree loops when the bus goes) are different from the one weget for small vehicles such as bicycle or motorcycle. Thus, thesystem successfully detects and distinguishes small and largevehicles even when they move in parallel at small distances.In all the previous discussions and experiments, it was as-sumed that the vehicle moves at a constant speed when itmoves above the loop detector. The vehicles may move at avery low or high speed but the signature (the shape of thevariations in the output with time) remains same. The signaturesobserved when a bicycle was running at about 10 km/h andat 2.5 km/h are shown in Fig. 16 and can be seen that thesignatures are almost same. It is important that the samplingrate of the DAS must be suitably selected to avoid aliasingwhile detecting vehicles moving at high speeds. Let us considera case where the vehicle stops when it was directly on topof the loop detector for a short duration and then proceeds.Such a situation rarely occurs in a normal trafﬁc. In such acondition, the signature may look different, compared with thesignature of the same vehicle while running at a constant speed.Fig. 17 shows a continuous record of the output from a loopwhen a bicycle moving in the forward direction was stoppedfor short intervals at three different positions (marked as 1, 2,and 3 in Fig. 17) with respect to the loop. The bicycle reachesposition-1 when its front wheel reaches the middle portion of aloop. It is indicated as position-2 when its pedal region reachesthe middle section of the loop. Similarly, position-3 is markedwhen its rear wheel reaches the middle portion of the loop. Thebicycle was moving at an almost constant speed except at therest positions 1, 2, and 3. The duration of the rest was 1.5 mineach. The position of the bicycle with respect to the loop forFig. 17. Output from a loop, when a bicycle moves with a nonuniform speedabove the loop. During the experiment, the bicycle was stopped, for 1.5 min, atthree positions on the loop as indicated.each condition is also indicated in Fig. 17. It can be observedin Fig. 17 that the signature in this case is different, but thechange in amplitude for different positions of the vehicle withrespect to the loop remained same. In such a case, a simplesignature-based detection and classiﬁcation algorithm may notwork well. As the amplitudes with respect to the positions ofthe vehicle are not affected, a threshold-level-based algorithmcan still sense and distinguish the vehicles successfully.V. CONCLUSIONAn automated vehicle detection system suitable for a hetero-geneous and lane-less trafﬁc condition has been a major stum-bling block for the implementation of many Intelligent Trafﬁcmanagement System (ITS) applications in several countries.This paper presented a possible solution to this problem using avehicle detection system based on a new multiple loop inductivesensor technique. The principle of operation of the proposedmultiple loop inductive sensor is detailed herein. Test resultsfrom a prototype system developed are provided to establishthe efﬁcacy of the proposed method. The results show that themultiple inductive loop system sense and segregate the numberof vehicles and their type. The developed inductive loop sensordetects large (e.g., bus) as well as small (e.g., bicycle) vehicles,thus making it suitable for heterogeneous trafﬁc conditions.The proposed multiple loop system is useful for roads withany type of (with or without discipline and homogeneous orheterogeneous) trafﬁc. The data provided by the measurementsystem is in digital form and hence, it is easy for transmissionto trafﬁc management centers for real-time applications. Thedeveloped system enables ITS implementations in countries
1360 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 61, NO. 5, MAY 2012with heterogeneous and lane-less trafﬁc, resulting in bettermanagement of existing roadways with reduced congestion.REFERENCES Trafﬁc Detector Handbook: 3rd Edition-Volume-I, U.S. Dept. Transp.,Fed. Highway Admin., Washington, DC, Oct. 2006, Pub. No. FHWA-HRT-06-108. R. L. Anderson, “Electromagnetic loop vehicle detectors,” IEEE Trans.Veh. Technol., vol. VT-19, no. 1, pp. 23–30, Feb. 1970. M. J. Prucha and M. View, “Inductive loop vehicle presence detector,”U.S. Patent 3 576 525, Apr. 27, 1971. R. J. Koerner and C. Park, “Inductive loop vehicle detector,” U.S. Patent3 989 932, Nov. 2, 1976. H. M. Patrick and J. L. Raymond, “Vehicle presence loop detector,” U.S.Patent 4 472 706, Sep. 18, 1984. M. A. G. Clark, “Induction loop vehicle detector,” U.S. Patent 4 568 937,Feb. 4, 1986. S. H. Lee, Y. Oh, and S. Lee, “New loop detector installation guidelinesfor real-time adaptive signal control system,” J. Eastern Asia Soc. Transp.Studies, vol. 6, pp. 2337–2348, 2005. Y.-K. Ki and D.-K. Baik, “Model for accurate speed measurement us-ing double-loop detectors,” IEEE Trans. Veh. Technol., vol. 55, no. 4,pp. 1094–1101, Jul. 2006. W.-H. Lin, J. Dahlgren, and H. Huo, “Enhancement of vehicle speedestimation with single loop detectors,” Transp. Res. Rec., J. Transp. Res.Board, vol. 1870, pp. 147–152, 2004. Y.-K. Ki and D.-K. Baik, “Vehicle-classiﬁcation algorithm for single-loopdetectors using neural networks,” IEEE Trans. Veh. Technol., vol. 55,no. 6, pp. 1704–1711, Nov. 2006. S. Meta and M. G. Cinsdikici, “Vehicle-classiﬁcation algorithm based oncomponent analysis for single-loop inductive detector,” IEEE Trans. Veh.Technol., vol. 59, no. 6, pp. 2795–2805, Jul. 2010. M. Pursula and I. Kosonen, “Microprocessor and PC-based vehicle classi-ﬁcation equipments using induction loops,” in Proc. IEEE 2nd Int. Conf.Road Trafﬁc Monit., 1989, pp. 24–28. Y. Oh and C. Lee, “The decision of the optimal shape of inductive loopdetector for real-time signal control,” J. Transp. Res. Soc. Korea, vol. 13,no. 3, pp. 67–86, 1995. Y. Oh and C. Lee, “Evaluation of conﬁdence of vehicle-detector infor-mation for real-time trafﬁc signal control,” J. Transp. Res. Soc. Korea,vol. 14, no. 3, pp. 91–125, 1996. J. Ametha, S. Tumer, and S. Darbha, “Formulation of a new methodologyto identify erroneous paired loop detectors,” in Proc. IEEE Intell. Transp.Syst., Oakland, CA, Aug. 25–29, 2001, pp. 591–596. J. Gajda, R. Sroka, M. Stencel, A. Wajda, and T. Zeqlen, “A vehicleclassiﬁcation based on inductive loop detectors,” in Proc. 18th IEEEIMTC, May 2001, vol. 1, pp. 460–464. S. S. M. Ali, B. George, L. Vanajakshi, V. Jayashankar, and V. J. Kumar,“A multiple loop vehicle detection system for heterogeneous and lane-less trafﬁc,” in Proc. IEEE I2MTC, Hangzhou, China, May 10–12, 2011,pp. 1413–1417. 16 Bit, 400 kS/s Isolated M Series MIO DAQ, Bus Powered, NIUSB-6216. [Online]. Available: http://sine.ni.com/nips/cds/print/p/lang/en/nid/207099S. Sheik Mohammed Ali received the B.E. degreein electronics and communication engineering fromAnna University, Chennai, India, in 2009. He is cur-rently working toward the M.S. degree (by research)with the Department of Electrical Engineering, In-dian Institute of Technology (IIT) Madras, Chennai.He joined the IIT Madras in 2009 as a ProjectAssociate. His research interests are in the area oftransducers, virtual instrumentation, system design,and Intelligent Transportation System.Boby George was born in Kannur, Kerala, India, in1977. He received the M.Tech. and Ph.D. degreesin electrical engineering from the Indian Institute ofTechnology (IIT) Madras, Chennai, India, in 2003and 2007, respectively.He was a Postdoctoral Fellow with the Instituteof Electrical Measurement and Measurement Sig-nal Processing, Technical University of Graz, Graz,Austria, from 2007 to 2010. Since 2010, he hasbeen with the Department of Electrical Engineering,IIT Madras as an Assistant Professor. His areas ofinterests include measurements, sensors, and instrumentation.Lelitha Vanajakshi received the B.S. and M.S. de-grees in civil engineering from the University ofKerala, Kerala, India, and the Ph.D. degree fromTexas A&M University, College Station.She is currently an Assistant Professor with theTransportation Engineering Division, Department ofCivil Engineering, Indian Institute of Technology,Madras, India. Her research interests include traf-ﬁc ﬂow modeling, trafﬁc operations, and intelligenttransportation systems.Jayashankar Venkatraman graduated from the In-stitute of Technology, Benaras Hindu University,Varanasi, India, in 1982. He obtained the Master’sdegree (by research) from the Indian Institute ofTechnology (IIT) Madras, Chennai, India, in 1991and the Ph.D. degree from the College of Engineer-ing, Guindy Anna University, Chennai, in 1994, bothin electrical engineering.He served the National Institute of Ocean Tech-nology (NIOT) from 1994 to 2001 as Head of thenon conventional energy group and did pioneeringwork in the areas of Ocean wave energy and Ocean thermal energy. He joinedthe faculty of electrical engineering, IIT Madras in 2001 and is currently aprofessor. His areas of research include high voltage insulation, non conven-tional energy sources, condition monitoring of power apparatus and biomedicalinstrumentation.