See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303761444

Intelligent Carpooling System

Chapter  in  Intelligent Systems Reference Library · June 2016

DOI: 10.1007/978-3-319-32168-4_2

CITATIONS READS

2 2,231

4 authors, including:

Elena Nechita Gloria Cerasela Crisan

University of Bacau University of Bacau
74 PUBLICATIONS   250 CITATIONS    69 PUBLICATIONS   646 CITATIONS   

SEE PROFILE SEE PROFILE

Sebastian Damian
Benemérita Universidad Autónoma de Puebla
1 PUBLICATION   2 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Creative Robotic Intelligences View project

iML interactive Machine Learning View project

All content following this page was uploaded by Gloria Cerasela Crisan on 10 May 2021.

The user has requested enhancement of the downloaded file.

Chapter 2
Intelligent Carpooling System
A Case Study for Bacău Metropolitan Area

Elena Nechita, Gloria-Cerasela Crişan, Sergiu-Mădălin Obreja

and Constantin-Sebastian Damian

Abstract Mobility is one of the most basic features of our modern society. People
and goods move around the entire Earth in a continuous and broad attempt to fulfill
economic, safety and environmental goals. The Mobility Management or
Transportation Demand Management is a collection of strategies for encouraging
more efficient traffic patterns towards achieving specific planning objectives. For
example, people can choose to switch from peak hours to non-peak time, or to cycle
instead of using the car. Administrative regulations could introduce incentives or
reimbursements when alternative commuting modes are used. Governmental poli-
cies could include fuel tax increases or pay-as-you-drive freeway taxes or car
insurances. The goal of this chapter is to present several alternative travel modes,
their social impact and their utility. As an example, we present two applications for
shared-use mobility in the metropolitan area of Bacău, Romania. The applications
integrate diverse computing languages with platforms, standards and technologies.
The experimental results are encouraging, allowing us to consider that seamless
integration of hybrid management systems for transportation could have tremen-
dous economic and social impact at global scale.

Keywords Traffic management
Carpooling
Transportation on demand
GPS

data sets

2.1 Introduction

Gathering people into common trips leads to individual and social efficiency. At a
personal level, it reduces the total travelling cost and the driving stress as well.
Although it is less comfortable than using the personal car and people usually need
more time for performing the travel, the broad acceptance of the shared-use

E. Nechita (&)
G.-C. Crişan
S.-M. Obreja
C.-S. Damian

Department of Mathematics, Informatics and Educational Sciences, “Vasile Alecsandri”
University of Bacău, Calea Mărăşeşti 157, 600115 Bacău, Romania
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 43

K. Nakamatsu and R. Kountchev (eds.), New Approaches in Intelligent Control,
Intelligent Systems Reference Library 107, DOI 10.1007/978-3-319-32168-4_2
44 E. Nechita et al.

mobility shows its viability, with practical developments not entirely explored.
From the social point of view, less fuel consumption, less CO2 emission, less traffic
congestion and more social interaction are the benefits of sharing cars or vans. By
saving natural resources like oil, by producing less pollution and by cutting the
travel expense, using a car for many persons is an advance to sustainable trans-
portation systems. As an alternative to public transportation and taxi services,
sharing a car combines the benefits of a shared cost with the flexibility of a taxi ride.
The main idea of shared-use mobility is to share the transportation cost between
multiple participants.
There are many ways people could use the same car or van. Car sharing means
reserving and using a car for a short period of time (usually for several hours) and
returning it before the reservation expires. It is a very convenient option for people
needing rare and short rides: common vehicle costs show that this is the most
efficient decision for less than 10.000 km/year. With an annual fee, and paying by
hour when needing a ride, one could avoid buying a car and paying for its regis-
tration, insurance, maintenance, etc. Carpooling means using the same car for
simultaneous transport of several people from a common starting point to a com-
mon end point. The most usual case is when neighbours work at the same facility
and agree to travel using only one car. Modern systems use the internet and
communication technologies in order to dynamically organise optimal trips. Such
examples are the smartphone apps like Lift, Sidecar or Uber.

2.2 Shared-Use Mobility—Historical Evolution

With more than 70 years of development, the shared-use mobility had different
materializations, addressing specific needs, and using the available resources at that
time:
• During the World War II, the American companies, churches and social associ-
ations were encouraged to “…organise state and local transportation committees
and car sharing clubs” [1] in order to preserve resources for the war (Fig. 2.1a).
After the war, the model of the modern family and the increasing quality of life
brought less concern in sharing cars under institutional frameworks, but instead
self-organised “fampools” (family and friends) became natural.
• The energy crises from the 70s spiked up the gasoline price, so people again
turned to the idea of grouping for common travels. Governments supported such
initiatives through High-Occupancy Vehicle (HOV) lanes, park-and-ride facil-
ities (Fig. 2.1b), or by sponsoring ridesharing projects. A unique project still in
use is the Morgantown Personal Rapid Transit (PRT) system from the West
Virginia University, USA [2]. Small cars are dispatched only at request, but in
crowded hours the system switches to a classic public transportation one. This is
a hybrid system, but is worth mentioned as it is running for almost 40 years,
with a reliability rate of 98.5 %.
2 Intelligent Carpooling System 45

Fig. 2.1 a Car-sharing advertising (World War II). Source Ames Historical Society’s World
War II Veterans Project. b HOV lane sign and free carpool parking sign in Ontario, Canada.
Source http://www.mto.gov.on.ca

• Until the end of the last century, several early systems were organised either by
large-scale employers, or by transportation associations, in order to mitigate the
traffic congestion, to lower the air pollution and to reduce the parking lot strain.
The matches were made by-hand, after collecting data on the employees’
addresses. Local transportation firms supplied vans for longer commutes when
around ten employees came from the same residential area [3] (Fig. 2.2a). For
one-time carpooling, telephone-based ride matching systems were set-up. The
transition to the next systems is made by enhancing the systems through
Personal Digital Assistants (PDAs) and Geographic Information Systems
(GIS) capabilities.
• After 2000, the new communication technologies and the internet broad facil-
ities had great impact on the reliability and on the responsiveness of the
ridesharing systems. The clients manage to post online their commands, and
connected services are now offered to the interested public. For example, the
San Francisco 511 platform offers complex traffic information in the SF Bay

Fig. 2.2 a Vanpool advertising in Vermont, USA. Source Vermont Agency of Transportation.
b Red Nose 2014 advertising in Canada. Source http://www.mto.gov.on.ca
46 E. Nechita et al.

area, including rideshare options [4]. Ecolutis offers aggregate services for firms
and public in France [5]. There are special occasion ride offers, organised by the
local authorities. An example is the Red Nose operation in Canada: the drivers
are suggested to call for a volunteer who can safely drive them home during
December (Fig. 2.2b).
• The current developments of mobile communications provide real-time or
dynamic ridesharing today. The drivers post their trip while they are driving,
and the potential riders warn just before the intentioned departure time. The
mobile application notifies each part’s smartphone about the available pairing,
which can be accepted by a single tap on the smartphone screen. Villefluide is an
online platform for ridesharing in France [6], Ridefinder works globally for
Europe [7]. Successful smartphone apps are Carma, Lift, and Sidecar.
Complimentary rewards are offered to people choosing to use these services. For
example, NuRide computes and reports all its green activities in a specific area
[8] (Fig. 2.3).

Fig. 2.3 NuRide Rhode Island report on December 1st, 2014. Source www.ripta.com
2 Intelligent Carpooling System 47

Dynamic carpooling, also known as casual carpooling because it arranges

shared rides on very short notice, arises supplementary challenges regarding the
matching of drivers and travelers in real time. Adding this request to the flex-
ibility required in routes leads to complicated algorithms, possibly leading to
combinatorial explosion [9]. A review on the available automatic and heuristic
data processing routines to support efficient matching in carpool schemes is
presented in [10]. An application for dynamic carpooling also needs safety
authentication of the rider and driver before making the match. Feedback or
reputation systems may be useful in providing information about who to trust, as
inter-personal constraints (such as punctuality, smoking versus non-smoking,
male versus female preferences) could also interfere [11]. A state of the art on
dynamic carpooling and an identification of the issues against the adoption of
carpooling systems together with some suggestions to solve them are presented
in [12].
For a comprehensive ridesharing classification scheme and for an evolution of
the concept the reader can access the paper of Chan and Shaheen [13].
The following examples furthermore give some details on the functional vehicle
sharing solutions, illustrating the potential of these initiatives, with implications in
traffic flow improvement, public safety and optimisation of movement of people
and goods.
• Car2Go [14] is the Smart division of Daimler AG, offering services in European
and Nord-American cities. The company provides Smart Fortwo highly
energy-efficient vehicles, available through booking online with applications for
mobile devices. Moreover, the system is based on a “one-way” model, which
makes it very flexible. The clients are charged per minute, hour or day. The rates
include renting, fuel costs, insurance, maintenance and parking charges.
Sometimes, a small annual fee may be applied. In May 2014, the company
owned over 10,000 “car2gos” in eight countries, distributed in 26 cities and
servicing more than 700,000 clients. According to [15] the service is providing
electric cars exclusively in USA since November 2011.
• Zipcar [16] is the largest car sharing service, available in USA, Canada, Great
Britain, Spain and Austria. The fleet is diverse, the cars may be driven by hour
or day but they have to be returned to the same reserved parking spot. The driver
pays a registration fee, an annual fee and reservation fees for the regular usage.
The fuel costs, parking, and insurance are included, and facilities are provided
for business and for universities.
• CityCarClub [17] is a UK car sharing network, available in 15 cities. The charge
includes insurance, tax and fuel. The fleet is composed of low emission cars,
experimenting alternative technologies such as biodiesel, hybrid vehicles and
Stop-Start technology [18]. The network members use smart cards to access the
vehicles.
48 E. Nechita et al.

2.3 Conceptual Models for Shared Mobility Systems

The current transportation domain encompasses many stakeholders with hetero-

geneous needs and goals. The Intelligent Transportation System (ITS) must operate
with a large number of concepts, must balance the requests with the available
resources, and are supposed to manage a wide variety of models and implemen-
tations. We expect that the ITSs will bring major social and economic benefits, due
to the greater efficiency of the transport system and increase security. One direction
in ITS research is to design specific ontologies, representing dictionaries of relevant
concepts from traffic domains organised in a hierarchical structure of classes and
taxonomy.
Examples of top-level ontologies that can derive transportation concepts are:
Suggested Upper Merged Ontology (SUMO) [19], OpenCyc [20]. ISO 14825:2011
specifies the data model and physical encoding formats for geographic databases for
ITS applications and services [21]. Semantic web for Earth and Environmental
Terminology (SWEET) is a collection of dedicated top-level ontologies, maintained
by NASA Jet Propulsion Laboratory [22].
By simultaneously treating Intelligent Vehicles and Intelligent Infrastructure, the
low-level ontologies allow comprehensive and targeted approaches to
Transportation problems. The Open Geospatial Consortium [23] offers geospatial
and location standards, providing community-specific OGC Geography Markup
Language (GML) application schemas (AIXM for aviation, GeoSciML for geo-
sciences, etc.). Transportation simulations using ontologies are presented in [24,
25], and specific carpooling simulations based on ontologies are published in [26].
The complexity of the current transportation situations and the need of rapid
actions when sensitive events occur (i.e. bottlenecks, accident, and sudden road
closures) request decision support systems to assist traffic engineers. The
knowledge-based systems are designed to enhance the dynamicity and respon-
siveness, to mitigate the effects of the traffic incidents, and to correctly predict them
[27]. Data analysis or data mining can be used for real-time suggestions for human
operators. One such example is the opening of new lanes when the sensors register
heavy traffic. Another approach is to install mobile applications which can rec-
ommend to the driver new paths for completing his/her daily trip from home to
work.
Conceptual models and simulations are used to get insights on the strengths and
weaknesses of the carpooling systems for a given region. A simulation-based
methodology which emphasizes the construction process of a logic flow diagram
that translates the proposed methodology is presented in [28]. Multiple, heteroge-
neous models have been integrated using ontology techniques [26], especially for
predicting the demand. User-centred research has also been developed. This
approach is helpful for assessing the potential of the carpooling concept in general,
as well as to improve the interaction of the users with the existing web-based
platforms [29]. Statistical testing shows diverging requirements between different
2 Intelligent Carpooling System 49

groups; age, gender and individual economic situation (employment, for example)
are factors that influence carpooling acceptance.
Culture-specificity is also very important. A case study for Switzerland is pre-
sented in [30] and an analysis on Vermont carpool data is available in [31]. The
paper includes some considerations on the perceptions of carpooling, which have
been found to play a larger role than cost or convenience. For Portugal, the work of
[32] reveals the results of a stated preference experiment meant to assess the
enhancement possibilities for carpooling and carpool clubs. For China, where the
private car ownership had exceeded 88 million in 2012, a design of a commute
carpooling system is given in [33] and an investigation of human mobility patterns,
for predicting owners’ driving trajectories and real-time position for carpooling, is
presented in [34]. As taxicab systems play a prominent role in transportation
models, recommendation systems for carpoolable taxicabs come from China, too.
In [35, 36] the authors design such systems, based on a data driven approach.

2.4 Modelling a Carpooling System for Bacău, Romania

This section presents two implementations for carpooling services in the Bacau
area, Romania. The models are based on multi-agent interactions, with specific
characteristics and use multiple communication and representation technologies.

2.4.1 Country Context

Romania is located at the crossroads of many routes connecting Eastern to Western

Europe and Northern to Southern Europe. Of the total National road network,
5,868 km (37.3 %) are classified as European roads [37]. Moreover, Romania’s
location on the transit axes connecting Europe to Asia generated numerous studies
regarding the infrastructure and transport capabilities. The Romanian Government
identified [38] several key transport-related issues:
• Domestic transport, although diversified, has insufficient capacity for trans-
porting freight and passengers;
• The transport infrastructure is insufficiently developed and requires significant
investment in order to meet European standards;
• Access to the West-European corridors, as well as to the Eastern and Southern
Europe ones, is limited;
• The access roads from national roads to town centres and cities are inadequate
and most towns located on National and European roads lack bypasses.
At the end of 2013, more than 5 million privately-owned vehicles have been
recorded in Romania [39]. Nevertheless, the number of new cars has registered a
significant decrease during the last 5 years. This is due mainly to the poor
50 E. Nechita et al.

purchasing power of the population and to the fluctuations of the pollution tax.
Meanwhile, the market of the second-hand vehicles, imported from the Western
Europe, has increased. Under these circumstances, the pollution could also increase.
Romanian law no 3/2001 ratifies the Kyoto Protocol of the United Nations
Framework Convention on Climate Change. According to it, Romania is obliged to
reduce its emissions of greenhouse gases with 8 %, considering the year 1989 as
baseline. If we refer to the emissions from road transport, the average carbon
dioxide emissions from new passenger cars was, in 2013, 132.1 gCO2/km (de-
creasing with 5.22 % compared to 2012) in Romania, while the EU Member States’
average was 126.7 gCO2/km (decreasing with 4.1 % compared to 2012) [40]. The
Romanian government encouraged the rolling stock renewal through national
policies, in order to eliminate the vehicles with significant pollutant capacity.
However, this initiative should be complemented with “soft” measures, such as
intelligent traffic management and transport systems.
The following subsection presents an overview of the data which are relevant for
Bacău, influencing the transportation conditions and the transit system as they stand
at the moment of our study.

2.4.2 City Context

In order to implement a sharing mobility system that is functional and beneficial for
a certain area, specific factors also have to be taken into account. This is the
approach that we have considered for Bacău metropolitan area. Bacău, the 15th
largest city in Romania, is the residence city of Bacău County, part of the historical
region Moldavia and of the North-East Development Region (according to the EU
regional policy [41]). The city is situated at about 300 km from Bucharest, the
capital of Romania. Two European routes cross Bacău: E85 links the South area
with the Northern cities Iaşi and Suceava, while E574 provides access to the
South-Western part, with Braşov, Piteşti and Craiova (the last two are important
centres of the European automotive industry). There is no beltway for Bacău;
therefore the transit through the city often faces congestion, consequently leading to
increased travel time, vehicle operating costs, accidents and environmental damage.
Current data (June, 2014, [42]) show a population slightly over 170,000
inhabitants, distributed within the city limits of 43.1 km2, while the whole county
has more than 706,000 inhabitants. Bacău metropolitan area [43] integrates Bacău
and the neighbouring communities, with a population of around 250,000.
The transportation needs are high and have increased during the last 5 years,
along with the development of SMEs around Bacău. Several divisions of important
international companies are hosted in or in the vicinity of Bacău as well. The rolling
stock has increased. Consequently, the traffic has reached in Bacău a level where
the key performance indicators (especially the indices for traffic efficiency and for
pollution reduction [44]) are alarming.
2 Intelligent Carpooling System 51

For Bacău County, the car ownership has been reported at 111,500 at the end of
2013 [42]. The public transport is based on three major median busways and seven
secondary ones in the city. In addition there are regular buses for the neighbouring
communities. This system is far from providing comfortable connections between
the rural area and Bacău, where most of the employers are concentrated. Thus there
is an important flow of vehicles to and from Bacău, especially during the working
days. An important share of them uses E85 regularly, therefore generating
agglomeration with all its consequences.
The development plan of Bacău [45] follows an integrated approach to the
transport and land use. Key directions include improvement of infrastructure,
management and control of land use citywide and traffic decongestion of the city
centre, but important investments (which are not scheduled in the near future) are
needed. In the central area of the city, roadside parking is limited in location and
duration and is well enforced. The number of off-street parkings is very small.
As far as we know, no shared-use mobility study has been performed for the city
of Bacău. The two implementations which are presented in Chap. 5 aim at
encouraging the authorities to consider new policies regarding the local trans-
portation means and to raise awareness among the local community towards the
green solutions.

2.4.3 Information Requirements for an Efficient Model

In order to incentive the traveller’s choice, a carpooling system has to provide

several features [46], among which we mention:
• Accessibility. The stops need to be located near residential neighbourhoods
and/or in the business area of the city.
• Affordability. The rates should be reasonable; it is important that the system is
available for short trips.
• Convenience. If the vehicles are easy to check in and out at any time, this would
highly encourage their use.
• Reliability. It is recommendable that the vehicles are available and have minimal
failures.
Besides the above listed aspects, the targeted users in Bacău would certainly
require to access the service on smart devices, hence the necessity of a modern,
adequate implementation.
Considering these characteristics, the model we propose focuses mainly on fuel
economy and reduction of transport-related greenhouse gas emissions through an
improved degree of occupancy of the vehicles. The mobility of users in the
metropolitan area and the connectivity to the public transportation systems have
52 E. Nechita et al.

been simulated through the flexibility of the implementations. Therefore, for the
target area we have studied:
• the main transport routes and the most important congestion points
• the most important, necessary stops
• the zones with maximum passengers flow
• traffic monitoring (traffic volume, daily travel data)
• sample data on travel time and speed
• socioeconomic information (workplaces, schools, public institutions and num-
ber of their attendees)
• public transport and parking surveys
• consumers’ satisfaction and perception data on the public transport systems
• environmental impact.
The corresponding data have been provided by the Committee for urbanism,
territorial planning and environmental protection of Bacău County Council [47].
For an efficient carpooling system, the characteristics of the vehicle fleet are also
essential. In order to score improvements resulting from its implementation, fuel
consumption and emissions of CO2 must be reduced. This can be done gradually,
introducing alternative vehicles (such as hybrid electric or fully electric) and/or
fuels (bio- and coal-based, compressed natural gas or liquefied petroleum gas). As
these facilities are scarcely present in Bacău County, we only simulate their
influence by means of two parameters: the average fuel consumption and the
average CO2 emission per car, for the system fleet. As a conclusion, our model
strengthens the role of shifting from privately-owned vehicles to carpooling, and
assesses the reliability of the solution considering the measures suggested in
[48, 49].
According to these goals, the followings represent input data for the model:
• number of main stops
• number of secondary stops
• number of routes
• the set of routes
• number of clients in each main stop, on time intervals
• average fuel consumption
• average greenhouse gas emissions for the vehicles involved in the simulation.
For a given number of clients, the carpooling simulation algorithm allows us to
load a car with 1, 2, 3, 4 or with a random number (between 1 and 4) of clients. For
the chosen routes and stops, the calculations provide as output data:
• number of needed cars (for each of the five variants of loading)
• total fuel consumption
• total greenhouse gas emissions.
2 Intelligent Carpooling System 53

2.5 Applications and Experiments

The applications presented in this section integrate diverse computing languages

with platforms, standards and technologies. They have been used to perform a
preliminary experiment regarding the appropriateness of implementing a carpooling
service in the metropolitan area of Bacău.

2.5.1 SplitCar

The SplitCar application implements the model with Microsoft solutions: the Visual
Studio 2013 suite, the .NET Framework 4.5 package, C# and XML, the components
Windows Presentation Foundation (WPF) for controlling the map delivered by
Bing, and the REST service for locations and routes, offered by Bing Maps.
Given that the majority of the handled data represent locations, distances and
routes, we have chosen as support the Bing Maps tool. In order to use maps, the
Map control will be placed in a WPF window. Besides the Map control, an active
Microsoft account and a Bing key are needed for the use of the map-related ser-
vices: locations finding, distance calculator, route finding. After setting the
Microsoft account and getting the key access to the Bing services, this key is
assigned to the Map control CredentialsProvider attribute:
<WPF:Map CredentialsProvider=’’key’’/>
The map is initialized using the attributes Mode (RoadMode, AerialMode,
AerialModeWithLabels), ZoomLevel, AnimationLevel, Heading, which can be
modified at any time through the controller, using C#. For a complete control over
the map, several functionalities have been implemented in the main window of the
application, as Fig. 2.4 presents.
Several child controls may be added on the Bing map with the command
<map_name>.Children.Add(<control>);
These controls could represent locations, routes (MapPolyline objects), surfaces
(MapLayer objects) or other objects. Only the first two are compulsory for our
model implementation.
The locations are defined through Pushpin type controls, whose attributes allow
the customization desired by the user. In Fig. 2.4, the main stops are coloured in
green and the secondary ones appear in blue. Both types are created with a left-click
of the mouse in the Map control. Right click on a location removes it from the map.
In the main window of the application, several items are designed to offer an
improved control over the map. The user can set the maximum numbers of main
stops (between 2 and 4) and secondary or via-stops (between 0 and 9).
Routes can be defined on the map between any two points, with a MapPolyline
control, according to the following conventions: any route starts in a main stop and
54 E. Nechita et al.

Fig. 2.4 The main window of SplitCar. Routes and stops

ends with a main stop (different from the origin); any route can pass through zero,
one or more secondary stops; routes are displayed on the map with different col-
ours, automatically computed according to the number of routes; the routes take
into account the one-way roads; if a location (part of one or more routes) is deleted,
the user implicitly agrees to delete all the routes that pass through that location/stop.
Any number of routes can be created, and the definition of a route can be cancelled
at any time.
The routes are created when activating the Set routes button (which opens the
dialog window only if at least two main stop are already defined). For the accuracy
of the implementation, we have used the services REST from Bing Maps. These
services receive information on locations and routes, interpret them and display
them on the map. The distance between two points on the map is computed in
kilometres. In Fig. 2.4, the length of the routes appears in the low, right corner.
The implementation of a class that finds a certain location on the map is simple.
The button Search for address opens a window where the user can introduce an
address. After providing it, the class LocationFromAddress returns the corre-
sponding location (latitude and longitude coordinates) on the map:
2 Intelligent Carpooling System 55

class LocationFromAddress
{
private string BaseUrl =
"http://dev.virtualearth.net/REST/v1/Locations?query=";
private string BingAdditional =
"&output=xml&include=queryParse";
private string BingMapsKey = "&key=my_private_key";
private string BingQuery = string.Empty;
private Location point;
public LocationFromAddress();
public void SetQuery(string);
public Location Run();
private Uri AssembleLink();
private string NormalizeQuery(string);
}

Conversely, the class AddressFromLocation returns an address if the location is

given in latitude and longitude:

class AddressFromLocation
{
private string BaseUrl
"http://dev.virtualearth.net/REST/v1/Locations/";
private string BingAdditional =
"?output=xml&includeEntityTypes=
Address&includeNeighborhood=1";
private string BingMapsKey = "&key=my_private_key";
private Location point;
public AddressFromLocation();
public void SetPoint(Location);
public string Run();
private Uri AssembleLink();
}

The next step is to compute the optimal routes and their lengths. The route
between two main stops (with or without intermediary stops) is generated with the
class RouteBetweenTwoPoints, whose components are as follows:

class RouteBetweenTwoPoints
{
public enum MapOptions { Start = 0, End,
ViaPoint, Optimize, TravelMode, Points, Tolerance };
public enum OptimizeRoute { Time = 0, Distance,
TimeAvoidClosure, TimeWithTraffic } ;
public enum TravelMode { Driving = 0, Walking, Transit };
private LocationCollection viaPoints;
private Location Start, End;
private List<string> Options = new List<string>();
private string BaseUrl =
"http://dev.virtualearth.net/REST/v1/Routes?o=xml";
private string BingMapsKey = "&key=my_private_key";
public RouteBetweenTwoPoints();
public string SetRequestParameter(MapOptions, object);
public MapPolyline Run();
public string GetLink();
private Uri AssembleLink();
}
56 E. Nechita et al.

We notice that the function SetRequestParameter receives as first parameter an

object of MapOptions type and holds the information on what needs to be estab-
lished, while the second parameter is an object which depends on the type of the
first parameter. For example, if RouteBetweenTwoPoints is used to define a route,
the function SetRequestParameter sets the following:
• the start and stop points (main stops)
• the intermediary points of the route (the secondary stops, if any)
• the optimisation type (there are four possibilities, to optimise upon: time, dis-
tance, traffic jams avoidance, time and traffic data)
• the travel type (with three options: driving, walking, transit)
• the type of the desired output, indicating if we need the points returned by the
REST service to create the route)
• the tolerance allowed for the points to determine the route.
If all the data have been appropriately setup with SetRequestParameter, the route
is received in a MapPolyline object, with the function Run. Afterwards, the com-
putations are made with the class DistanceBetweenTwoPoints.
For each route and four time intervals (6:00–10:00, 10:00–14:00, 14:00–18:00,
18:00–22:00), a number of persons requiring the carpooling services can be
introduced when activating the button Input data, as in Fig. 2.5. Moreover, the
average fuel consumption and the average greenhouse gas emissions are required,
per private, regular car and per carpooling vehicle.
When all the input data are available, the module RouteData performs the
computations of the output information (total fuel consumption and greenhouse gas
emissions) for a certain route. The data are presented in several formats: text, tables
and histograms, according to the users’ choice, when activating the button Show
model calculations. An example is given in Fig. 2.6, where the value “5” on the

Fig. 2.5 Data acquisition window for SplitCar

2 Intelligent Carpooling System 57

Fig. 2.6 A diagram for SplitCar output data

horizontal axis encodes what was denoted as “random setting” in Table 2.2, namely
a random number (between 1 and 4) of occupants in a car. The total fuel con-
sumption is measured in litres.

2.5.2 Buddy

The Buddy application is a web-based solution. We have chosen to use ASP .NET
for the server-side, JavaScript, JQuery, Ajax as programming languages, HTML,
CSS and other applications for improving the design of the web interface
(JQueryUI, Bootstrap, jChartFX), all of them embedded in Visual Studio. We also
used the API provided by Google for the well-known mapping service Google
Maps. The API also allows the use of maps stored on sites of third parties, and
includes a locator for organisations (as well as other objectives) placed on the
territories of numerous countries all over the world.
58 E. Nechita et al.

Fig. 2.7 The interface of Buddy

The interface of Buddy is simple (Fig. 2.7), with two main zones. For its
dynamic configuration (resizing in accordance with the browser window or with the
device display) we have used the Bootstrap style class collection. The right zone is
a container that includes the Google Maps API. This API must be included in the
header of the web page with:

<script
type="text/javascript"
src="https://maps.googleapis.com/maps/
api/js?sensor=false">
</script>

and an object map is defined with:

var mapOptions =
{
mapTypeControl: false,
mapTypeId: google.maps.MapTypeId.ROADMAP,
center: new
google.maps.LatLng(46.571289, 26.925170999999978),
zoom: 14 //zoom 14 times on Bacau
};
map = new google.maps.Map(
document.getElementById("#map-canvas"),
mapOptions);
geocoder = new google.maps.Geocoder();

The left zone of the interface uses the accordion widget from the JQuery UI
package and offers access to three tabs: Location, Map type, and Routes.
2 Intelligent Carpooling System 59

The packages jQuery and jQuery UI have to be included in the header of the web
page with:

<head runat="server">
<script src="jquery-1.9.0.min.js" type="text/javascript">
</script>
<script src="jquery-ui-1.10.4.js" type="text/javascript">
</script>
<link href="jquery.ui.accordion.css"
rel="stylesheet" type="text/css" />
</head>

As the widgets in jQuery UI use the jQuery classes, it is compulsory that the
order of the lines in the above sequence code is maintained.
The user can search for a new location, either introducing its name or its
coordinates. When searching for an address upon its name, the function findAdress()
gets the input text and (if this is consistent) uses the geocode method, as follows:

geocoder.geocode(
{ 'address': address }, function (results, status)
{
if (status == google.maps.GeocoderStatus.OK)
{
map.setCenter(results[0].geometry.location);
$("#location-coordinates").html
"Lat: " + map.getCenter().lat() + ", Lng: " +
map.getCenter().lng());
}
else
alert('Geocode was not successful
for the following reason: ' + status);
});

If the corresponding search is successful and the function returns a list of

locations, the map is centred in the first result, which is considered to be the most
significant. When searching for an address upon its coordinates, the map is centred
in the point defined by the specified coordinates (if they are correct) with
map.setCenter(new google.maps.LatLng(lat, lng));
60 E. Nechita et al.

Fig. 2.8 A hybrid view over Bacău, with Google Maps

With Map type, the user changes the view, choosing among: Roadmap (as in
Fig. 2.7), Satellite, Hybrid (as in Fig. 2.8) or Terrain.
The tab Routes is the most complex. It provides functionalities that allow to
create and to modify routes on the Google map. When the web application is
launched, an Ajax function calls an ASP.NET function from the server. It returns (if
any) the routes previously saved in the database. The results provided by this
function appear to the user as in Fig. 2.7, and are stored in a JavaScript object of
JSON type. For the routes displayed in Fig. 2.7, the JSON object is as follows:

46.58299327827027-26.912510097026825/46.56989553861151-
26.91491335630417;
46.57880470562833-26.930191218852997/46.57160668424229-
26.92186564207077/46.56328646931478
26.922981441020966/46.56912845580439-
26.910965144634247/46.56706317894082-26.903841197490692;
46.57302276393836-
26.90118044614792/46.556971695640904-26.911823451519012;
2 Intelligent Carpooling System 61

The function parseExistingRoutes receives the above string and decomposes it in

values representing the coordinates of the routes:

function parseExistingRoutes(string)
{
string = string.substring(0, string.length - 1);
var routes_strings = string.split(';');
for (var i = 0; i < routes_strings.length; i++)
{
var route_string = routes_strings[i].split('/');
var lcs = new Array();
for (var j = 0; j < route_string.length; j++)
{
var locations_strings =
route_string[j].split('-');
lcs[j] = new Object();
lcs[j].lat = locations_strings[0];
lcs[j].lng = locations_strings[1];
}
r++;
showRoute(lcs, function (distance) {});
}
}

For each location, an object such as

{ lat: 46.57302276393836, lng: 26.90118044614792 }

is created. For each route, the application generates a vector of objects. This is
further sent to the function showRoute() which recomposes the route. The first
element of the vector always becomes the start point and the last one is seen as the
destination. If the route is made of more than three locations, the intermediary ones
become via-stops:

var start = new google.maps.LatLng(ls[0].lat, ls[0].lng);

var end = new google.maps.LatLng(ls[len - 1].lat,
ls[len - 1].lng);
var waypoints = [];
for (var i = 1; i < len-1; i++)
waypoints.push({location: new google.maps.LatLng
ls[i].lat, ls[i].lng), stopover: true});

where the function google.maps.LatLng(x,y) creates Google objects of type point.

62 E. Nechita et al.

To display a route, the algorithm finds the shortest path that connects the starting
point with the destination, preserving the position and the order of the intermediary
locations. This is done with a request object, which is further used as a parameter
for the functions in the Google API:

var request =
{
origin: start,
destination: end,
waypoints: waypoints,
optimizeWaypoints: true,
travelMode: google.maps.TravelMode.DRIVING
};

Each point is identified with a Marker type object on the Google map. Ten
images (for the marker objects) in ten different colours are stored in the resources
file of the system, to enhance the routes readability on the map. The marker objects
are initialised with

var icon = "Resources/marker" + r%10 + ".png";

for (var i = 0; i < len; i++)
{
var marker = new google.maps.Marker({
position: new google.maps.LatLng(ls[i].lat, ls[i].lng),
map: map,
icon: new google.maps.MarkerImage(icon)});
newmarkers.push(marker);
}

where r is the index of the route.

The computations and the route display are done through a call of the Google
API:

var directionsDisplay =
new google.maps.DirectionsRenderer({
polylineOptions:
{strokeColor: colors[(r - 1)%10] },
suppressMarkers: true, preserveViewport: true });
directionsDisplay.setMap(map);

which specifies the desired colour for the route, cancels the standard markers dis-
play and indicates no zoom over the created route.
2 Intelligent Carpooling System 63

If the call is successful, the route is displayed (with the customized markers) and
its length (in kilometres) is computed:

if (status == google.maps.DirectionsStatus.OK)
{
directionsDisplay.setDirections(response);
for (var i = 0; i < markers.length; i++)
if (markers[i]) markers[i].setMap(null);
var thisroute = response.routes[0];
var distance = 0;
for (var i = 0; i < thisroute.legs.length; i++)
distance += thisroute.legs[i].distance.value;
callback(distance / 1000);
callback(response.routes[0].legs[0].distance.value/1000);
}

The application offers to the user the possibility of defining his/her own routes,
when activating Create new route. This option calls the function createRoute(),
whose main functionalities are to place markers in the desired locations and to reuse
the already existing markers (both with right-click of the mouse). Instances of
Listener type objects of the API are instantiated to activate these facilities:

listener1 = google.maps.event.addListener
map, 'rightclick', function (event)
{
var marker=
new google.maps.Marker
({position:event.latLng,map:map,title:'Marker-'+(l+1)});
markers[l] = marker;
var loc = new Object();
loc.lat = event.latLng.lat().toString();
loc.lng = event.latLng.lng().toString();
locations.push(loc);
…

for (var i = 0; i < newmarkers.length; i++)

newmarkers_listeners[i]=
google.maps.event.addListenerOnce
newmarkers[i],'rightclick',
function ()
{
var marker =
new google.maps.Marker
({ position: this.getPosition(), map: map, title:
'Marker-' + (l + 1) });
markers[l] = marker;
var loc = new Object();
loc.lat = this.getPosition().lat().toString();
loc.lng = this.getPosition().lng().toString();
locations.push(loc);
…
64 E. Nechita et al.

Fig. 2.9 Defining a route to visit the locations defined by user markers

The chosen locations are stored in an array and their coordinates are displayed in
a temporary table, as in Fig. 2.9. The user can remove some locations, if these are
misplaced. When the configuration is final, the route can be saved.
The system performs an ajax call of the function saveRoute() on the server:

$.ajax({
type: "POST",
url: "Default.aspx/saveRoute",
data: "{jsonstring:'" + JSON.stringify(locations) + "'}",
contentType: "application/json; charset=utf-8",
dataType: "json",
success: function (msg)
{
…

The array locations is sequenced and sent as a JSON object. After saving the
route, the right-click events detection over the markers and the over the map is
stopped. Moreover, the route length (returned by the function showRoute() above)
is also saved (Fig. 2.10) through an ajax call of the function saveRouteDistance()
on the server:

$.ajax({
type: "POST",
url: "Default.aspx/saveRouteDistance",
data: "{jsonstring:'" + route_id + "-" + distance + "'}",
contentType: "application/json; charset=utf-8",
dataType: "json",
success: function (msg)
{
…
2 Intelligent Carpooling System 65

Fig. 2.10 The definition of the route is completed

The simulation starts when activating the button Start model simulation, which
opens the windows where the user presents:
• the parameters (the same as in the first application: number of passengers, the
average fuel consumption and the average greenhouse gas emissions per
privately-owned car and per carpooling vehicle);
• data for each route: number of travellers on that route, per day and time
intervals.
Both applications presented here are available at [50], where we have collected a
series of applications designed to illustrate the role of Informatics in developing
intelligent systems to support the sustainable development of our society. The
results presented in the following Section were obtained on a desktop PC with a
2 GHz quad-core processor and 2 GB RAM capacity.

2.6 Results and Discussion

The application Buddy generates, for each route, a table (see Table 2.1) where the
user can compare the figures related to each type of mobility setting, defined by the
column Persons per car (4). The lines shadowed in grey correspond to the mean
value of the vehicles loading, determined as the average of the sequence of random
numbers (between 1 and 4) whose sum cover the travel request. The values in
column (7), computed as the product between the route lengths, the number of cars
and the quantity of CO2 emissions per car and per kilometre, have been rounded to
the nearest integer.
66 E. Nechita et al.

Table 2.1 Comparison between five mobility choices

The data in columns (6) and (7) can also be seen in charts, built with jChartFX
support. The package jChartFX (which requires the presence of jQuery tools) must
be included in the header of the web page:

<script src=
"jchartfx.advanced.js" type="text/javascript"></script>
<script src=
"jchartfx.system.js" type="text/javascript"></script>
<script src=
"jchartfx.coreVector.js" type="text/javascript"></script>

The simulation has been run for a flow of 10,000 persons (which represents an
estimation for the number of the commuters in a working day for Bacău city) and
for an average route of 12 km length per car and per day. We used various routes
2 Intelligent Carpooling System 67

Table 2.2 Sample output data

Number Total fuel Total CO2 Fuel economy and
of needed consumption emissions CO2 emissions
cars (l) (kg) decrease (%)
(a) Personal ownership 5,000 4,200 7,800 –
(2 persons per car)
(b) Carpooling (1 to 4 3,900 3,276 6,084 22.00
persons per car;
random setting)
Set 1 of parameters
(c) Carpooling (1 to 4 3,900 2,340 4,680 44.28
persons per car;
random setting)
Set 2 of parameters

and two sets of parameters. Set 1 models the case when the cars used for carpooling
have the same characteristics as the regular ones. As most of the passenger cars
used in Romania complies with the Euro 4 emission standard, we have set the fuel
consumption per 100 km at 7 L (of gasoline) and the CO2 emissions at 130 g/km
[39]. Set 2 simulates a modern carpooling fleet, using vehicles with superior per-
formances, for which the two parameters were set to 5 L/100 km and 100 g/km,
respectively.
Table 2.2 presents a sample of output data, which display substantial fuel sav-
ings and important CO2 emissions decrease. Although the applications allow the
user to simulate a vehicle loading with any number (from 1 to 4) of passengers, the
table includes only the lines for two persons per car and for the random loading,
which is the most plausible situation to happen in real life. For the computation in
the last column, the reference line is (a), for both (b) and (c).
If we consider an average price of 1.1 euros per gasoline litre in 2014 [40], we
get a daily saving of 1,016 euros for case (b) and 2,046 euros for case (c).
According to the SWOT analysis performed by Bacău County Council [45], the
investments of the inhabitants in health-care and education decreased during the last
years. The savings could be directed towards these domains, which could improve
the quality of life, both at personal and community level. Moreover, the impact of
the shared mobility services on the environment must not be disregarded. The
results of the simulations are represented in Figs. 2.11 and 2.12.
Despite these encouraging results, we should question the attitude of the public
in Bacău metropolitan area towards the carpooling option. A study on this issue is
part of our future work, but preliminary research show that there are both psy-
chological barriers and lack of involvement from the local stakeholders regarding
this mobility solution.
68 E. Nechita et al.

Fig. 2.11 The total CO2 emissions for single-driver case (indigo) and for random number of
passengers case (grey)

Fig. 2.12 The total fuel consumption for single-driver case (indigo) and for random number of
passengers’ case (grey)

2.7 Conclusions and Further Work

Transportation activities have major adverse impact on most of the urban

agglomerations, as the traffic is more and more busy and the pollution becomes
difficult to mitigate.
The cars cannot be eliminated from our lives but the unwanted effects of their
extensive use could be reduced through intelligent strategies. There are several
solutions for that: ecological car design, efficient metropolitan networks, multi-
modal planning etc. Among them, car-sharing and carpooling are two modern
2 Intelligent Carpooling System 69

concepts and should be put forth; specific solutions and facilities should be
developed, as both are cost-effective and social rewarding.
Car-sharing and carpooling become more effective and more attractive when
seen as an alternative means of transport that can be used together with the personal
vehicles and/or with the public transport systems. Practically, all the
shared-mobility companies started by manually managing their services: the users
place a booking of a vehicle to a human operator, get the key and register the
personal data in a form stored in the car. As the programmes have extended their
fleet rapidly, the manual-operated systems became expensive and inconvenient,
generating errors. Automated reservations, key management and billing offer make
the share-used solutions more attractive and effective.
In Europe, car-sharing and carpooling have already proved to be effective in
several countries, including: Austria, Germany, Switzerland and Netherlands. In
Romania, the participation of several municipalities in European projects (such as
Bucharest in the CIVITAS TELLUS—Transport and Environment Alliance for
Urban Sustainability [51], Sibiu and Timişoara in TRANSPOWER—Supervised
Implementation of Sustainable Urban Transport Concepts [52], Suceava, in
MIDAS—Measures to Influence transport Demand to Achieve Sustainability [53],
Brăila in PILOT—Planning Integrated Local Transport [54]) and organisations
(such as The Romanian Public Transport Association in LINK—The European
forum on intermodal passenger travel [55]) started to implement several forms of
ICT-based intelligent mobility solutions.
There is a ride sharing community and several initiatives that Romania joined,
such as [56, 57]. EkoRoad [58] and 4 in masina [59] are two Romania-based
car-sharing/carpooling services available on mobile devices. The former has
expanded its area at European level [60].
In Romania, the market share of electric and hybrid vehicles is now less than
0.01 % [61]. A single public charging point is available in Bucharest (however, it is
estimated [62] that in 2020, the market share will reach over 5 %). Fuel prices have
increased in recent years and their evolution is now uncertain. As the development
of infrastructure is not expected soon, the city of Bacău could significantly benefit
from the implementation of several “soft” measures for improving its transportation
capacity, efficiency and comfort (see [63] for a review on Clean Transport Systems
and related issues). Bike-sharing (which will be implemented in the future) and
car-sharing/carpooling are such measures which may have a high benefit-cost ratio.
Our future work will focus on developing a carpooling application for mobile
devices. The already implemented features (such as routes definition, or operation
on maps) will be preserved and some more will be added, such as finding the
nearest vehicle available on the way, therefore approaching a dynamical perspective
designed to fit the users’ needs and to optimise the efficiency of the shared system.
Another research direction that we will investigate refers to the use of ACO
metaheuristic [64, 65] for solving the Daily Car-Pooling Problem and to the
assessment of this metaheuristic performance when compared to the results in [66].
70 E. Nechita et al.

Acknowledgments This research was supported by the project “Bacău and Lugano—Teaching
Informatics for a Sustainable Society”, co-financed by Switzerland through the Swiss-Romanian
Cooperation Programme to reduce economic and social disparities within the enlarged European
Union.

References

1. Coates, A.: Guide to victory. Popular Gov. 9(1–4), (1943). https://archive.org/details/

populargovernmen914inst. Accessed Nov 2014
2. Transportation and Parking, West Virginia University. http://transportation.wvu.edu/prt.
Accessed Nov 2014
3. DART’s RideShare program site. http://www.ridedart.com/services/rideshare. Accessed Nov
2014
4. San Francisco Bay 511 program site. http://www.511.org. Accessed Nov 2014
5. Covouiturage Ecolutis site. http://www.ecolutis.com. Accessed Nov 2014
6. Villefluide project site. http://www.villefluide.fr. Accessed Nov 2014
7. Carpooling and Ridesharing across Europe site. http://www.ridefinder.eu. Accessed Nov 2014
8. NuRide project site. http://www.nuride.com. Accessed Nov 2014
9. Knapen, L., Yasar, A., Cho, S., Keren, D., Dbai, A.A., Bellmans, T., Janssens, D., Wets, G.,
Shuster, A., Sharfman, I., Bhaduri, K.: Exploiting graph-theoretic tools for matching in
carpooling applications. J. Ambient Intell. Humaniz. Comput. 5(3), 393–407 (2014)
10. Ferrari, E., Manzini, R., Pareschi, A., Persona, A., Regattieri, A.: The car pooling problem:
Heuristic algorithms based on saving functions. J. Adv. Trans. 37, 243–272 (2003)
11. Nagare, D.B., Moore, K.L., Tanwar, N.S., Kulkarni, S.S., Gunda, K.C.: Dynamic carpooling
application development on Android platform. Int. J. Innov. Technol. Explor. Eng. 2(3),
136–139 (2013)
12. Graziotin, D.: An analysis of issues against the adoption of dynamic carpooling. http://arxiv.
org/abs/1306.0361 (2010)
13. Chan, N.D., Shaheen, S.A.: Ridesharing in North America: past, present and future.
Transp. Rev. 32(1), 93–112 (2012)
14. Austin car2go site. https://www.car2go.com. Accessed Nov 2014
15. U-T San Diego News site. http://www.utsandiego.com. Accessed Nov 2014
16. Zipcar car sharing project site. http://www.zipcar.com. Accessed Nov 2014
17. City Hire Club site. http://www.citycarclub.co.uk/. Accessed Nov 2014
18. DeBruin, L.A.: Energy and feasibility analysis of gasoline engine start/stop technology.
Undergraduate Honors Thesis, Ohio State University (2013). https://kb.osu.edu/dspace/
bitstream/handle/1811/54555/Start-Stop_Thesis.pdf. Accessed Nov 2014
19. Suggested Upper Merged Ontology site. http://www.adampease.org/OP/
20. Open Cyc Platform site. http://www.cyc.com/platform
21. ISO 14825:2011 site. http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detail.
htm?csnumber=54610
22. SWEET Overview site. http://sweet.jpl.nasa.gov/
23. Open Geospatial Consortium site. http://www.opengeospatial.org/
24. Becker, M., Smith, S.: An ontology for multi-modal transportation planning and scheduling.
Tech. report CMU-RI-TR-98-15, Robotics Institute, Carnegie Mellon University (1997)
25. Houda, M., Khemaja, M., Oliveria, K., Abed, M.: A public transportation ontology to support
user travel planning. In: Loucopoulos P., Cavavero J.L. (eds.) Proceedings of the International
Conference on Research Challenges in Information Science, RCIS’2010 19-21, Nice, France
(2010)
2 Intelligent Carpooling System 71

26. Cho, S., Kang, J.I., Yasar, A.U.H., Knapen, L., Bellemans, T., Janssens, D., Wets, G., Hwang,
C.S.: An activity-based carpooling microsimulation using ontology. procedia computer
science. In: The 4th International Conference on Ambient Systems, Networks and
Technologies (ANT 2013), The 3rd International Conference on Sustainable Energy
Information Technology (SEIT-2013), vol. 19, pp. 48–55 (2013)
27. Dimitrakopoulos, G., Bravos, G., Nikolaidou, M., Anagnostopoulos, D.: Proactive,
knowledge-based intelligent transportation system based on vehicular sensor networks.
Intell. Trans. Syst. 7(4), 454–463 (2013)
28. Correia, G., Viegas, J.M.: A conceptual model for carpooling systems simulation. J. Simul. 3,
61–68 (2009)
29. Wilkowska, W., Farrokhikhiavi, R., Ziefle, M., Vallée, D.: Mobility requirements for the use
of carpooling among different user groups. In: Ahram, T., Karwowski, W., Marek, T. (eds.)
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics
(AHFE 2014) (2014)
30. Ciari, F.: Why do people carpool? Results from a Swiss survey. Swiss Transport Research
(2012). http://www.strc.ch/conferences/2012/Ciari.pdf
31. Watts, R., Belz, N., Fraker, J., Gandrud, L., Kenyon, J., Meece, M.: Increasing carpooling in
Vermont: opportunities and obstacles. University of Vermont Transportation Research Center
(2010). http://www.uvm.edu/*transctr/trc_reports/UVM-TRC-10-010.pdf
32. Correia, G., Viegas, J.M.: Carpooling and carpool clubs: clarifying concepts and assessing
value enhancement possibilities through a stated preference web survey in Lisbon, Portugal.
Trans. Res. Part A 45, 81–90 (2011)
33. Zhou, G., Huang, K., Mao, L.: Design of commute carpooling based on fixed time and routes.
Int. J. Veh. Technol. (2014). http://www.hindawi.com/journals/ijvt/2014/634926/
34. Liu, N., Feng, Y., Wang, F., Liu, B., Tang, J.: Mobility crowdsourcing: toward zero-effort
carpooling on individual smartphone. Int. J. Distrib. Sens. Netw. (2013). http://www.hindawi.
com/journals/ijdsn/2013/615282/
35. Zhang, D., He, T., Liu, Y., Stankovic, J.A.: CallCab: a unified recommendation system for
carpooling and regular taxicab services. In: Proceedings of the 2013 IEEE International
Conference on Big Data, pp. 439–447 (2013)
36. Zhang, D., Li, Y., Zhang, F., Lu, M., Liu, Y., He, T.: coRide: carpool service with a win-win
fare model for large-scale taxicab networks (2013). http://www-users.cs.umn.edu/*tianhe/
Papers/coRide.pdf
37. Asakaite, I., Celik, B.: Mapping of logistics infrastructure of central and Eastern Europe for
automotive Industry. Master Thesis, Göteborg University (2006). https://gupea.ub.gu.se/
bitstream/2077/9603/1/2006_72.pdf
38. Government of Romania. Ministry of transport (2007) Sectorial operational programme
transport 2007–2013. http://www.fonduri-ue.ro/res/filepicker_users/cd25a597fd-62/Doc_prog/
prog_op/5_POST/2_POST_Eng.pdf. Accessed Nov 2014
39. Romania—Open datasets site. http://date.gov.ro/. Accessed Nov 2014
40. European Environment Agency (2014) monitoring CO2 emissions from passenger cars and
vans in 2013. EEA Technical report 19/2014. http://www.eea.europa.eu/publications/
monitoring-co2-emissions-from-passenger. Accessed Nov 2014
41. Inforegio EU Regional Policy site. http://ec.europa.eu/regional_policy/index_ro.cfm.
Accessed Nov 2014
42. National Institute of Statistics (2014). http://www.insse.ro. Accessed Oct 2014
43. European Statistics site. http://ec.europa.eu/eurostat. Accessed Oct 2014
44. Kaparias, I., Bell, M.G.H.: Key performance indicators for traffic management and
intelligent transport systems. 7th framework programme, contract no. 218636 (2011). http://
www.polisnetwork.eu/publicdocuments/download/516/document/d-3-5—key-performance-
indicators-for-traffic-management-and-its-final.pdf. Accessed Nov 2014
45. Bacău County Council (2014) strategy for sustainable development of Bacău county for the
period 2010–2030. http://www.bids-see.ro/. Accessed Nov 2014
72 E. Nechita et al.

46. European Commission, DG Enterprise & Industry (2009) The potential of intelligent transport
systems for reducing rod transport related greenhouse gas emissions. A Sectorial e-Business
watch study by SE Consult, Final Report. http://ec.europa.eu/enterprise/archives/e-business-
watch/studies/special_topics/2009/documents/SR02-2009_ITS.pdf. Accessed Nov 2014
47. Bacău County Council site. http://www.csjbacau.ro. Accessed Nov 2014
48. Millard-Ball, A., Murray, G., Schure, J., Fox, C., Burkhardt, J.: TCRP Report 108:
Car-sharing: Where and How it Succeeds. Transportation Research Board of the National
Academies, Washington DC (2005)
49. Shaheen, S.A., Wright, J., Sperling, D.: California’s zero-emission vehicle mandate: linking
clean-fuel cars, car sharing, and station car strategies. Trans. Res. Rec. 1791, 113–120 (2001).
TRB Paper 02-3587, Institute of Transportation Studies, UC Davis
50. BLISS project site. http://www.bliss.ub.ro. Accessed Nov 2014
51. Transport and Environment Alliance for Urban Sustainability (CIVITAS TELLUS project).
http://www.civitas.eu/content/tellus. Accessed Nov 2014
52. Transport Research and Innovation portal. http://www.transport-research.info/web/projects/
project_details.cfm?id=11355. Accessed Nov 2014
53. Midas Europe project site. http://www.midas-europe.com. Accessed Nov 2014
54. PILOT—Planning Integrated Local Transport. http://www.pilot-transport.org. Accessed Nov
2014
55. TRANSPOWER—Supervised Implementation of Sustainable Urban Transport Concepts.
http://www.transport-research.info/web/projects/project_details.cfm?id=35675. Accessed Nov
2014
56. Roadsharing services project site. http://www.roadsharing.com. Accessed Nov 2014
57. Carpool Rideshare project site. http://www.carpoolworld.com. Accessed Nov 2014
58. Ekoroad travel project for Romania site. http://www.ecodrum.ro. Accessed Nov 2014
59. 4 in masina project site. http://4inmasina.ro. Accessed Nov 2014
60. Ekoroad travel project site. http://www.ekoroad.com. Accessed Nov 2014
61. Varga, B.O.: Electric vehicles, primary energy sources and CO2 emissions: Romanian case
study. Energy 49, 61–70 (2013)
62. APIA (2014) Romanian association of the vehicle producers and importers. http://apia.ro.
Accessed Nov 2014
63. European Commission, Directorate-General for Mobility and Transport (2011). Study on clean
transport systems. Project 2010 final report. http://ec.europa.eu/transport/themes/urban/studies/
doc/2011-11-clean-transport-systems.pdf. Accessed Nov 2014
64. Crişan, G.C., Nechita, E.: Solving TSP with Ant algorithms. J. Comput. Commun. Control 3,
228–231 (2008)
65. Pintea, C.M., Crişan, G.C., Chira, C.: Hybrid Ant models with a transition policy for solving a
complex problem. Log. J. IGPL 20(3), 560–569 (2012)
66. Swan, J., Drake, J., Özcan, E., Goulding, J., Woodward, J.: A comparison of acceptance
criteria for the daily car-pooling problem. Comput. Inf. Sci. III, 477–483 (2013)
http://www.springer.com/978-3-319-32166-0

View publication stats