taxi efficiency measurements based on motorcade-sharing...

Research ArticleTaxi Efficiency Measurements Based on Motorcade-SharingModel Evidence from GPS-Equipped Taxi Data in Sanya

Jiawei Gui 12 and Qunqi Wu 12

1School of Economics and Management Changrsquoan University Xirsquoan Shanxi 710064 China2Center of Comprehensive Transportation Economic Management Changrsquoan University Xirsquoan Shanxi 710064 China

Correspondence should be addressed to Jiawei Gui gjwchdeducn

Received 16 July 2018 Revised 8 October 2018 Accepted 28 October 2018 Published 11 November 2018

Guest Editor Razi Iqbal

Copyright copy 2018 Jiawei Gui and Qunqi Wu This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

Urban traffic congestion has become a global problem and has garnered special importance in recent years in the transportationsector especially in taxi markets To unlock the potential of Internet of Vehicles (IoV) in Intelligent Transportation Systems (ITS)it was vital to make efficiencymeasurements In this study Distance Formula was built to calculate distances by GPS data based onmathematical equations andMotorcade-Sharing (MS)Model was proposed to improve the efficiency of collaborative vehiclesTheexperimental data of 2191 GPS-equipped taxis in Sanya of China was adopted to make comparisons between original results andmodelled resultsMeasurement results showed thatMSModel had 1054more leisure taxis reduced 5 overdriving taxis and saved3373 running distance in total compared to the originalThis indicated that the application of MSModel could not only alleviateurban traffic congestion but also optimize urban taxi markets and it has a bright future in the field of taxi and other collaborativevehicles Future directions could be improving MS Model and expanding data

1 Introduction

Nowadays there are plenty of urgent social issues in thetransportation research field For example urban traffic con-gestionhas become a global problemandhas garnered specialimportance in recent years in the transportation sectorespecially in taxi markets With the worsening situationof urban congestions and the development of autonomousvehicles taxi markets are facing increasing opportunities aswell as challenges It is essential that appropriate countermeasures must be taken to suppress or reverse or at leastalleviate the worsening situation

At the same time there aremore andmore advanced tech-nologies as well (ie intelligent traffic system autonomousvehicles etc) To unlock the potential of Internet of Vehicles(IoV) in Intelligent Transportation Systems (ITS) it is vitalto make efficiency measurements In this study Motorcade-Sharing Model was proposed for promoting taxi efficiencyand GPS-enabled taxi data in Sanya was adopted to supportcomparisons

The rest of this paper is organized as follows Section 2reviews relatedworks Section 3 introduces formulasmodelsdata and tools Section 4 describes results from the perspec-tive of path and efficiency Section 5makes a brief discussionSection 6 summarizes main conclusions contributions andproposes future directions

2 Literature Review

In this section there are several similar problems andsome related works till date including Internet of Vehiclescollaborative vehicle routing and taxi service improving

21 Internet of Vehicles Problem Intelligent TransportationSystems (ITS) had significant impact on our life [1] andthey will consist of a crucial component of our modernizedsociety [2 3] There is an amount of advanced technologiesinvolved in ITS for example radio frequency identification[4] wireless sensor networks [5 6] network coding [7 8]network energy [9] database storage [10 11] data mining

HindawiJournal of Advanced TransportationVolume 2018 Article ID 4360516 10 pageshttpsdoiorg10115520184360516

2 Journal of Advanced Transportation

[12] and cloud computing security [13] In recent years asignificant change in ITS was that much more data werecollected from a variety of sources and can be processedinto various forms [14] And numerous advanced multidisci-plinary journals began publishing a special issue of big datafor example Nature in 2008 [15] and Science in 2011 [16]

Internet of Things (IoT) is a novel paradigm that israpidly gaining ground in the scenario of modern wirelesstelecommunications as defined by Atzori et al [17] In the IoTparadigm many of the objects that surround us will be on thenetwork [18ndash20] According toMiorandi et al (2012) [21] it ispredictable that the Internet will exist as a seamless fabric ofclassic networks and networked objects by 2022 And manyof the hurdles such as digitization services of the Smart CityProject can easily be solved by IoT as discussed by Ahmedand Rani [22]

Vehicular Ad Hoc Networks (VANETs) are self-organizing networks that can significantly improve trafficsafety and travel comfort as defined by Chaqfeh et al [23]As an emerging concept transmuting the notion of VANETsInternet of Vehicles (IoV) has a bright future For exampleUmer et al (2018) proposed a dual ring connectivity modelfor VANETs under a heterogeneous traffic flow [24]

22 Collaborative Vehicle Routing Problem Vehicle RoutingProblem (VRP) is to find a plan to assign vehicles in such amanner that demands are satisfied and total mileage coveredby the fleet is a minimum [25] A lot of related works in thisfield have already been done Some of themwere with variouswindow constraints [26] while some of them were underintelligent algorithms [27 28]

In recent years more andmore companies tend to reducedistribution costs by creating collaborative environment onthe Internet so as to share various information and resourcessuch as customers warehouses and fleets [29] And numer-ous modeling methods were developed by researchers Forexample Yuan et al (2017) proposed Kminus2FMM approachto improve the accuracy level of measuring travel timevariability [30] Wang et al (2017) proposed a developedsemi-nonparametric generalized multinomial logit model toextend the standard Gumbel distribution [31] Tang et al(2017) proposed an improved hierarchical fuzzy inferencemethod based on map-matching algorithm [32] Al-Mayoufet al (2018) proposed an intersection-based segment awarealgorithm for geographic routing in VANETs [33] Lahaand Putatunda (2018) proposed three incremental learningmethods for next pickup location prediction problems [34]Xu and Cai (2018) proposed a variable neighborhood searchalgorithm for the consistent Vehicle Routing Problem [35]

Collaborative Vehicle Routing Problem (CVRP) is to findthe best solution to assign vehicles in such a manner thatdemands are satisfied by comprehensive efficiency being amaximum It helps decreasing the waiting time of passengersbut its related studies were much fewer than the ones relatedto VRP

23 Taxi Service Improving Problem Taxi service includesvaries of aspects such as safety convenience efficiency com-fort and economy Utilizing large-scale GPS data to improve

taxi services has become a popular research problem in theareas of data mining intelligent transportation geographicalinformation systems and the Internet of Things [36]

Above all traffic safety is vital Projects of safety havebeen more and more popular all over the world such as thezero-accident goal by 2050 in EU the Hazardous MaterialCooperative Research Program in the United States andthe 5-year safety production plan in China [37] And thereare already some researches focusing on taking measures toenhance traffic safety For example Zou et al (2018) proposeda developed empirical Bayes method and applied it to actualcrash datasets [38] Jain et al (2018) proposed an algorithm tocontrol traffic congestion and road safety for Social Internetof Vehicles [39] According to the evaluation results of atransportation system as discussed by Lam [40] journeytime was a significant factor in passenger waiting time whenvehicle shortages occur In addition according to La et al(2013) time pressure could impact crashes for taxi drivers[41]

In recent years rapid technical growth of autonomousvehicles tended to accelerate the process of autonomous taximarket For example an autonomous taxi service has comeinto use on a 45km campus road at Seoul National Uni-versity [42] With the development of technique and policyautonomous taxi market has a bright future Meanwhileunderstanding origin-destination distribution of taxi trips isvery important for improving effects of enhancing quality oftaxi services [43]

To sum up some helpful works have already been doneHowever few researches proposed MS Model and took bigdata into consideration Besides autonomous vehicles needassigning continuous directives in time while it is difficultfor complex algorithms to respond immediately under thebackground of big data

How to achieve collaborative vehicle routing and improvetaxi service byMSModel based on Internet of Vehicles In thenext section methods and data are involved to explore thisquestion

3 Methods and Materials

31 Measurement Steps To solve this problem 4 steps arenecessary (see Figure 1) First is acquiring data includingthe demands of passengers and the supplies of taxis Secondis measuring distances between passengers and taxis basedon the real-time locations of them Third is working outmodelled results based on the MS Model Last is makingcomparisons between original results and modelled results

There were two key points during the measurement Oneis Distance Formula (see Section 32) and the other isMotorcade-Sharing Model (see Section 33)

32 Distance Formula There are five steps in this sectionand the notations for formulas are in the Appendix (seeAppendix A)

The primary function of Distance Formula was to cal-culate distances by GPS data First in order to describe the

Journal of Advanced Transportation 3

Real-time LocationTaxi Supply

DistanceMeasurement

Real-time LocationPassenger Demand

Original Results


Motorcade-Sharing Model

DistanceMeasurement


Modelled Results

Comparisons

Figure 1 Measurement steps design

location of taxis in a mathematical way the polar coordinatesystem was established in Formula (1)

119875 (120588 120579 120593) (1)

where 120588 ge 0 0 le 120579 le 120587 and 0 le 120593 lt 2120587 The variable 120588represents the radial distance The variable 120579 represents thepolar angle The variable 120593 represents the azimuthal angle

Second in order to calculate the distance between twolocations the rectangular coordinate system was establishedin Formula (2)

997888997888119874119875 = (119909 119910 119911) (2)

where 119909 = 120588 sin 120579 cos120593 119910 = 120588 sin 120579 sin 120593 and 119911 = 120588 cos 120579Suppose that there are two locations of taxis

called 1198751 and 1198752 According to Formulas (1) and (2)and 120588 = 119877 their polar coordinates 1198751(119877 1205791 1205931) and1198752(119877 1205792 1205932) can be converted into rectangular coordinates9978889978881198741198751 = (119877 sin 1205791 cos1205931 119877 sin 1205791 sin 1205931 119877 cos 1205791) and 9978889978881198741198752 =(119877 sin 1205792 cos1205932 119877 sin 1205792 sin 1205932 119877 cos 1205792)

Third in order to calculate the distance between twolocations the cosine value of the angle between vectors 9978889978881198741198751and 9978889978881198741198752 was calculated by Formula (3)

cos⟨99788899788811987411987519978889978881198741198752⟩ =

9978889978881198741198751 sdot9978889978881198741198752100381610038161003816100381610038161003816

99788899788811987411987511003816100381610038161003816100381610038161003816100381610038161003816100381610038169978889978881198741198752

100381610038161003816100381610038161003816(3)

Formula (3) can be simplified into Formula (4) and thederivation process is in the Appendix (see Appendix B)

cos⟨99788899788811987411987519978889978881198741198752⟩ = sin 1205791 sin 1205792 cos (1205932 minus 1205931)

+ cos 1205791 cos 1205792(4)

Fourth according to the equation of arc length calculationthe distance between vectors 9978889978881198741198751 and

9978889978881198741198752 was calculated byFormula (5)

119889 = 119877 ⟨99788899788811987411987519978889978881198741198752⟩ (5)

where119877 = 63781370(119896119898) which represents the radius of theearth

According to Formula (3) Formula (5) can be simplifiedinto Formula (6)

119889= 119877 arccos (sin 1205791 sin 1205792 cos (1205932 minus 1205931) + cos 1205791 cos 1205792)

(6)

In other words the distance between two locations of taxiscan be calculated approximately by the longitudes and lati-tudes of them based on GPS data and the radius of the earth

Suppose that the variable 119879119863 represents total distancebetween two locations of taxis the variable 119879119874119863 representstotal occupied distance between them and the variable 119879119880119863represents total unoccupied distance between them Andthen 119879119863 can be computed by Formula (7) [44]

119879119863 = 119879119874119863 + 119879119880119863 (7)

where 119879119863 = sum119868119894=1

sum119879119905=1

119889119894119905 119879119874119863 = sum119868119894=1

sum119879119905=1

119889119894119905 | 119901 ge 1and 119879119880119863 = sum119868

119894=1sum119879119905=1

119889119894119905 | 119901 = 0At last the target of this study is to minimize the total

distance and it can be indicated in Formula (8)

min1le119894le1198681le119905le119879

119879119863 = min1le119894le1198681le119905le119879

119868

sum119894=1

119879

sum119905=1

119889119894119905 (8)

Suppose that the demand of passengers is invariable whichmeans that 119879119874119863 is constant And then Formula (8) can besimplified into Formula (9)

min1le119894le1198681le119905le119879

119879119863 = 119879119874119863 + min1le119894le1198681le119905le119879

119868

sum119894=1

119879

sum119905=1

119889119894119905 | 119901 = 0 (9)

Minimizing the total distance is equivalent tominimizing thetotal unoccupied distance And in the next sectionMSModelis established to solve this problem


Real-time LocationDemand

Real-time LocationSupply

Taxi Passenger

Online Platform

(a)

TaxiInformation

PassengerInformation

Taxi Passenger

Online Platform

(b)

Figure 2 Information exchange design (a) Information flows in online platform (b) Information flows out from online platform

Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(a)

Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(Wait)

(Wait)

(b)

Figure 3 Information exchange of a hypothetical example (a) Information flows in online platform (b) Information flows out from onlineplatform

33 Motorcade-Sharing Model Motorcade-Sharing (MS)Model was established to minimize the total unoccupieddistance and it contained two main strategies ProximityStrategy and Passivity Strategy

(i) Proximity Strategy The unoccupied taxi with closerdistance has its priority to pick up passengers immediately inother words passengers will be picked up by the unoccupiedtaxi whose position is closest to them at that time

(ii) Passivity Strategy All the unoccupied taxis pick uppassengers passively on the basis of the above-mentionedProximity Strategy and they keep parking after passengersarrive That is to say only occupied taxis keep driving whileothers keep parking until necessary

MS Model needs an online platform to exchange infor-mation between taxis and passengers Taxis and passengerssend their real-time locations to the online platform (seeFigure 2(a)) and get partnersrsquo information in return (see Fig-ure 2(b)) Suppose that there are only 3 taxis and 3 passengersonline (see Figure 4(a)) In the original setting Taxi1 picksup Passenger1 Taxi2 picks up Passenger2 and Taxi3 picksup Passenger3 According toMSModel taxis and passengerssend their real-time locations to the online platform (seeFigure 3(a)) and get partnersrsquo information in return (seeFigure 3(b)) As a modelled result (see Figure 4(b)) Taxi3picks up Passenger3 as usual However Taxi2 is going topick up Passenger1 because Taxi2 is closer to Passenger1than Taxi1 Besides Taxi1 and Passenger2 keep waiting fora second If there is still no better choice in the followingseconds Taxi1 picks Passenger2 in the endAutonomous taxishave no conflicts of interests and need an online platform toshare information Thus MSModel is especially suitable and

useful for autonomous vehicles to improve comprehensiveefficiency in the future

When there is more than one taxi online it is pos-sible for those taxis to improve comprehensive efficiencyby cooperating with others by MS Model It seems thatmodelled routing without overlapping has higher efficiencythan original setting however it needs evidences from realdata to verify

34 Data and Tool (i) Data Source Data in this study wascollected from the big data platform Travel Cloud whichwasdeveloped by Ministry of Transport of the Peoplersquos Republicof China (see Data Availability) The GPS data of Taxis inthis study was provided by Sanya Traffic and TransportationBureau in Hainan Province in China Major data itemsinclude vehicle ID longitude latitude and time (see the filenamed ldquoDataxlsxrdquo in the Supplementary Materials (availablehere))

(ii) Data Processing According to the origin data thelocations of 2506 taxis in Sanya were recorded every 15seconds from 900 am to 959 am on Nov 15th in 2016adding up to 766042 records First all the locations of taxisare averaged every 60 seconds so as to improve the accuracyof data that is to say average locations of those taxis wererecorded every 60 seconds entirely Second the flaw data wasremoved so as to ensure the integrity of data As a result therewere 2191 taxis and 131460 records in the experimental data

(iii) Implement Tool Several kinds of statistics softwarewere adopted in this study as follows Excel was appliedfor original data processing and result storage Matlab was


Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2

(a)

Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2(Wait)

(Wait)

(b)

Figure 4 Routing comparison of a hypothetical example (a) Original vehicle routing (b) Improved vehicle routing

1818

519

195

20La

titud

e (O

rigin

al)

109 1095 110 11051085Longitude (Original)

(a)

1819

20La

titud

e (M

odel

led)

185

195

109 1095 110 11051085Longitude (Modelled)

(b)

Figure 5 Path comparison of taxi positions located minutely at a fixed monitor where the longitude was in the range of [108∘111∘] while thelatitude was in the range of [18∘21∘] (a) Original positions (b) Modelled positions

applied for programming (see the file named ldquoProgram-mingpdfrdquo in the Supplementary Materials) Stata was appliedfor drawing

4 Results

In this section modelled results of 2191 GPS-equipped taxisin Sanya were compared with original results from theperspectives of path and efficiency

41 Path Comparison We adopted scatter diagram ratherthan line diagram because the former seems clearer thanthe latter Taxi positions were located minutely at a fixedmonitor Thus crowded districts lied in dark areas To beginwith there was seldom difference between original results(see Figure 5(a)) and modelled results (see Figure 5(b))in path comparison results of all the 2191 taxis because of

compression However disparities gradually appeared whenfocusing on specific local areas It came to a specific resultby a narrow scope that the longitude was in the rangeof [1094∘1096∘] while the latitude was in the range of[182∘184∘] (see Figure 6) And at last it came to anotherspecific result by a far narrow scope that the longitude wasin the range of [109495∘109505∘] while the latitude was inthe range of [18245∘18255∘] (see Figure 7)

It can be seen from Figures 5ndash7 that modelled paths wereobviously smoother and more unobstructed than originalpaths thus MS Model helped alleviating urban traffic con-gestion However it needs digital comparisons to quantify

42 Efficiency Comparison All the 2191 taxis were dividedinto different groups by their running distances First theefficiencies of running distances were compared (see Table 1)There were two main findings in Table 1 On the one hand


182

182

518

318

35

184

Latit

ude (

Orig

inal

)


(a)

182

182

518

318

35

184

Latit

ude (

Mod

elle

d)


(b)

Figure 6 Path comparison of taxi positions located minutely at a fixed monitor where the longitude was in the range of [1094∘1096∘] whilethe latitude was in the range of [182∘184∘] (a) Original positions (b) Modelled positions

1095 109505109495Longitude (Original)

182

4518

25

182

55La

titud

e (O

rigin

al)

(a)

182

4518

25

182

55La

titud

e (M

odel

led)

1095 109505109495Longitude (Modelled)

(b)

Figure 7 Path comparison of taxi positions located minutely at a fixed monitor where the longitude was in the range of [109495∘109505∘]while the latitude was in the range of [18245∘18255∘] (a) Original positions (b) Modelled positions

modelled total distance (2329205km) was 3373 lowerthan original total distance (3514737km) which impliedthat modelled scheme was more fuel-efficient and thuseconomical in another word On the other hand modelledscheme obtained reducing distance results in groups above20 km and increasing results in groups below 20 km whichimplied that modelled scheme was more efficient in generalrather than unilateral

Second the efficiencies of running taxis were compared(see Table 2) There were three main findings in Table 2Above all modelled taxis with 0 km (289) were about fivetimes the quantity of original taxis with 0 km (58) andincreasing 1054 leisure taxis implied that modelled schemewasmore fuel-efficient and thus economical than the originalFurthermore modelled taxis in groups above 60 km (5) werehalf the original taxis in groups above 60 km (10) andreducing 5 overdriving taxis implied that modelled schemehad fewer overdriving taxis than the original and thus wassafe in another word Last but not least modelled scheme

obtained reducing taxi results in groups above 20 km andincreasing results in groups below 20 km which implied thatmodelled scheme was more efficient in general rather thanunilateral

It can be seen from Tables 1 and 2 that modelled schemehad higher fuel-efficient more economical and safer taxis ingeneral compared with the original Thus MS Model helpedoptimizing urban taxi markets

5 Discussion

Whilemaking comparisons in the last section there were stillfive factors that had not been contemplated within which aredescribed as follows

First Proximity Strategy has limitations In reality itwas difficult for taxis to pick up passengers according totheir relative positions even if information could be sharedadequately It required the online platform to keep good


Table 1 Efficiency comparison of running distancesThe data in the 2nd and 4th columns are accumulation of corresponding distances unitkm

Distance Original results Modelled results119889 le 10 226316 644 435411 186910 lt 119889 le 20 1185774 3374 868347 372820 lt 119889 le 30 1082263 3079 507667 218030 lt 119889 le 40 522342 1486 275253 118240 lt 119889 le 50 265303 755 130322 56050 lt 119889 le 60 151005 430 76144 327119889 gt 60 81734 233 36061 155Total 3514737 10000 2329205 10000

Table 2 Efficiency comparison of running taxis The data in the 2nd and 4th columns are numbers of taxis with corresponding distances

Distance Original results Modelled results119889 = 0 58 265 289 13190 lt 119889 le 10 654 2985 952 434510 lt 119889 le 20 780 3560 612 279320 lt 119889 le 30 448 2045 209 95430 lt 119889 le 40 152 694 80 36540 lt 119889 le 50 61 278 30 13750 lt 119889 le 60 28 128 14 064119889 gt 60 10 046 5 023Total 2191 10000 2191 10000

balance of allocation That is to say MS Model might havemuch lower efficiency because of mistakes

Second Passivity Strategy has limitations In realityit was difficult for taxis to keep parking after passen-gers reach their destinations because there were otherrequirements (ie traffic control refueling driversrsquo turningover etc) Sometimes it was more reasonable for taxis toapproach busy areas rather than staying That is to say MSModel might have much lower efficiency because of trafficconditions

Third information sharing was essential for MS ModelSeveral management plans for vehicular traffic (ie bigdata collections cloud computing platform constructionautonomous technique etc) should also be prepared com-pletely That is to say MS Model might have much lowerefficiency because of failure facilities

Fourth in most cities in China morning rush hours areoften centered in the period of 700-900 am [45] Thusthe experimental data in this study was not in rush hoursHowever traffic conditions may vary over different timeperiods more data should be analyzed and presented in thefuture research That is to say MS Model might have muchhigher efficiency during rush hours

At last this study focused on the efficiency comparisonbetweenMSModel and the originalThus path optimizationswere not considered and taxis drove the original path in MSModel That is to say MS Model might have much higherefficiency by optimizing routes

To sum up the results could bemore accurate by improv-ing the model or expanding the sample size Neverthelessthe trend would not change that modelled scheme would still

result in higher efficiency than the original In addition MSModel is suitable and useful for autonomous vehicles

6 Conclusions

In this study the authors proposed Motorcade-Sharing (MS)Model for promoting taxi efficiency and adopted GPS-enabled taxi data in Sanya to make efficiency comparisonsbyDistance Formula Given the arguments mentioned abovethe major findings in this article include several contentswhich are described as follows

(i) The application of MS Model could alleviate urbantraffic congestion and it resulted in less obstructed path andmore leisure taxis than the original (see Section 41)

(ii) The application of MS Model could optimize urbantaxi markets and it resulted in higher fuel-efficient moreeconomical and safer taxis than the original (see Section 42)

(iii) The application of MS Model could improve effectsaccording to statistic results (see Section 42) and it hasa bright future in the field of taxi and other collaborativevehicles

The contributions of this study could be as follows(i) Distance Formula was built to calculate distances byGPS data based on mathematical equations (ii) Motorcade-Sharing Model was proposed to improve the efficiency ofcollaborative vehicles (iii) It was verified that modelledscheme had higher efficiency than the original

Future directions for research can be in two ways One isimproving MS Model and the other is expanding the samplesize


Appendix

A Notations for Formulas

See Table 3

B Derivation Process of Formula (4)

cos⟨99788899788811987411987519978889978881198741198752⟩

= 119877 sin 1205791 cos1205931119877 sin 1205792 cos1205932 + 119877 sin 1205791 sin 1205931119877 sin 1205792 sin 1205932 + 119877 cos 1205791119877 cos 1205792radic(119877 sin 1205791 cos1205931)2 + (119877 sin 1205791 sin 1205931)2 + (119877 cos 1205791)2radic(119877 sin 1205792 cos1205932)2 + (119877 sin 1205792 sin 1205932)2 + (119877 cos 1205792)2

= 1198772 (sin 1205791 sin 1205792 cos1205931 cos1205932 + sin 1205791 sin 1205792 sin 1205931 sin 1205932 + cos 1205791 cos 1205792)1198772radic(sin 1205791 cos1205931)2 + (sin 1205791 sin 1205931)2 + (cos 1205791)2radic(sin 1205792 cos1205932)2 + (sin 1205792 sin 1205932)2 + (cos 1205792)2

= sin 1205791 sin 1205792 (cos 1205931 cos1205932 + sin 1205931 sin 1205932) + cos 1205791 cos 1205792 = sin 1205791 sin 1205792 cos (1205932 minus 1205931) + cos 1205791 cos 1205792

(B1)

Table 3 Notations for Formulas

Notation Explanation

119874 The center of the rectangular coordinate systemThecenter of the earth

119875 The location120588 The radial distance120579 The polar angle120593 The azimuthal angle119894 The taxi ID119905 The time119868 Themaximum of taxi ID119879 Themaximum of time119875119894 The location of taxis119877 The radius of the earth120579119894 The longitude of the location of taxis120593119894 The latitude of the location of taxis119889 The distance between two locations of taxis119901 The number of passengers in a taxi119879119863 Total distance between two locations of taxis119879119874119863 Total occupied distance between two locations of taxis

119879119880119863 Total unoccupied distance between two locations oftaxis

Data Availability

The data collected during the study are freely available athttpstransportdatacntraffictravelopendetailid=188

Conflicts of Interest

The authors declare that there are no conflicts of interestregarding the publication of this article

Authorsrsquo Contributions

Jiawei Gui and Qunqi Wu contributed to study conceptionand design Jiawei Gui contributed to data collection andsoftware analysis and interpretation of results manuscriptpreparation-draft and manuscript preparation-revised ver-sion All authors reviewed the results and approved the finalversion of the manuscript

Acknowledgments

The authors thank Razi Iqbal for providing objective com-ments twice to improve this paper and thank Aya Moezfor providing editing guidance Jiawei Gui was rewardedby the National Scholarship of China for doctoral studentsand appreciated that The work described in this articlewas supported in part by the Strategic Planning ResearchProject of Ministry of Transport of China [2018-7-9] and[2018-16-9] in part by the National Social Science Fundof China [17BJY139] and in part by Changrsquoan UniversityExcellent Doctoral Dissertation Cultivation Project of Chi-nese Universities Scientific Fund of China [2017023002]

Supplementary Materials

(i) A file named ldquoDataxlsxrdquo in this file the major data usedin this study is recorded including original mode modelledmode and the comparison results between them Becauseof space limitation tables in this file (824KB in size) recordonly 51 taxis (No1-50 and No2191) of full table (327MB insize) (ii) A file named ldquoProgrammingpdfrdquo in this file majorcodes programmed by Matlab are recorded (48KB in size)(Supplementary Materials)

References

[1] F-Y Wang ldquoParallel control and management for intelligenttransportation systems concepts architectures and applica-


tionsrdquo IEEE Transactions on Intelligent Transportation Systemsvol 11 no 3 pp 630ndash638 2010

[2] G Zhang M Li and J Wang ldquoApplication of the AdvancedPublic Transport System in Cities of China and the Prospectof Its Future Developmentrdquo Journal of Transportation SystemsEngineering and Information Technology vol 7 no 5 pp 24ndash302007

[3] X Cheng L Yang and X Shen ldquoD2D for intelligent trans-portation systems A feasibility studyrdquo IEEE Transactions onIntelligent Transportation Systems vol 16 no 4 pp 1784ndash17932015

[4] P Vrba F Macurek and V Marik ldquoUsing Radio FrequencyIdentification in Agent-BasedManufacturingControl Systemsrdquoin Proceedings of the International Conference on IndustrialApplications of Holonic and Multi-Agent Systems vol 3593 pp176ndash187 Springer Berlin Heidelberg Copenhagen Denmark2005

[5] V Y Liu and Z Yu ldquoWireless Sensor Networks for Internetof things A systematic review and classificationrdquo InformationTechnology Journal vol 12 no 16 pp 3581ndash3585 2013

[6] S Ali A Ahmad R Iqbal S Saleem and T Umer ldquoJoint RRH-Association Sub-Channel Assignment and Power Allocation inMulti-Tier 5G C-Ransrdquo IEEE Access vol 6 pp 34393ndash344022018

[7] F Jamil A Javaid T Umer andM H Rehmani ldquoA comprehen-sive survey of network coding in vehicular ad-hoc networksrdquoWireless Networks vol 23 no 8 pp 2395ndash2414 2017

[8] S H Ahmed S H Bouk M A Yaqub D Kim and H SongldquoDIFS Distributed Interest Forwarder Selection in VehicularNamed Data Networksrdquo IEEE Transactions on Intelligent Trans-portation Systems vol 19 no 9 pp 3076ndash3080 2018

[9] G Auer V Giannini C Desset et al ldquoHow much energy isneeded to run a wireless networkrdquo IEEE Wireless Communi-cations Magazine vol 18 no 5 pp 40ndash49 2011

[10] F Chang J Dean S Ghemawat et al ldquoBigtable a distributedstorage system for structured datardquo ACM Transactions onComputer Systems vol 26 no 2 Article ID 1365816 2008

[11] M Iwazume T Iwase K Tanaka and H Fujii ldquoBig datain memory Benchmarking in memory database using thedistributed key-value store for constructing a large scale infor-mation infrastructurerdquo in Proceedings of the 38th Annual IEEEComputer Software and Applications Conference WorkshopsCOMPSACW 2014 pp 199ndash204 New York NY USA 2014

[12] C-W Tsai C-F Lai M-C Chiang and L T Yang ldquoDatamining for internet of things a surveyrdquo IEEE CommunicationsSurveys amp Tutorials vol 16 no 1 pp 77ndash97 2014

[13] K Hashizume D G Rosado E Fernandez-Medina and E BFernandez ldquoAn analysis of security issues for cloud computingrdquoJournal of Internet Services amp Applications vol 4 no 1 pp 1ndash132013

[14] J Zhang F-Y Wang K Wang W-H Lin X Xu and C ChenldquoData-driven intelligent transportation systems a surveyrdquo IEEETransactions on Intelligent Transportation Systems vol 12 no 4pp 1624ndash1639 2011

[15] F Frankel and R Reid ldquoBig data Distillingmeaning from datardquoNature vol 455 no 7209 p 30 2008

[16] W Los and J Wood ldquoDealing with data Upgrading infrastruc-turerdquo Science vol 331 no 6024 pp 1515-1516 2011

[17] L Atzori A Iera and G Morabito ldquoThe internet of things asurveyrdquoComputer Networks vol 54 no 15 pp 2787ndash2805 2010

[18] J Gubbi R Buyya S Marusic and M Palaniswami ldquoInternetof Things (IoT) a vision architectural elements and futuredirectionsrdquo Future Generation Computer Systems vol 29 no 7pp 1645ndash1660 2013

[19] M Gohar S H Ahmed M Khan N Guizani A Ahmedand A U Rahman ldquoA Big Data Analytics Architecture for theInternet of SmallThingsrdquo IEEE Communications Magazine vol56 no 2 pp 128ndash133 2018

[20] T A Butt R Iqbal and S C Shah ldquoSocial Internet ofVehicles Architecture and enabling technologiesrdquoComputersampElectrical Engineering vol 69 pp 68ndash84 2018

[21] D Miorandi S Sicari F de Pellegrini and I Chlamtac ldquoInter-net of things vision applications and research challengesrdquo AdHoc Networks vol 10 no 7 pp 1497ndash1516 2012

[22] S H Ahmed and S Rani ldquoA hybrid approach Smart Street usecase and future aspects for Internet of Things in smart citiesrdquoFuture Generation Computer Systems vol 79 pp 941ndash951 2018

[23] M Chaqfeh A Lakas and I Jawhar ldquoA survey on datadissemination in vehicular ad hoc networksrdquo Vehicular Com-munications vol 1 no 4 pp 214ndash225 2014

[24] T Umer M K Afzal E U Munir and M Alam ldquoA Dual RingConnectivity Model for VANET Under Heterogeneous TrafficFlowrdquoWireless Personal Communications vol 98 no 1 pp 105ndash118 2018

[25] G B Dantzig and J H Ramser ldquoThe truck dispatchingproblemrdquoManagement Science vol 6 no 1 pp 80ndash91 1959

[26] M M Solomon ldquoAlgorithms for the vehicle routing andscheduling problemswith timewindowconstraintsrdquoOperationsResearch vol 35 no 2 pp 254ndash265 1987

[27] M Filipec D Skrlec and S Krajcar ldquoGenetic algorithmapproach for multiple depot capacitated vehicle routing prob-lem solving with heuristic improvementsrdquo International Journalof Modelling and Simulation vol 20 no 4 pp 320ndash328 2000

[28] H S Mahmassani ldquoDynamic Network Traffic Assignment andSimulation Methodology for Advanced System ManagementApplicationsrdquo in Networks and Spatial Economics vol 1 pp267ndash292 2001

[29] B SunMWei and J Yao ldquoCollaborative vehicle route problembased on vehicle task reliabilityrdquo Application Research of Com-puters vol 30 no 8 pp 2280ndash2287 2013

[30] S Yuan B Wright Y Zou and Y Wang ldquoQuantification ofvariability of valid travel times with fmms for buses passengercars and taxisrdquo IET Intelligent Transport Systems vol 11 no 1pp 1ndash9 2017

[31] K Wang X Ye R M Pendyala and Y Zou ldquoOn the devel-opment of a semi-nonparametric generalized multinomial logitmodel for travel-relatedchoicesrdquoPLoS ONE vol 12 no 10 2017

[32] J Tang S Zhang Y Zou F Liu and X Ma ldquoAn adaptivemap-matching algorithm based on hierarchical fuzzy systemfrom vehicular GPS datardquo PLoS ONE vol 12 no 12 Article ID0188796 2017

[33] Y R Al-Mayouf N F Abdullah O A Mahdi et al ldquoReal-Time Intersection-Based Segment Aware Routing Algorithmfor UrbanVehicularNetworksrdquo IEEE Transactions on IntelligentTransportation Systems vol 19 no 7 pp 2125ndash2141 2018

[34] A K Laha and S Putatunda ldquoReal time location predictionwith taxi-GPS data streamsrdquo Transportation Research Part CEmerging Technologies vol 92 pp 298ndash322 2018

[35] Z Xu and Y Cai ldquoVariable neighborhood search for consistentvehicle routing problemrdquo Expert Systems with Applications vol113 pp 66ndash76 2018


[36] R Wang C-Y Chow Y Lyu et al ldquoTaxiRec Recommendingroad clusters to taxi drivers using ranking-based extremelearning machinesrdquo IEEE Transactions on Knowledge and DataEngineering vol 30 no 3 pp 585ndash598 2018

[37] C Ren Q Wu C Zhang and S Zhang ldquoA Normal Dis-tribution-Based Methodology for Analysis of Fatal Accidentsin Land Hazardous Material Transportationrdquo InternationalJournal of Environmental Research and Public Health vol 15 no7 Article ID 15071437 2018

[38] Y Zou J E Ash B-J Park D Lord and L Wu ldquoEmpiricalBayes estimates of finitemixture of negative binomial regressionmodels and its application to highway safetyrdquo Journal of AppliedStatistics vol 45 no 9 pp 1652ndash1669 2018

[39] B Jain G Brar J Malhotra S Rani and S H Ahmed ldquoAcross layer protocol for traffic management in Social Internetof Vehiclesrdquo Future Generation Computer Systems vol 82 pp707ndash714 2018

[40] S Lam J Taghia and J Katupitiya ldquoEvaluation of a trans-portation system employing autonomous vehiclesrdquo Journal ofAdvanced Transportation vol 50 no 8 pp 2266ndash2287 2016

[41] Q N La A H Lee L B Meuleners and D Van DuongldquoPrevalence and factors associated with road traffic crashamong taxi drivers in Hanoi Vietnamrdquo Accident Analysis ampPrevention vol 50 pp 451ndash455 2013

[42] S-W Kim G-P Gwon W-S Hur et al ldquoAutonomous CampusMobility Services Using Driverless Taxirdquo IEEE Transactions onIntelligent Transportation Systems vol 18 no 12 pp 3513ndash35262017

[43] J Tang S Zhang X Chen F Liu and Y Zou ldquoTaxi tripsdistribution modeling based on Entropy-Maximizing theoryA case study in Harbin citymdashChinardquo Physica A StatisticalMechanics and its Applications vol 493 pp 430ndash443 2018

[44] H Yang and S C Wong ldquoA network model of urban taxiservicesrdquo Transportation Research Part B Methodological vol32 no 4 pp 235ndash246 1998

[45] R Sun M Li and Q Wu ldquoResearch on Commuting TravelMode Choice of CarOwnersConsidering Return Trip Contain-ing Activitiesrdquo Sustainability vol 10 no 10 Article ID 101034942018

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


[12] and cloud computing security [13] In recent years asignificant change in ITS was that much more data werecollected from a variety of sources and can be processedinto various forms [14] And numerous advanced multidisci-plinary journals began publishing a special issue of big datafor example Nature in 2008 [15] and Science in 2011 [16]

Internet of Things (IoT) is a novel paradigm that israpidly gaining ground in the scenario of modern wirelesstelecommunications as defined by Atzori et al [17] In the IoTparadigm many of the objects that surround us will be on thenetwork [18ndash20] According toMiorandi et al (2012) [21] it ispredictable that the Internet will exist as a seamless fabric ofclassic networks and networked objects by 2022 And manyof the hurdles such as digitization services of the Smart CityProject can easily be solved by IoT as discussed by Ahmedand Rani [22]

Vehicular Ad Hoc Networks (VANETs) are self-organizing networks that can significantly improve trafficsafety and travel comfort as defined by Chaqfeh et al [23]As an emerging concept transmuting the notion of VANETsInternet of Vehicles (IoV) has a bright future For exampleUmer et al (2018) proposed a dual ring connectivity modelfor VANETs under a heterogeneous traffic flow [24]

22 Collaborative Vehicle Routing Problem Vehicle RoutingProblem (VRP) is to find a plan to assign vehicles in such amanner that demands are satisfied and total mileage coveredby the fleet is a minimum [25] A lot of related works in thisfield have already been done Some of themwere with variouswindow constraints [26] while some of them were underintelligent algorithms [27 28]

In recent years more andmore companies tend to reducedistribution costs by creating collaborative environment onthe Internet so as to share various information and resourcessuch as customers warehouses and fleets [29] And numer-ous modeling methods were developed by researchers Forexample Yuan et al (2017) proposed Kminus2FMM approachto improve the accuracy level of measuring travel timevariability [30] Wang et al (2017) proposed a developedsemi-nonparametric generalized multinomial logit model toextend the standard Gumbel distribution [31] Tang et al(2017) proposed an improved hierarchical fuzzy inferencemethod based on map-matching algorithm [32] Al-Mayoufet al (2018) proposed an intersection-based segment awarealgorithm for geographic routing in VANETs [33] Lahaand Putatunda (2018) proposed three incremental learningmethods for next pickup location prediction problems [34]Xu and Cai (2018) proposed a variable neighborhood searchalgorithm for the consistent Vehicle Routing Problem [35]

Collaborative Vehicle Routing Problem (CVRP) is to findthe best solution to assign vehicles in such a manner thatdemands are satisfied by comprehensive efficiency being amaximum It helps decreasing the waiting time of passengersbut its related studies were much fewer than the ones relatedto VRP

23 Taxi Service Improving Problem Taxi service includesvaries of aspects such as safety convenience efficiency com-fort and economy Utilizing large-scale GPS data to improve

taxi services has become a popular research problem in theareas of data mining intelligent transportation geographicalinformation systems and the Internet of Things [36]

Above all traffic safety is vital Projects of safety havebeen more and more popular all over the world such as thezero-accident goal by 2050 in EU the Hazardous MaterialCooperative Research Program in the United States andthe 5-year safety production plan in China [37] And thereare already some researches focusing on taking measures toenhance traffic safety For example Zou et al (2018) proposeda developed empirical Bayes method and applied it to actualcrash datasets [38] Jain et al (2018) proposed an algorithm tocontrol traffic congestion and road safety for Social Internetof Vehicles [39] According to the evaluation results of atransportation system as discussed by Lam [40] journeytime was a significant factor in passenger waiting time whenvehicle shortages occur In addition according to La et al(2013) time pressure could impact crashes for taxi drivers[41]

In recent years rapid technical growth of autonomousvehicles tended to accelerate the process of autonomous taximarket For example an autonomous taxi service has comeinto use on a 45km campus road at Seoul National Uni-versity [42] With the development of technique and policyautonomous taxi market has a bright future Meanwhileunderstanding origin-destination distribution of taxi trips isvery important for improving effects of enhancing quality oftaxi services [43]

To sum up some helpful works have already been doneHowever few researches proposed MS Model and took bigdata into consideration Besides autonomous vehicles needassigning continuous directives in time while it is difficultfor complex algorithms to respond immediately under thebackground of big data

How to achieve collaborative vehicle routing and improvetaxi service byMSModel based on Internet of Vehicles In thenext section methods and data are involved to explore thisquestion

3 Methods and Materials

31 Measurement Steps To solve this problem 4 steps arenecessary (see Figure 1) First is acquiring data includingthe demands of passengers and the supplies of taxis Secondis measuring distances between passengers and taxis basedon the real-time locations of them Third is working outmodelled results based on the MS Model Last is makingcomparisons between original results and modelled results

There were two key points during the measurement Oneis Distance Formula (see Section 32) and the other isMotorcade-Sharing Model (see Section 33)

32 Distance Formula There are five steps in this sectionand the notations for formulas are in the Appendix (seeAppendix A)

The primary function of Distance Formula was to cal-culate distances by GPS data First in order to describe the



DistanceMeasurement


Original Results



DistanceMeasurement


Modelled Results

Comparisons



119875 (120588 120579 120593) (1)



997888997888119874119875 = (119909 119910 119911) (2)




cos⟨99788899788811987411987519978889978881198741198752⟩ =

9978889978881198741198751 sdot9978889978881198741198752100381610038161003816100381610038161003816

99788899788811987411987511003816100381610038161003816100381610038161003816100381610038161003816100381610038169978889978881198741198752

100381610038161003816100381610038161003816(3)



+ cos 1205791 cos 1205792(4)



119889 = 119877 ⟨99788899788811987411987519978889978881198741198752⟩ (5)




(6)



119879119863 = 119879119874119863 + 119879119880119863 (7)

where 119879119863 = sum119868119894=1

sum119879119905=1

119889119894119905 119879119874119863 = sum119868119894=1

sum119879119905=1

119889119894119905 | 119901 ge 1and 119879119880119863 = sum119868

119894=1sum119879119905=1



min1le119894le1198681le119905le119879

119879119863 = min1le119894le1198681le119905le119879

119868

sum119894=1

119879

sum119905=1

119889119894119905 (8)


min1le119894le1198681le119905le119879

119879119863 = 119879119874119863 + min1le119894le1198681le119905le119879

119868

sum119894=1

119879

sum119905=1

119889119894119905 | 119901 = 0 (9)





Taxi Passenger

Online Platform

(a)

TaxiInformation


Taxi Passenger

Online Platform

(b)


Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(a)

Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(Wait)

(Wait)

(b)












Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2

(a)

Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2(Wait)

(Wait)

(b)


1818

519

195

20La

titud

e (O

rigin

al)


(a)

1819

20La

titud

e (M

odel

led)

185

195


(b)



4 Results







182

182

518

318

35

184

Latit

ude (

Orig

inal

)


(a)

182

182

518

318

35

184

Latit

ude (

Mod

elle

d)


(b)



182

4518

25

182

55La

titud

e (O

rigin

al)

(a)

182

4518

25

182

55La

titud

e (M

odel

led)


(b)






5 Discussion















6 Conclusions








Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




DistanceMeasurement


Original Results



DistanceMeasurement


Modelled Results

Comparisons



119875 (120588 120579 120593) (1)



997888997888119874119875 = (119909 119910 119911) (2)




cos⟨99788899788811987411987519978889978881198741198752⟩ =

9978889978881198741198751 sdot9978889978881198741198752100381610038161003816100381610038161003816

99788899788811987411987511003816100381610038161003816100381610038161003816100381610038161003816100381610038169978889978881198741198752

100381610038161003816100381610038161003816(3)



+ cos 1205791 cos 1205792(4)



119889 = 119877 ⟨99788899788811987411987519978889978881198741198752⟩ (5)




(6)



119879119863 = 119879119874119863 + 119879119880119863 (7)

where 119879119863 = sum119868119894=1

sum119879119905=1

119889119894119905 119879119874119863 = sum119868119894=1

sum119879119905=1

119889119894119905 | 119901 ge 1and 119879119880119863 = sum119868

119894=1sum119879119905=1



min1le119894le1198681le119905le119879

119879119863 = min1le119894le1198681le119905le119879

119868

sum119894=1

119879

sum119905=1

119889119894119905 (8)


min1le119894le1198681le119905le119879

119879119863 = 119879119874119863 + min1le119894le1198681le119905le119879

119868

sum119894=1

119879

sum119905=1

119889119894119905 | 119901 = 0 (9)





Taxi Passenger

Online Platform

(a)

TaxiInformation


Taxi Passenger

Online Platform

(b)


Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(a)

Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(Wait)

(Wait)

(b)












Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2

(a)

Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2(Wait)

(Wait)

(b)


1818

519

195

20La

titud

e (O

rigin

al)


(a)

1819

20La

titud

e (M

odel

led)

185

195


(b)



4 Results







182

182

518

318

35

184

Latit

ude (

Orig

inal

)


(a)

182

182

518

318

35

184

Latit

ude (

Mod

elle

d)


(b)



182

4518

25

182

55La

titud

e (O

rigin

al)

(a)

182

4518

25

182

55La

titud

e (M

odel

led)


(b)






5 Discussion















6 Conclusions








Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





Taxi Passenger

Online Platform

(a)

TaxiInformation


Taxi Passenger

Online Platform

(b)


Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(a)

Taxi3 Passenger3

Online Platform

Taxi1 Passenger1

Taxi2 Passenger2

(Wait)

(Wait)

(b)












Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2

(a)

Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2(Wait)

(Wait)

(b)


1818

519

195

20La

titud

e (O

rigin

al)


(a)

1819

20La

titud

e (M

odel

led)

185

195


(b)



4 Results







182

182

518

318

35

184

Latit

ude (

Orig

inal

)


(a)

182

182

518

318

35

184

Latit

ude (

Mod

elle

d)


(b)



182

4518

25

182

55La

titud

e (O

rigin

al)

(a)

182

4518

25

182

55La

titud

e (M

odel

led)


(b)






5 Discussion















6 Conclusions








Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2

(a)

Taxi2

Passenger1

Taxi3

Passenger3

Taxi1

Passenger2(Wait)

(Wait)

(b)


1818

519

195

20La

titud

e (O

rigin

al)


(a)

1819

20La

titud

e (M

odel

led)

185

195


(b)



4 Results







182

182

518

318

35

184

Latit

ude (

Orig

inal

)


(a)

182

182

518

318

35

184

Latit

ude (

Mod

elle

d)


(b)



182

4518

25

182

55La

titud

e (O

rigin

al)

(a)

182

4518

25

182

55La

titud

e (M

odel

led)


(b)






5 Discussion















6 Conclusions








Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



182

182

518

318

35

184

Latit

ude (

Orig

inal

)


(a)

182

182

518

318

35

184

Latit

ude (

Mod

elle

d)


(b)



182

4518

25

182

55La

titud

e (O

rigin

al)

(a)

182

4518

25

182

55La

titud

e (M

odel

led)


(b)






5 Discussion















6 Conclusions








Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia














6 Conclusions








Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Appendix


See Table 3


cos⟨99788899788811987411987519978889978881198741198752⟩




(B1)






Data Availability






Acknowledgments




References



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



















































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


taxi efficiency measurements based on motorcade-sharing...

Documents