Content uploaded by Chengbin Chu
Author content
All content in this area was uploaded by Chengbin Chu on Jan 22, 2024
Content may be subject to copyright.
HAL Id: hal-02014342
https://hal.science/hal-02014342
Submitted on 11 Feb 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
A survey of models and algorithms for optimizing shared
mobility
Abood Mourad, Jakob Puchinger, Chengbin Chu
To cite this version:
Abood Mourad, Jakob Puchinger, Chengbin Chu. A survey of models and algorithms for optimizing
shared mobility. Transportation Research Part B: Methodological, 2019, Volume 123 (123), pp.323-
346. �10.1016/j.trb.2019.02.003�. �hal-02014342�
A survey of models and algorithms for optimizing shared mobility
Abood Mourada,b, Jakob Puchingera,b, Chengbin Chuc
aLaboratoire G´enie Industriel, CentraleSup´elec, Universit´e Paris-Saclay, Gif-sur-Yvette, France
bInstitut de Recherche Technologique SystemX, Palaiseau, France
cUniversit´e Paris-Est, ESIEE Paris, D´epartement Ing´enierie des Syst`emes, Noisy-le-Grand, France
Abstract
The rise of research into shared mobility systems reflects emerging challenges, such as rising traffic con-
gestion, rising oil prices and rising environmental concern. The operations research community has turned
towards more sharable and sustainable systems of transportation. Shared mobility systems can be collapsed
into two main streams: those where people share rides and those where parcel transportation and people
transportation are combined. This survey sets out to review recent research in this area, including different
optimization approaches, and to provide guidelines and promising directions for future research. It makes
a distinction between prearranged and real-time problem settings and their methods of solution, and also
gives an overview of real-case applications relevant to the research area.
Keywords: Optimization, passenger and freight transportation, prearranged and real-time ridesharing,
exact and heuristic methods.
1. Introduction
The concept of shared mobility has gained popularity in recent years, attracting attention from the
operations research community, especially after the world of transportation witnessed a mini-revolution
with the launch of shared mobility services like V´
elib, Autolib, Zipcar, Car2Go and others. Emerging
challenges, such as growing levels of traffic congestion and limited oil supplies with their increasing prices,
together with the rising environmental concerns have pushed research towards more sharable and sustainable
systems of transportation. Applying this sharing concept in real-life transportation systems is expected to
afford a set of potential benefits, whether for people sharing their daily trips or for combined passenger and
freight transportation.
Shared mobility comes with many benefits, such as decreasing congestion and pollution levels and
reducing transportation costs for both people and goods, but it also has challenges that are holding back
widespread adoption. Furuhata et al. (2013) identified three major challenges for agencies providing shared
rides to passengers. These are: designing attractive mechanisms, proper ride arrangement, and building
trust among unknown passengers in online systems. Thus, in order to be adopted more widely, a shared
mobility service should be easy to establish and provide a safe, efficient and economical trip. As such, it
should be able to compete with the immediate access to door-to-door transportation that private cars provide
(Agatz et al. (2012)).
Email addresses: abood.mourad@centralesupelec.fr (Abood Mourad), jakob.puchinger@centralesupelec.fr
(Jakob Puchinger), chengbin.chu@esiee.fr (Chengbin Chu)
Preprint submitted to Elsevier December 14, 2018
Another important aspect is the emergence of autonomous mobility services and their potential applica-
tion to existing shared mobility systems. Fully autonomous vehicles are expected to reduce traveling costs
and provide a safer and more comfortable and sustainable mode of transportation (Meyer et al. (2017)). If
those assumptions translate to reality, autonomous vehicles will dramatically change the urban landscape,
and if they can be used as a shared transportation service, they could reshape the future of shared mobility
systems (Chen et al. (2016b)).
From a logistics system perspective, swiftly-growing urbanization rates, and consequently the potential
change in people’s demands for goods in urban areas, justify the need to develop new urban logistics sys-
tems. These new systems should ensure efficient urban mobility, not just for people, but for goods as well
(Fatnassi et al. (2015)). Thus, much of the recent research has focused on increasing the sustainability of
mobility systems. Projects have focused on improving existing transportation systems and service quality
and designing new systems that can offer a more sustainable and ecological approach and thus contend with
rising urban challenges. One innovative idea is to combine individual freight and passenger transportation
streams in an urban area, prompting efforts to study the efficiency gains made when people and goods share
rides and identify the potential challenges facing this combination.
The increasing need for new technologies and services that support the development of sustainable and
innovative shared mobility systems is coupled with the need to develop new operations research models and
optimization approaches. An increasing amount of research is thus directed towards building new models
and methods that can efficiently operate these systems. Reviewing the literature on shared mobility systems
for passenger transportation, Furuhata et al. (2013) surveyed the existing ridesharing systems and identified
their key challenges. The paper also classified these systems according to their different features, matching
search strategies, pricing methods and target demand segments. Agatz et al. (2012) surveyed the different
operations research models that allow travelers (drivers and riders) to be matched in real-time, and reviewed
the optimization challenges that arise in such real-time systems and the methods used to operate them. A
more recent survey by Molenbruch et al. (2017) reviewed the literature on demand-responsive ridesharing
systems, called dial-a-ride problems (DARPs). The authors introduce a taxonomy classifying the reviewed
papers according to their real-time characteristics, service design, and solution methods. Similarly, Ho et al.
(2018) presented an up-to-date review of recent studies on dial-a-ride problems (DARPs) with their different
variants and solution methodologies. Moreover, the paper introduced references to benchmark instances,
investigates their application areas, and suggests directions for future research. City logistics is a major field
of innovation in freight transportation, so the rising importance of sharing aspects in last mile distribution
makes it equally important to investigate the latest developments in city logistics. Savelsbergh and Van
Woensel (2016) reviewed the most recent trends and challenges in city logistics and identified opportunities
for research. Sampaio Oliveira et al. (2017) studied the crowd-sourcing logistics model, which aims to
use available capacity on trips already taking place, called the crowd, to deliver goods in urban areas. The
paper reviewed the latest developments in crowd logistics along with their different features, applications,
deployment issues and impacts on city logistics.
Whereas these reviews on shared mobility have focused on either people or freight transportation con-
sidering one variant of the problem (dynamic ridesharing systems and carpooling services (Agatz et al.
(2012); Furuhata et al. (2013)), DARPs (Molenbruch et al. (2017); Ho et al. (2018)), city logistics (Savels-
bergh and Van Woensel (2016)), crowd-logistics (Sampaio Oliveira et al. (2017)) and other variants), here
we review different variants of the shared mobility problem for both people and goods. We thus focus on
shared mobility systems where (i) travelers share their rides to reduce travel costs, usually called rideshar-
ing systems, or where (ii) passenger and freight transportation are combined. We find that although the
different variants can share similar modeling features, formulations and solution approaches, their context
2
of application varies. For example, A DARP-based formulation can be used to model both types of shared
mobility systems, but some of its features can vary depending on the context in which it is applied. We
thus study these variants according to their modeling choices, defining features and solution methods, and
we identify their common and varying characteristics. This survey brings several key contributions: (1) a
comprehensive overview of recent papers on shared mobility for transportation of people and goods, (2)
an extensive study of the different variants of the problem based on their application contexts, models, fea-
tures and solution approaches, (3) an overview of the latest trends in research on real-time shared mobility
systems, shared autonomous mobility and crowd-based logistics, (4) a tabulated overview for each section
summarizing the reviewed papers and their problem characteristics and solution methods, and (5) a review
of recent shared mobility case studies analyzed and classified according to their scope and the approaches
used.
This paper is organized as follows. In section 2, we present the different variants of the shared mobility
problem, study their different features and modeling approaches, and explain how we build on them. In sec-
tion 3, we focus on mobility systems that allow people to share their rides, including those with real-time
settings and those that consider shared autonomous vehicles. In section 4, we investigate the latest develop-
ments in city logistics and go on to review the integrated passenger and freight transportation problem with
its solution methods and applications. This review of the literature on ridesharing and combined systems is
split into two separate sections to facilitate the organization of the survey and help readers easily identify
the parts of the literature that interest them most. In addition, each section comes with a set of case studies
on the relevant shared mobility topics. Finally, in section 5, we summarize the key findings and suggest
directions for future research.
2. Shared mobility problems
2.1. Background
As introduced earlier, the concept of shared mobility applies not just to people transportation but also
to combined people and freight transportation to make better use of available transportation resources. The
literature has introduced a number of variants of the shared mobility problem for people and freight trans-
portation (Figure 1). Shared mobility systems for people transportation aim to minimize the number of
vacant seats in vehicles in order to reduce the number of used vehicles, and thus traffic congestion and pol-
lution. This can be achieved using a number of concepts, such as; ridesharing,carpooling,vanpooling,
car-sharing,dial-a-ride and others. Ridesharing allows people with similar itineraries and time schedules
to share a vehicle for a trip so that each person’s travel costs (i.e. fuel, toll, parking expenses, etc.) are re-
duced (Furuhata et al. (2013)). Based on this definition, we use the term ”ridesharing” throughout the paper
to represent this category of systems in which people share their rides. The idea of ride-sharing has many
benefits including reducing travel cost and time, decreasing fuel and energy consumption, alleviating traffic
congestion and thus reducing air pollution. There are several variants of the ridesharing problem , most
of which develop efficient mobility systems that allow travelers to share their trips and thus enhance their
travel experience (Agatz et al. (2012)). Planning for rideshared trips can be categorized into ’prearranged’,
or ’static’ ridesharing, and ’dynamic’ ridesharing. In prearranged ridesharing, travelers’ demand (drivers
and riders) is known beforehand (i.e. travelers’ origins, destinations, and departure and arrival times are
given in advance) and can thus be used to plan their shared trips. Such prearranged services are mainly used
for planning regular commuter trips as well as shared long-distance trips (e.g. inter-city trips). However,
long-distance trips generally have more flexible time schedules than commuting trips. Dynamic rideshar-
ing focuses on matching drivers and riders on-the-fly. In other words, new drivers, offering rides, and riders,
requesting rides, can enter and leave the system at any time, and the system then tries to match their trips at
3
Figure 1: Shared mobility - Problem variants
short notice (or even en-route). In their review of dynamic ridesharing systems, Agatz et al. (2012) focused
on the optimization problem of efficiently matching drivers and passengers. This ride-matching problem
has two steps. First, it determines efficient vehicle routes, and then it assigns passengers to those vehicles
taking into consideration the conflicting objectives of maximizing the number of matched travelers and
minimizing the operation cost and passenger inconvenience (these real-time systems are further explored in
section 3.2).
One variant of the ridesharing problem is called the carpooling problem. Carpooling was first intro-
duced by large companies in an effort to encourage their employees to pick up colleagues while driving
to/from work. The idea was to minimize the number of cars traveling to their sites every day (Baldacci
et al. (2004)). Carpooling is generally used for commuting but has become increasingly popular for longer
one-offjourneys. The carpooling problem aims to determine the subsets of travelers that will share the
same trip and the paths these shared trips should follow in order to maximize sharing and minimize travel
costs. In order to increase the flexibility of carpooling services, which are usually prearranged, the concept
of flexible carpooling has been introduced (Shaheen et al. (2016)). Flexible carpooling, also called ca-
sual carpooling or slugging, is a semi-organized service in which destination, meeting point and departure
times are all known in advance among potential participants. The main difference is that rideshares are
formed spontaneously at the meeting point on a first-come first-served basis (Chan and Shaheen (2012)).
This enhanced flexibility opened the door to deploying new carpooling services, not only for daily com-
mutes but for long-distance trips as well (see SlugLines,SmartSlug and KangaRide for example). Along
similar lines, Kaan and Olinick (2013) consider the vanpooling problem with its optimization models and
solution algorithms. In this problem, commuters in the vanpool drive to an intermediate location, called
a park-and-ride location, and then take a van and ride together to the target destination. Car/vanpooling,
4
which can be operated on daily or long-term bases (Wolfler Calvo et al. (2004)), provide regular and cost
efficient means of transportation, they do not accommodate unexpected changes of schedule. By contrast,
the dial-a-ride (DARP) provides shared trips between any origin and destination in response to advanced
passenger requests within a specific area (see Molenbruch et al. (2017); Ho et al. (2018) for recent review).
The DARP models a demand-responsive transportation mode in which the aim is to define a set of routes
in order to satisfy passenger requests at minimized costs (Masson et al. (2014); Ritzinger et al. (2016)).
Each request consists of transporting a passenger from his/her origin location to his/her destination loca-
tion where passengers with similar route and time preferences can share the same vehicle as long as there
is capacity. As such, solving the DARP is about minimizing the total travel distance, and thereby travel
time, while respecting rider-specified time restrictions and any vehicle restriction constraints (more prob-
lem features and objectives are discussed in sections 2.2 and 2.3 respectively). These demand-responsive
systems often focus on providing service to people with reduced mobility (e.g. elderly, handicapped etc.).
The main difference between a DARP and a dynamic ridesharing problem is that a driver in the DARP can
provide service to a wide set of passengers, as the drivers in this case are part of the service, and thus have
less restrictions regarding route and time. In contrast, a driver in a dynamic ridesharing context can only
provide service to passengers with similar route and time schedules to the driver (Gu et al. (2016)). In other
words, in DARP, all drivers are professional and typically operate out of one or more depots, whereas in
dynamic ridesharing each driver is often an individual who has a specific origin and destination and may
have preferences to be considered (like a maximum detour time, maximum number of stops, etc.).
Another variant of the ridesharing problem is the shared-taxi problem introduced by Hosni et al. (2014)
as a multi-vehicle dynamic DARP. In the shared-taxi problem, passengers indicate their desired pickup and
drop-offlocations, their earliest/latest acceptable pickup/drop-offtime, and a maximum trip time. Solving
the shared-taxi problem aims to optimally assign passengers to taxis and determine the optimal route for
each taxi, which means this problem shares the same characteristics, such as demand-responsiveness, as
the DARP. However, the main difference is that most shared-taxi services aim to minimize the response
time to passenger requests whereas dial-a-ride systems aim to minimize vehicle operating cost by reduc-
ing the number of vehicles used to serve given passenger demands (Jung et al. (2016)). When considering
ridesharing variants, it is important to differentiate ridesharing from carsharing, which is a different concept.
Carsharing is a car rental service in which people who are interested in making only occasional use of a
vehicle can rent cars for short periods of time (Agatz et al. (2012)). Although carsharing shares the aim
of reducing car usage and increasing mobility with ridesharing, the optimization challenges that arise in
both systems are different. Those challenges include, determining depot locations and assigning and redis-
tributing vehicles among these depots. Although the carsharing concept allows people to occasionally use
a network of vehicles for short periods of time, they do not necessarily share their trips with other travel-
ers, which rules carsharing systems outside the scope of this review. Here we use the different variants of
ridesharing introduced so far to classify recent studies on shared mobility systems for people transportation
(see section 3).
On the other hand, combining passenger and freight flows has the potential to improve the performance
of existing transportation services as their needs can be satisfied with fewer resources (Trentini et al. (2015)).
In this kind of combined delivery system, spare capacity in public transport systems can be used for retail
store replenishment, or taxis can move or deliver freight when carrying a passenger or during idle time. In
an integrated system, when transporting freight, we need to decide whether to use a pure freight or people
transportation network or a combination of the two (Savelsbergh and Van Woensel (2016)). This choice
depends on the origin location, destination, and due time of freight. In our paper, the focus is on systems
in which people and freight transportation are combined. Li et al. (2014) introduced the share-a-ride
5
problem (SARP) in which people and parcels are handled in an integrated way by the same taxi network.
In this problem, a number of taxis drive around in an urban area to serve passenger requests but can also
deliver some freight (parcels), from their origins to their final destinations, as long as these deliveries do
not add significant extra time to their passengers’ trips. Further, Ghilas et al. (2016a) explored an integrated
solution for simultaneous passenger and freight transportation so that fewer vehicles are required. In their
problem, a set of pickup and delivery vehicles is used to serve a set of parcel delivery requests where a part
of the delivery process is carried out on a scheduled passenger transportation service. Trentini et al. (2015)
introduced another integrated system in which goods are transported in city buses, which have some spare
capacity, from a distribution center to a set of bus stops before they can be delivered to final customers by a
fleet of near-zero emissions city freighters.
Increasing interest in such combined systems has led to the concept of crowd-sourced delivery. Crowd-
sourcing allows activities that were traditionally performed by a certain agent or company to be outsourced
to a large pool of individuals (Goetting and Handover (2016)), which aligns it to the concept of sharing
economy. Crowd-sourced delivery, or crowd-shipping, is based on sharing excess and underused assets,
which here translates as using excess capacity on journeys already taking place in order to make deliv-
eries. As such, crowd-sourced delivery could revolutionize delivery by increasing operational efficiency
and reducing transportation costs. The problem of combining passenger and freight transportation shares
many features with the ridesharing problem where only passengers are considered. However, it has some
complicating features as well, such as transfers, synchronization, capacity constraints, multiple echelons,
etc. (Savelsbergh and Van Woensel (2016)). The key to successfully combining passenger and freight
transportation is to ensure there is no significant negative effect on people when goods are transported or
delivered during their journey. We explore and discuss these combined systems in section 4.
Variant Goods
transport
On-
demand
Daily
commute
Long-
distance
Pre-
arranged
Real-time
Carpooling X X
Flexible Carpooling X X X X
Vanpooling X X
Prearranged Ride-sharing X X X
Long-distance Ride-sharing X X
Dynamic Ride-sharing X X
DARP X X X X
Shared-Taxi X X
Combined Delivery X X X
Share-a-Ride Problem X X X
Crowd-sourcing X X X X
Table 1: Shared mobility variants for people transportation - Different characteristics
Table 1gives a roll-up summary of the different variants of recent shared mobility problems. Below
we give a more detailed analysis of these variants and the modeling approaches and optimization methods
commonly used.
2.2. Modeling and features
In the shared mobility problem, a set of transportation requests, representing passengers or passengers-
plus-goods, need to go from their origins to their destinations while satisfying certain criteria and respecting
6
certain service specifications. The service provider receives these different requests and then arranges with
its available transportation resources (vehicles, car parks, drivers etc.) for the delivery process. This plan-
ning of shared trips is one of the main tasks in shared mobility. In this problem, the service is shared in the
sense that multiple requests might be serviced using the same resource (e.g. vehicle) at the same time. In
order to establish this shared service, a set of features and constraints should be considered. Shared mobil-
ity problems are usually modeled using different vehicle routing problem (VRP) formulations that represent
these features as a set of additional constraints characterizing each variant of the problem. Many of these
features can be found in both ridesharing systems and systems combining passenger and freight transporta-
tion, but other features can relate to either ridesharing systems or combined systems, but not both. In the
following, we summarize the different types of features and constraints reported in the literature for the
shared mobility problem. Furthermore, we identify problem variants that consider each type of constraint
discussed in order to get a clear picture of these variants and their common and different characteristics.
Routing constraints (RC):
In shared mobility systems, every request needs to be transported from its origin to its destination, and
the origin location has to be visited first. This feature applies to both passengers and goods but can have
some variations. For example, in some ridesharing applications, a passenger can be picked up or dropped
offat an intermediate location, usually called meeting point, which can lead to shorter detours (Stiglic et al.
(2016a)). Another example is found in multi-echelon transportation systems where goods are transported
through a scheduled line to a public transport station and then delivered by vehicle to their final destination
(Ghilas et al. (2016b)). While most models insist that each transportation request is served by one vehicle at
most (as in Hosni et al. (2014) for a shared-taxi problem and Li et al. (2014) for a multi-echelon combined
system), some models allow these requests to be transferred using multiple vehicles (as in Herbawi and
Weber (2011) for a dynamic multi-hop ride-sharing problem and Masson et al. (2014) for a DARP with
transfers).
Furthermore, in demand-responsive transportation systems (including DARP and shared-taxi systems)
and many logistics systems in which a fleet of vehicles is located at specific locations (depots) and ready
for service, there is an additional constraint imposed on the route each vehicle will follow: each vehicle
should return to one of the depots when its trip is finished. In some simplified problem settings, a vehicle
might have to return to the same depot from which it started its trip. Moreover, any shared mobility model
must ensure each vehicle reaches and leaves a corresponding location (request origin or destination, depot,
intermediate meeting point or a public transport station). This constraint ensures conservation of flow, and
is very common in shared mobility problems. In some combined systems, passenger requests are given
higher priority when building routes to serve both passengers and goods (Li et al. (2014)). The routes are
first constructed based on passenger requests, then freight requests are only inserted when passenger trips
are not significantly affected. Routing constraints are usually considered hard constraints, because violating
them might lead to detached vehicle routes or a request being picked up but not delivered to destination.
Thus, these constraints need to be strictly respected when modeling and solving a shared mobility problem.
Time constraints (TC):
Besides indicating where a transportation request needs to be picked up and where it should be trans-
ported to, a shared trip must also indicate when this process can take place. This is usually done by asso-
ciating a time window with each transportation request, whether for a passenger or freight. In ridesharing
systems, this time window is usually given by each passenger indicating the earliest departure time from
7
his/her origin and the latest arrival time at his/her destination. Thus, in order for a passenger to participate
in a shared trip, he/she should be picked up at origin and dropped offat destination within the time window
he/she has specified (Stiglic et al. (2016a)). Like passenger requests, freight delivery requests may also be
associated with a time window. In some cases, two time windows are used to represent these time restric-
tions: a pickup time window, indicating when a request should be picked up, and a drop-offtime window,
indicating when it should be delivered (Ghilas et al. (2016a)).
In addition, there could be added restrictions on the duration of the shared trips. In most ridesharing
systems, a set of drivers, offering rides, and riders, looking for rides, are matched to share their trips. In
order to accommodate the riders, the driver might have to make a detour from his original itinerary and
make some extra stops (Furuhata et al. (2013)). The length of this potential detour depends on how far the
driver is willing to extend his/her trip time. Moreover, if drivers have sufficient time flexibility, they might
provide rides to multiple riders simultaneously. Of course, pick-up and drop-offof several riders in a single
trip adds layers of complexity to the planning decisions (Agatz et al. (2011)). Thus, a successful ridesharing
respects the departure and arrival times for all participants, as well as the maximum detour time for the
driver. In DARP-like systems, drivers are employed by the service provider (like in shared-taxi services),
and thus have no preferences in terms of departure, arrival and detour times. In such systems, other time
restrictions might be considered, such as: maximum working hours for drivers, a maximum response time
for processing a passenger request, and the maximum service time for vehicles, which is usually related to
recharging and maintenance operations (Li et al. (2014)). Most of the previous scheduling constraints also
apply to combined systems transporting passengers and goods through the same network. One difference is
that in combined systems, every participant specifies a trip excess time which indicates his/her readiness to
extend the trip in order to pick up and deliver some goods (Li et al. (2016a)). Thus, successful integration of
passengers and goods in a shared trip should respect the maximum extra time that participating passengers
are ready to accept.
Unlike routing constraints, time constraints, also called scheduling constraints, are considered soft con-
straints, because violating them might not detach vehicle routes or intercept the flow, but may result in
passengers or freight arriving late to their destinations, especially in real-world conditions. Violating these
constraints may thus be allowed if it increases the likelihood of finding a solution, but discouraged through
a penalty cost.
Capacity constraints (CC):
A capacity constraint is a factor that prevents a shared transportation resource from being overused. In
ridesharing systems, a capacity constraint limits the number of passengers sharing the same vehicle at the
same time to the number of vacant seats in that vehicle (Santos and Xavier (2015)). Besides limiting the
maximum number of passengers, many vanpooling systems also require a minimum number of passengers
to form a vanpool for a shared trip (Kaan and Olinick (2013)). Number of participants in a shared vanpool-
ing trip must therefore be within these two limits. In logistics systems, a capacity constraint ensures that
the volume of goods to be transported does not exceed the available space provided by the transportation
service (Savelsbergh and Van Woensel (2016)). This constraint holds valid whether goods are transported
using a fleet of vehicles (Li et al. (2014)), public transport (Behiri et al. (2018)) or any other transportation
service. In addition, in integrated models where passengers and goods are transported together, constraints
on both capacities may need to be considered. This is because most of the reviewed literature assumes
that passengers and goods are transported in separate compartments (Ghilas et al. (2016c)). In an uncertain
environment, where passenger or freight demand is stochastic, these capacity constraints might be violated,
and should thus be treated using stochastic approaches.
8
Cost constraints (OC):
In some problem settings, a ridesharing participant may specify a maximum travel cost that he/she is
willing to pay for the shared trip, and should thus be matched to shared trips that stay under the maximum
amount specified. Furthermore, in order to attract more participants, travel costs in ridesharing systems
should be competitive with other modes of transportation. A good example can be found in vanpooling
systems where passengers are only assigned to vanpools that are cheaper than their current commuting
costs (Kaan and Olinick (2013)). However, integrating goods delivery with passenger trips that already take
place could decrease travel costs for participants and transportation costs for goods (Crainic and Montreuil
(2016)). Even if passengers would have longer detours when freight delivery is added to their trip, they
would still get lower travel cost than if no deliveries are added. However, the bulk of research on these
combined systems does not consider travel and transportation cost as a feature or constraint in the system
but more as an objective to be minimized given its importance for service operators (as we will see in
Section 2.3).
Synchronization constraints (SC):
Many recent research papers have focused on studying different synchronization constraints in shared
mobility problems. An extensive review by Drexl (2012) identified five different types of synchronization
constraints for VRPs. The first type of synchronization constraint, called task synchronization, is required
when multiple transportation units are capable of serving a task (i.e. a transportation request). In other
words, a task synchronization constraint ensures that each request (passenger or freight) is served exactly
once by one or more vehicles (Fink et al. (2018)). Furthermore, when the operations performed by differ-
ent transportation units need to be coordinated in terms of space and time, operation synchronization is
required. In other words, a schedule for a vehicle in a shared mobility system should be built to take into
account the schedules of other vehicles, so their schedules need to be synchronized. Logistics systems offer
good examples of when operation synchronization is needed: for example, a system where two different
vehicles arrive at different customer locations, one delivering the product and the other carrying the crew
to install it (Hojabri et al. (2018)), or a system in which multiple vehicles cooperate in order to transport
one big-size cargo (Hu and Wei (2018)). Another type of synchronization constraints is called movement
synchronization. In some ridesharing systems where passengers are allowed to transfer from one vehi-
cle to another on the way to their destination, the arrival and departure of vehicles to and from transfer
points need to be synchronized (see Masson et al. (2014) for an example). Another example is found in
multi-echelon systems where goods are transported with passengers through a scheduled transport line after
being collected by a fleet of vehicles. Such systems also need to ensure synchronization between requests
and the collecting vehicles, and between requests and the scheduled line departures (Ghilas et al. (2016b)).
Load synchronization ensures that the right amount of load is collected and delivered to a customer, or in
other words, no load is lost when transferred between different transportation units. This is the case when
deliveries are done using two distinct fleets of vehicles where a request is transferred from one vehicle to
another at satellite locations before it can be delivered to a customer (Grangier et al. (2016)). The same load
synchronization constraint is needed when deliveries are transferred between pickup and delivery vehicles
and a public transport line in a multi-echelon transportation system. Finally, resource synchronization
ensures that the use of resources common to different transportation units is limited to availability (Drexl
(2012)). Number of drivers, vehicle fleet size, available parking slots, vacant seats for riders to share, and
the available space and capacity in transportation units in both ridesharing and combined systems are all
examples of limited resources whose use needs to be synchronized. Xiang et al. (2006) considered a DARP
9
with passenger-driver and passenger-vehicle compatibility constraints. They classified passengers into dif-
ferent levels, and ruled that vehicles could only accommodate passengers of corresponding levels, i.e. a
passenger can only use a vehicle of the same or higher level. As a rule, modeling synchronization con-
straints yields more complex and non-linear mathematical formulations (e.g. implications) which need to
be handled using linearization techniques. However, these constraints are important for modeling realistic
settings and, from an algorithmic perspective, can be used to decompose hard problems (e.g. they can be
used as coupling constraints in a column generation based approach; see Fink et al. (2018)).
2.3. Objective functions
Most objectives that shared mobility problems aim to optimize can be classified into two main cat-
egories; operational objectives and quality-related objectives. Operational objectives are usually about
optimizing system-wide operating costs, such as minimizing vehicle miles and transportation time, maxi-
mizing the number of serviced requests, minimizing the number of required vehicles, and others. Quality-
related objectives are about enhancing the quality of service provided. For example, minimizing total pas-
senger ride or waiting time might yield a better performance from the passenger perspective but not from
a system-wide perspective. Furthermore, minimizing system-wide travel time does not necessarily mean
shorter travel times for every passenger. This difference between collective and individual perspectives in
shared mobility systems justifies the need for methods that consider both operational and quality-related
objectives. A good example can be found in Kalczynski and Miklas-Kalczynska (2018) where a decentral-
ized approach takes carpool participant preferences into account while maintaining the same system-wide
savings that can be obtained in centralized approaches.
Much of the research on shared mobility is focused on optimizing a single operational objective, but
there are papers that consider multiple-objective systems combining operational with quality-related objec-
tives. In single objective systems, service quality considerations are represented as constraints in the model
to ensure a minimum service level (Molenbruch et al. (2017)). In other words, a set of constraints limiting
passenger extra ride times, caused by deviations from their original routes, are added when optimizing the
system. Likewise, most of ridesharing research has focused minimizing the system-wide travel distance
(vehicle miles) or total travel time. For example, in Wolfler Calvo et al. (2004), the system-wide travel
time in a carpooling system is minimized with an added penalty cost for unserved requests. In a dynamic
environment, where transportation demand is revealed in real-time, satisfying full demand may not be at-
tainable, in which case it becomes pertinent to maximize the number of served requests as it extends the
reach of the transportation service (Berbeglia et al. (2012)). Some studies have considered maximizing the
total profit obtained from the ridesharing system (see Hosni et al. (2014) for a shared-taxi problem and Par-
ragh et al. (2015) for a DARP) or minimizing the total cost of operating it (see Kaan and Olinick (2013) for a
vanpooling problem and Braekers et al. (2016) for a DARP). Moreover, some more problem-specific objec-
tives have been considered in the literature, such as; minimizing the number of required vehicles (Guerriero
et al. (2013)), minimizing vehicle emissions (Atahran et al. (2014)), maximizing passenger occupancy rate
(Garaix et al. (2011)), minimizing staffworkload (Lim et al. (2017)), and maximizing system reliability
(Pimenta et al. (2017)). Most studies on combined crowd-sourced systems have focused on either maxi-
mizing the profit obtained by integrating passenger and freight flows (Li et al. (2014)) or minimizing the
operational cost of these systems (Ghilas et al. (2016a)), but there have been efforts to consider additional
objectives, such as minimizing the number of vehicles required to operate the system (Trentini et al. (2015))
and minimizing the total wait time of demands before being serviced (Behiri et al. (2018)). The recent
shared mobility studies listed in Table 2and 5have been classified using these different objectives.
As mentioned above, multi-objective systems consider a combination of two or more of the above-
listed single objectives. Solving multi-objective problems requires different methods to those employed for
10
solving single-objective problems. The literature identifies three main techniques for dealing with multi-
objective problems. The first, and most popular approach is to aggregate the different objectives into a
weighted-sum objective with different measures. In this approach, a weight has to be defined for each of
the combined objectives. As such, the relative importance of each objective needs to be quantifiable and
well-defined. A good example can be found in Kirchler and Wolfler Calvo (2013) who used an aggregated
objective function combining six different objectives: minimizing routing cost, excess ride time, passenger
waiting time, route durations, early arrival times at pickup and delivery nodes, and number of unserved re-
quests. Another example of a weighted-sum objective can be found in Lehu´
ed´
e et al. (2014). One drawback
of the weighted-sum approach is that it might not be able to find the full set of non-dominated solutions for
optimization problems in which some variables are constrained to be integers (i.e. non-convex optimiza-
tion problems). In addition to weighted-sum approach, some papers consider a hierarchical, also called
lexicographical, objective function. In this approach, the different objectives are ordered according to their
importance, and first the main objective is optimized to generate a set of solutions, then a secondary objec-
tive is optimized whenever two solutions with the same quality, in terms of the main objective, are obtained.
Stiglic et al. (2016a) considered a ridesharing system with a lexicographic objective function. First, the sys-
tem generates solutions that maximize the number of matched participants and then the secondary objective
is used to select solutions that maximize the distance savings (see also Schilde et al. (2014)). This approach
is therefore efficient in problems where the different objectives can be classified into main and secondary
objectives. Finally, the third approach for dealing with multi-objective problems is to obtain the set of non-
dominated solutions in terms of the different criteria, called Pareto frontier (Paquette et al. (2013)). The
main advantage of this approach is that it helps decision makers analyze the relations between the different
objectives, as it provides all the possible optimal solutions. However, this approach might not be applicable
for dynamic shared mobility systems where decisions need to be taken in relatively short time frames, as
it requires obtaining the full set of optimal solutions and a human being to select the best solution among
them (Molenbruch et al. (2017)).
2.4. Computational complexity and solution approaches
As mentioned above, the shared mobility problem is a generalization of the vehicle routing problem
(VRP) and is NP-hard in general. In addition, simplified variants of the problem (e.g. with a single-driver
single-rider setting, single pickup and dropoffor a single-objective function) are still NP-hard (Gu et al.
(2016)). Furthermore, solving these problems becomes more complex when they have dynamic settings
and stochastic input data. Thus, both exact and heuristic solution approaches have been introduced in the
literature. Due to the complexity of shared mobility problems, most studies have focused on developing
approximation and heuristic approaches for solving them. That said, a number of studies have developed
exact methods for solving simplified variants of the problem. These exact methods are usually used to solve
static problem variants with deterministic data, e.g. a column generation-based method for the carpooling
problem (Baldacci et al. (2004)), a branch-and-cut algorithm for a multi-vehicle static DARP (Cordeau and
Laporte (2007)), a two phase method for generating optimal matches in a static ride-sharing problem (Stiglic
et al. (2016a)), and a branch-and-price algorithm for a crowd-sourced system with a scheduled line for trans-
porting passengers and goods (Ghilas et al. (2016a)). However, solving these static variants becomes more
complex when complicating features are added to the system, such as allowing passenger transfers, inte-
grating public transport, and considering vehicle/driver compatibility. To deal with these complex features,
a number of heuristic approaches have been introduced, such as a local search strategy for a static DARP
with complex constraints (Xiang et al. (2006)), an Adaptive Large Neighborhood Search (ALNS) heuristic
for the DARP with transfers (Masson et al. (2014)), a constraint-based Large Neighborhood Search (Jain
and V. Hentenryck (2011)), an integrated column generation in a Large Neighborhood Search (Parragh and
11
Schmid (2013)) for a static DARP, a Lagrangian decomposition heuristic for the static shared-taxi problem
(Hosni et al. (2014)), and another ALNS approach for the crowd-sourced delivery system with scheduled
line (Ghilas et al. (2016b)).
Nevertheless, even these heuristic algorithms often have large computation times limiting the size of
instances on which they can be tested, which consequently also limits their usability for large-scale and
dynamic systems which need to be re-optimized at regular intervals as new transportation requests enter
the system. As a result, heuristic approaches need to be improved so that good-quality solutions can be
obtained in short computational times. In order to clarify how a heuristic approach can be improved to
handle dynamic problem settings, we take the ALNS heuristic as an example. In a classical ALNS-based
method, a set of insertion and removal operators are used to enhance a current solution. Thus, at each
iteration, one insertion operator and one removal operator are selected and applied to the current solution
seeking an improvement. This process continues until an acceptable solution is found or a maximum number
of iterations is reached. In order to minimize the number of required iterations, and thus the computation
time, the classical ALNS can be improved by adding a score to each operator (Masmoudi et al. (2016)).
If using one operator, whether it is an insertion or removal operator, brings an improvement to the current
solution, then the score of the operator used will be increased. The probability of using this operator in the
next iterations will thus be higher, and so an acceptable solution would be reached in a shorter time.
For the uncertainty factor, more advanced techniques are needed for solving shared mobility problems
with one or more source of uncertainty. This is because a solution obtained by solving the deterministic
version of the problem might not be valid when uncertainty is revealed. The most common source of
uncertainty lies in transportation demands, where some of the data on transportation requests is unknown
at the moment the shared trips are planned. This uncertainty might be in request occurrence times or
locations (Ghilas et al. (2016c)). AAnother important aspect is the stochasticity of travel times, as traffic,
accidents, and other factors make it impossible to know travel times between different locations in advance
(Heilporn et al. (2011)). Due to the complexity of this uncertainty, most studies have not considered any
more than one source. However, there has been some research on integrating multiple sources of uncertainty
(e.g. considering stochastic travel times and delivery locations; Li et al. (2016b)). For solving shared
mobility problems that involve uncertainty, the literature has identified two categories of methods. The
most common approach is to make a decision and then minimize the expected (recourse) costs induced
by the consequences of this decision. This approach is called stochastic programming with recourse.
In the second approach, called multi-scenario approach, the expected costs are estimated by evaluating
a solution on a set of different scenarios. In this approach, heuristic algorithms can be efficiently used to
obtain a solution each time a new scenario is tested (for more details on stochastic models and their solution
approaches, interested readers are referred to Ritzinger et al. (2016)).
3. People sharing rides
This section focuses on introducing shared mobility problems for people transportation. The idea is to,
(i) investigate the potential benefits and planning aspects (Section 3.1), (ii) review the modeling choices and
optimization approaches in real-time settings (Section 3.2), and (iii) discuss the potential integration of new
automated services in such shared systems (Section 3.3). We also provide an overview table summarizing
the papers reviewed and their problem characteristics and solution approaches (Table 2).
3.1. Planning and potential benefits
As mentioned above, the increasing demand for passenger transportation has attracted more research
into enhancing the efficiency and quality of existing public transport systems and developing new systems
12
that can provide more sustainable solutions (Wolfler Calvo et al. (2004)). Ridesharing is one opportunity
to provide a reduced-cost mobility service that is as flexible as private cars but can also increase occupancy
rates and decrease traffic and pollution levels (Furuhata et al. (2013)).
In a ridesharing system, drivers and riders share the travel costs so that each benefits from the shared ride.
Benefit can be obtained when the travel cost of a shared ride is lower than the cost of the alternative means of
transport (individual car trips, taxis or public transport). While some users choose to participate in a shared
ride to reduce their travel expenses, others may be motivated by the potential social and environmental
benefits (Furuhata et al. (2013)). Besides the potential cost savings, ridesharing can also allow drivers to
reduce their travel time because they will be able to take high-occupancy lanes reserved for vehicles with
two or more occupants (Stiglic et al. (2015)). Riders, on the other hand, may appreciate dispensing with
the need to drive or own a vehicle. Despite its potential advantages, there are also major obstacles that
prevent wider uptake of ridesharing. According to Furuhata et al. (2013), the two main barriers to wider
adoption of ridesharing are coordinating passenger trips that have similar itineraries and time schedules,
and developing effective methods to encourage participation. Limited flexibility in participants’ itineraries
and time schedules may result in many of them not finding a match. Other issues like privacy, safety, social
discomfort and pricing are also challenges for ridesharing systems. For example, a potential participant may
be willing to share rides with colleagues and friends, but not with complete strangers (Agatz et al. (2012)).
As such, new methods for arranging the shared rides need to be developed, and reputation and profiling
systems for addressing these social and privacy concerns need to be built.
In order to attract more riders and facilitate matching them in shared rides, we identify some the concepts
in the literature that can help maximize the potential benefits of a ridesharing system. One of those concepts
is to consider a set of meeting points where a shared ride can take place. Thus, a rider might be picked up
at his/her origin location or at a pickup meeting point and dropped offat his/her destination location or at a
drop-offmeeting point. Meeting points would thus allow drivers to have smaller detours while maintaining
a good-enough quality of service for the riders. Stiglic et al. (2015) investigated benefits of using meeting
points in a ridesharing system and found that as they can lead to shorter detours, meeting points have the
potential to increase the system-wide distance savings as well as the number of participants that can be
matched in the system. With the aim of attracting more riders, especially from suburban areas, Stiglic et al.
(2018) examined the potential benefits of integrating ridesharing and public transport, and found that the
two can prove complementary. While ridesharing can bring passengers from less-densely-populated areas
to public transport, the public transport system allows ridesharing to provide service to more passengers
and reduce drivers’ detours. Another concept is to allow riders requesting a shared ride to transfer between
drivers, and thus use more than one driver to reach his/her final destination. Herbawi and Weber (2011)
considered a version of this multi-hop ridesharing problem in which the transportation network is formed
by driver ridesharing offers. Thus, drivers do not deviate from their original itineraries while riders have to
find routes that minimize their travel time, costs, and the number of transfers required to reach their final
destination. Masson et al. (2014) considered ridesharing settings in which riders are allowed to transfer
between vehicles at intermediate transfer points, and suggested that these transfers can lead to considerable
savings, especially when multiple transfers are allowed. To guarantee a certain level of service, Lee and
Savelsbergh (2015) investigated the benefits of deploying a number of dedicated drivers to provide service
to unmatched riders, and identified the environments in which dedicated drivers are most beneficial. When
the number of riders increases to a certain point, the need to deploy a set of dedicated drivers became
essential to maintain an acceptable service level.
13
3.2. Real-time ridesharing
As introduced earlier in section 2, a real-time ridesharing system aims to bring travelers together at
short notice. Furthermore, a real-time ridesharing system might need to be re-optimized at regular intervals
as more travelers enter or leave the system. In addition, travelers already en route need to be notified of
any change of plan at each time the system is re-optimized, as their original routes might be updated. This
automated process requires efficient models and algorithms for matching drivers and riders in very short
computation times. As a result, many recent studies on real-time ridesharing systems have focused on de-
veloping heuristic approaches, as they can provide good-quality solutions in relatively short computation
times. Nevertheless, such systems can also be addressed by enumeration (exact) algorithms (like branch-
and-bound). Due to their brute-force nature, using enumeration algorithms for such real-time systems may
require an additional preprocessing effort in order to fit the short computation times needed. Some prepro-
cessing techniques can tighten travelers’ time windows, eliminating unnecessary variables and constraints
and identifying inequalities for narrowing the solution space (Liu et al. (2014)). For example, Agatz et al.
(2011) introduced an efficient rolling horizon approach that can provide high-quality solutions for dynamic
ridesharing systems where drivers and riders continuously enter and leave the system. In a later survey,
Agatz et al. (2012) provided a review of the related operations research-based models in the academic
literature. Here we review the more recent studies and their solution approaches.
Huang et al. (2013) proposed a branch-and-bound algorithm and an integer programming algorithm for
solving the problem of large-scale real-time ridesharing, and introduced a kinetic tree algorithm capable of
better scheduling dynamic requests and adjusting routes on-the-fly. Liu et al. (2014) proposed a branch-
and-cut algorithm to solve a realistic DARP with multiple trips and request types and a heterogeneous fleet
of vehicles with configurable capacity and manpower planning. To solve the dynamic ridesharing prob-
lem over a full-day time horizon, Santos and Xavier (2015) suggested dividing the day into time periods,
after which a deterministic instance of the problem can be generated and solved by a greedy randomized
adaptive search procedure (GRASP). Ma et al. (2015) introduced a dynamic taxi-sharing system based on
a mobile-cloud architecture. In their proposition, the system first uses a search method, based on a spatio-
temporal index, to find candidate taxis for every ride request, and then a taxi that satisfies the request with
the shortest detour is selected through a scheduling process. Jung et al. (2016) later suggested applying
hybrid-simulated annealing (HSA) to dynamically assign passenger requests to shared taxis. In addition,
it investigates what type of objective functions and constraints could be employed to improve the system
and prevent excessive passenger detours. Braekers and Kovacs (2016) proposed different formulations for
the DARP with driver consistency (DC-DARP). For solving this problem, the authors developed a large
neighborhood search algorithm that finds near-optimal solutions in short computation times. Masmoudi
et al. (2017) propose three metaheuristics for solving the Heterogeneous Dial-a-Ride problem (HDARP).
These are: an improved ALNS-based method, Hybrid Bees Algorithm with Simulated Annealing (BA-SA),
and with Deterministic Annealing (BA-DA). More recently, Masoud et al. (2017) proposed an exact and
real-time ride-matching algorithm, and the approach maximizes the number of served riders while account-
ing for their travel preferences. The system also aims to minimize the number of transfers and waiting
times for riders, and make their shared trips more comfortable. As ridesharing participants might not accept
the matches proposed by the service provider on-the-fly, it becomes important to analyze how stable the
generated matches are. For this purpose, Wang et al. (2018) studies the stability of rideshare matches by
providing several mathematical programming methods to generate near-stable matches in real-time. Their
results suggested that taking stability considerations into account comes with only a small additional cost
to the system-wide performance in terms of traveled-distance savings.
To conclude, the development of new methods and algorithms for providing good-quality solutions
14
in short times is at the heart of the real-time ridesharing concept, which is why we found rising interest
from the OR research community to address the optimization issues in real-time ridesharing systems. In
many ridesharing systems, like in major metropolitan areas, thousands of drivers and riders might be trav-
eling between thousands of different locations at the same time, thus creating a need for fast optimization
approaches that can match their different trips quickly. Most recent studies have focused on developing
heuristic approaches that can solve large-scale ridesharing problems in real-time (see Table 2) and the door
is open for introducing new heuristic techniques. In what follows, we identify possible directions for future
research. First, (i), as few papers have considered synchronization aspects in such real-time systems (see
Table 2), more research should study these aspects and introduce them in future ridesharing systems. An
example would be to allow flexible driver-to-vehicle assignments and multi-depot settings which require
drivers and vehicles to be synchronized. Second, (ii), very few papers have considered cost restrictions
when matching travelers in share rides (cost constraints). An interesting avenue for research would be to
focus more on individual traveler benefits from ridesharing beside the system-wide cost considerations.
Third, (iii), decomposition techniques could be integrated into algorithms for solving multi-objective prob-
lems to consider more quality-related objective functions. This is because most of the reviewed papers have
considered a single operational objective with a minimum service quality level due to complexity aspects
(see Table 2). Finally, (iv) for exact approaches, we see three possible techniques to enhance their per-
formance on responding to real-time system needs. These are: developing preprocessing techniques that
can decrease the enumeration effort, decomposing the problem based on geographic partitioning or time
intervals to make the size of the problem to be solved at each time smaller, and developing faster algorithms
for solving the subproblem in a decomposition-based approach (e.g. branch-and-price, branch-and-cut, and
so on) where most of the computational effort is spent on solving the subproblems. That said, ridesharing
systems that can handle requests dynamically are clearly gaining the upper hand. As new innovations and
transport technologies are introduced, we need more research into responding to traveler needs in future
real-time ridesharing systems.
3.3. Ridesharing with autonomous vehicles
Autonomous vehicles (AVs), also dubbed driverless, automated or self-driving, are an emerging technol-
ogy expected to bring fundamental shifts in people transportation. AVs are expected to provide a sustainable
solution that can enhance road safety levels and traffic flows, reduce fuel consumption, and thus improve
passenger mobility in general (Katrakazas et al. (2015)). The potential deployment of autonomous vehicles
in tandem with the increasing need for shared mobility services has attracted the attention of the operations
research community, especially now that many large mobility providers (Tesla, Ford, Lyft and others) have
announced plans to deploy new autonomous mobility services (Sparrow and Howard (2017)). Furthermore,
recent studies on different cities have concluded that if AVs are shared, then the number of vehicles needed
to provide service to all travelers will significantly decrease (Levin et al. (2016)). Despite their potential
benefits, shared autonomous vehicles also come with security concerns. In other words, if autonomous
vehicles do not prove safer than human-driven vehicles, they might not be legally viable for widespread use
(Hevelke and Nida-R¨
umelin (2015)). In a study assessing public interest in such new technology, Daziano
et al. (2016) derived estimates of how much consumers are willing to pay to let vehicles drive for them.
Their results show that modeling flexible user preferences is an important determinant of the amount they
are willing to pay for automation. Krueger et al. (2016) showed that other service attributes, such as travel
cost, travel time and rider waiting time, might be critical factors for uptake of shared autonomous vehicles
(for related studies, see Bansal and Kockelman (2016),Yap et al. (2016), Bansal et al. (2016), Zmud and
Sener (2017)). Another concern is the potential increase in vehicle miles traveled due to repositioning trips
performed by shared autonomous vehicles in order to reach new travelers.
15
Reference Problem variant Method Objective Constraints
RC TC CC OC SC
Characteristic
Baldacci et al. (2004) Car-pooling E D, P X X X X M
Wolfler Calvo et al. (2004) Car-pooling H TXXX M
Xiang et al. (2006) DARP H CXXX M
Cordeau and Laporte (2007) DARP E, H DXXX M
Jain and V. Hentenryck (2011) DARP H DXXX M
Heilporn et al. (2011) DARP E CXXX M
Garaix et al. (2011) DARP E OX X X X M
Herbawi and Weber (2012) D. Ride-sharing H D, T, P X X X X M
Berbeglia et al. (2012) DARP H PXXX M
Kaan and Olinick (2013) Van-pooling H CXXXX M
Parragh and Schmid (2013) DARP E, H CXXX M
Huang et al. (2013) D. Ride-sharing E, H CX X S
Kirchler and Wolfler Calvo (2013) DARP H C, T, N X X X X M
Masson et al. (2014) DARP H DX X X X M
Lehu´
ed´
e et al. (2014) DARP H T, N X X X X M
Hosni et al. (2014) Shared-taxi H CX X X X M
Atahran et al. (2014) DARP H VXXX M
Liu et al. (2014) DARP E TXXX M
Stiglic et al. (2015) P. Ride-sharing E P, D XXX M
Lee and Savelsbergh (2015) D. Ride-sharing H CX X X S
Santos and Xavier (2015) D. Ride-sharing H PXXX M
Parragh et al. (2015) DARP E, H CXXX M
Ma et al. (2015) Shared-taxi H DXXXX M
Ritzinger et al. (2016) DARP H TXXX M
Jung et al. (2016) Shared-taxi H T, C XXX M
Masmoudi et al. (2016) DARP H CXXX M
Braekers and Kovacs (2016) DARP E, H CXXX M
Masmoudi et al. (2017) DARP H CXXX M
Masoud et al. (2017) D. Ride-sharing E PXXXX M
Pimenta et al. (2017) DARP H RXXX M
Alonso-Mora et al. (2017) D. Ride-sharing E CXXX M
Stiglic et al. (2018) P. Ride-sharing E P, D X X X X M
Kalczynski and Miklas-
Kalczynska (2018)
Car-pooling H DXXX M
Wang et al. (2018) D. Ride-sharing H DXXX S
Method: E: Exact approach, H: Heuristic approach.
Objectives: D: Min. Travel Distance, T: Min. Travel Time, P: Max. Number of Participants, C: Min.
Operational Cost, V: Min Vehicle Emissions, R: Max. System Reliability, O: Max. Occupancy Rate,
N: Min. Number of Used Vehicles.
Characteristic: S: Single rider per trip, M: Multiple rider per trip.
Table 2: Shared mobility - Ride-sharing systems
16
Two main AV ownership models are being considered for future transportation systems: AVs as a public
service, or privately-owned AVs. In the case of AVs as a public service, we consider a fleet of such vehicles
at specific locations (depots). AVs are invoked from their stations to satisfy mobility demands in an urban
area such that a single AV can serve multiple demands before going back to a depot. Privately owned AVs
cannot just bring their owners from their homes to their work locations in the morning and bring them back
in the evening while providing ridesharing opportunities to other users, but they can also serve other users
when their owners do not need them (e.g. while they are at work). Although some companies (Tesla and
Ford) have stated plans to sell AVs to consumers, many transportation companies have either explicitly
stated or implicitly implies that they initially plan to use AVs to provide public transportation services
rather than selling individual AVs to private consumers for personal use (Hyland and Mahmassani (2017)).
Given this potential shift from a society that is heavily reliant on privately-owned vehicles to one in which
transportation services are provided through fleets of vehicles operated by private companies, significant
research is needed to plan such new systems and maximize their efficiency.
That said, there is a surge of interest in developing new methods for operating autonomous vehicles.
Such methods consist of finding a path between different locations and determining the safest and most
feasible itinerary. Hyland and Mahmassani (2017) introduced a taxonomy for classifying vehicle fleet
management problems to inform future research on autonomous vehicle fleets. Their paper reviewed the
existing categories to classify scheduling and routing problems, then refine some of them as they relate to the
AV fleet problem, and proposed novel taxonomic categories for classifying AV fleet management problems.
K¨
ummel et al. (2017) proposed a framework for AVs based on the model of a family (where the father is
provider of physical services, the mother is strategic manager, and the children are individual AVs). In this
decentralized model, vehicles are allowed to inter-negotiate while the fleet manager can set fleet strategies
and pre-allocate vehicles to locations where increasing demand is expected. In another framework for
modeling shared AVs, Levin et al. (2016) proposed a heuristic for dynamically constructing shared rides
using AVs. The proposed approach consists of a dispatcher that checks whether there is any AV that is
already located or en route to where a travel demand has appeared and then assigns the AV to carry the
longest-waiting traveler. Furthermore, other travelers are allowed to join the shared trip if they are traveling
to the same or close-enough destination, although priority remains with the travelers already in the vehicle.
Alonso-Mora et al. (2017) proposed a mathematical model for a large-scale real-time ridesharing system
that dynamically finds optimal routes for vehicles serving online requests while taking into account their
actual locations. Their algorithm, which applies to fleets of AVs, uses constrained optimization to improve
an initial greedy assignment and return good quality-solutions that converge to the optimal assignment over
time. In addition, Pimenta et al. (2017) considered a dial-a-ride system in which a set of small AVs operates
between different sections in a closed industrial site. For routing decisions, the paper proposes a heuristic
approach based on GRASP and an insertion mechanism. Another study, by Ma et al. (2017), introduced a
linear programming model for an AV sharing and reservation (AVSR) system in which travelers book AVs
in advance and the system arranges their routes and schedules. Chen et al. (2017) studied potential use of
AVs and presented a mathematical framework for designing AV zones in a general network. The paper also
provides a numerical study to demonstrate the performance of the proposed model.
To conclude, there has been a surge in research on AVs in domains from computer science to robotics
and engineering, but far less research into how to plan and operate AV services. We believe there are two
main reasons for this gap. First, most of scientific and technological advances have been made by AV manu-
facturers and service providers who tend to keep their methods and techniques a commercial secret. Second,
many studies have suggested that the same methods and algorithms that operate conventional vehicles will
continue to apply to AVs, and so a switch from conventional vehicles to AVs does not necessarily entail any
17
real change in the way they operate in a transportation system. From a modeling perspective, this statement
holds for many cases. However, there are some variations in which AV-based systems need to be considered
differently. For example, privately-owned AVs might be allowed to operate while their owners do not need
them, and they might be able to use dedicated roads which could reduce their traffic-related issues compared
to conventional vehicles. In addition, AVs are expected to be electric, and so planning their charging and
maintenance operations might require different approaches, especially as they have a different service range
and they need time to recharge, which could be time-consuming at some intermediate locations (Hiermann
et al. (2019)). Further research should target (i) better understanding how AVs can be operated, owned and
shared in future transportation systems, (ii) identifying their impacts on people transportation and how AVs
respond to passenger mobility needs, (iii) analyzing how shared AVs perform in different scenarios and
real-life situations, including varying transportation demand and network topologies, (iv) identifying the
new features introduced by AVs and studying how these features could affect existing ridesharing models,
and (v) introducing efficient solution approaches that can operate large-scale AV systems and factor in the
critical issue of their recharging and maintenance operations.
3.4. Case studies
This section presents a set of case studies focused on analyzing different ridesharing systems and their
performance and potential impacts. We consider case studies on systems that operate conventional vehicles
or AVs, and classify them according to their research objectives and the approaches used. Research ob-
jectives have focused on studying either the performances, efficiencies and deficiencies of the ridesharing
systems, or their impact on peoples’ lives and future transport infrastructure. On the other hand, we also
observe that the studies considered have used either optimization-based, simulation-based or data-analysis-
based approaches to achieve the intended research objectives. We discuss the different studies and their
outcomes, and provide a table classifying them into different categories.
There have been a number of recent case studies conducted to test the viability of ridesharing systems
and evaluate their proposed solution approaches. Agatz et al. (2011) led a study based on 2008 travel
demand data from metropolitan Atlanta, and the results suggested that advanced optimization approaches
have the potential to increase the participant matching rates and system-wide travel cost savings obtained
in dynamic ridesharing systems. Ma and Zhang (2017) studied traffic flow patterns in a single bottleneck
corridor using a dynamic ridesharing mode and dynamic parking charges, and the results showed that system
performance over the traditional morning commute may not be significantly improved when ridesharing fees
and parking charges are fixed. Nonetheless, dynamic parking charges with appropriately set ridesharing fees
can improve system performance in terms of vehicle miles and hours traveled and in terms of allied travel
costs. Jiang et al. (2017) proposed a large-scale nationwide ridesharing system called CountryRoads which
was deployed in three different years to assess system performance improvement through a case study
of the ‘Chunyun’ spring festival travel season in China. Results indicate that the proposed system was
able to attract more users, achieve a higher success matching rate, and thus contribute to an increasingly
successful ridesharing experience. Ferreira and D’Orey (2015) proposed a dynamic and distributed taxi-
sharing system that was evaluated using a simulation modeling approach based on realistic taxi trips in
Porto (Portugal). Simulation results showed that the system has the potential to reduce taxi fares, operation
costs and total travel distance (up to 9%). Furthermore, Maciejewski et al. (2016) conducted a study to
evaluate a rule-based dispatching algorithm that manages a fleet of shared taxis based on data collected
by local taxi services in Berlin and Barcelona. Results indicated that despite its simplicity and efficiency,
rule-based dispatching suffers from a limited planning horizon. Linares et al. (2017) studied a dynamic
ridesharing system architecture that considers the Metropolitan area of Barcelona as a case study, and.
18
results showed that this transportation mode has the potential to reduce traffic flow and pollution levels in
big cities while offering travelers shorter travel times.
Using data collected by surveying more than 500 respondents in Turin and Rome, Gargiulo et al. (2015)
tested and evaluate a dynamic ridesharing service called VirtualBus. They found that users’ main concerns
were privacy, trust, and reliability of planning. More recently, Wang et al. (2017) investigated the cost
and benefits of ridesharing with friends through a study on travel demands in the Yarra Ranges (Australia).
Their study revealed that limiting ridesharing to friends while rejecting strangers might reduce ride choices
and increase detour distances but it does not generate significantly higher costs. Furthermore, prioritizing
friends can substantially increase matching rate. In an effort to understand how urban parameters affect
the fraction of individual trips that can be shared (or ’shareability’), Tachet et al. (2017) conducted a study
based on millions of taxi trip records in New York City, San Francisco, Singapore and Vienna with the aim
of computing the shareability curves for each city.
Other case studies have focused on studying ridesharing system impacts on existing transportation sys-
tems. Martinez (2015) used a simulation-based procedure to evaluate the impacts of introducing a shared-
taxi system in Lisbon. Barann et al. (2017) conducted another study using more than 5 million taxi trips
in New York City and found that ridesharing could potentially save over 2 million kilometers of travel
distance per week, which would significantly decrease CO2emissions. Similarly, Yu et al. (2017) evalu-
ated the direct environmental benefits of ridesharing in Beijing, and found that it enabled energy savings,
distance savings, and lower CO2emissions. Stiglic et al. (2016b) studied the impact of driver and rider flex-
ibility in an enhanced dynamic ridesharing experience and found that suggested driver and rider flexibility
on departure/arrival times was important to ridesharing system success, but that driver flexibility in terms
of accepting detours was even more important. Thus, the benefits and positive impacts of ridesharing are
linearly correlated to the flexibility of ridesharing participants. Table 3gives a roll-up summary of the case
studies presented.
Method
Scope
Assessing system performance Studying impacts
Optimization-based Agatz et al. (2011)
Jiang et al. (2017)
Stiglic et al. (2016b)
Lee and Savelsbergh (2015)
Simulation-based
Agatz et al. (2011)
Maciejewski et al. (2016)
Ferreira and D’Orey (2015)
Linares et al. (2017)
Ma and Zhang (2017)
Martinez (2015)
Data-analysis-based
Tachet et al. (2017)
Liu and Li (2017)
Sanchez et al. (2016)
Gargiulo et al. (2015)
Wang et al. (2017)
Barann et al. (2017)
Yu et al. (2017)
Table 3: Case studies - Ridesharing systems
Case studies on ridesharing systems (see Table 3) have mainly focused on assessing their performance
and giving cues and clues for further research to increase their efficiency and maximize their benefits.
Studies have since been conducted using optimization, simulation or data-analysis approaches, but there
have been fewer recent case studies analyzing the impacts of ridesharing on transportation systems, possibly
19
because ridesharing is not a new concept, and so the bulk of research is focused either on improving existing
ridesharing systems or operating new ones rather than studying their potential impacts, which are assumed
to be net-positive.
We now have two decades of extensive research into ridesharing systems, but little research on au-
tonomous mobility services for future transportation systems. To fill this gap, a number of recent studies
have focused on new driverless services and their potential impacts on urban mobility. Gruel and Stanford
(2016) identified the long-term effects of introducing driverless cars and explored the conditions that would
make them beneficial or damaging in transportation systems. The study also investigated how automation
could increase the attractiveness of traveling by car. Smolnicki and Sołtys (2016) studied different au-
tonomous mobility solutions and discussed their impacts on metropolitan spatial structures. Talebpour and
Mahmassani (2016) studied the influence of AVs on traffic flow stability and throughput and found that AVs
can improve stability and be more effective in preventing shockwave formation and propagation. Meyer
et al. (2017) simulated the influence of AVs on the accessibility of Swiss municipalities, and concluded that
AVs could dramatically increase accessibility rates and even replace public transport outside dense urban
areas. Correia and van Arem (2016) explored the possibility of replacing individually-owned conventional
vehicles with autonomous ones and what it would mean for traffic flow and parking demand in a city.
Considering the city of Delft in the Netherlands as case setting, they showed that despite increased traffic
congestion due to empty vehicle relocation trips, vehicle automation could lead to more trip requests satis-
fied while reducing travel costs. Milakis et al. (2017) investigated future development opportunities for AVs
in the Netherlands and gave estimates for the potential impacts on transport planning, traffic management
and travel behavior over time horizons up to 2030 and 2050. Exploring the impact of shared AVs on urban
parking demand, Zhang et al. (2015) suggested that for AV adopter-users, up to 90% of parking demand
might be eliminated (also see Le Vine et al. (2015)). Harper et al. (2016) studied the influence of travel
with AVs for the elderly and people with travel-restrictive medical conditions and found a 14% increase in
annual vehicle miles traveled for the United States population 19 and older. Furthermore, Aria et al. (2016)
investigated AV effects on driver behavior and traffic performance, and the simulation results revealed that
the positive effects of AV on roads are especially highlighted when the road network is crowded. Diels
and Bos (2016) discussed a potential increase of motion sickness issues in AVs. Wadud (2017) focused on
identifying which vehicle sectors would likely be the earliest adopters of full automation in the UK, and
their findings suggests that households with the highest income will get higher gains from automations as
they travel higher distances.
Studies on the deployment of AVs in shared mobility systems include Chen et al. (2016a) who ran a sim-
ulation study on the city of Austin, Texas. The results suggested that AVs offer a viable alternative to private
vehicle travel (also see Fagnant and Kockelman (2014,2016); Chen et al. (2016b)). The study revealed that
each shared AV can replace 5–9 privately owned vehicles while serving 96–98% of trip requests. Bischoff
and Maciejewski (2016b) led a similar study on the city of Berlin, Germany, simulating the replacement
of hundreds of thousands of vehicles all around the city by a fleet of autonomous taxis. Results suggested
that the car fleet in Berlin can be replaced by a fleet of 100,000 autonomous taxis while maintaining high
service quality for customers (also see Bischoffand Maciejewski (2016c)). Another study, by Scheltes and
de Almeida Correia (2017), simulated a system in which the last-mile segment of train trips was carried
out by a fleet of fully autonomous vehicles. Results obtained from applying the simulation model on Delft,
Netherlands, argue that such a system is able to compete with walking mode but needs to improve its perfor-
mance to be competitive with cycling. Through a case study using taxicab trip data from New York City, Ma
and Zhang (2017) concluded that an AV sharing and reservation system can significantly increase vehicle
mileage rates while reducing their ownership rates. Another case study by Lokhandwala and Cai (2018)
20
Method
Scope Assessing system performance Studying impacts
Optimization-based Ma et al. (2017)Correia and van Arem (2016)
Simulation-based
Bischoffand Maciejewski (2016a)
Fagnant and Kockelman (2016)
Chen et al. (2016b)
Bischoffand Maciejewski (2016c)
Chen and Kockelman (2016)
Levin and Boyles (2015)
Scheltes and de Almeida Correia (2017)
Lokhandwala and Cai (2018)
Gruel and Stanford (2016)
Zhang et al. (2015)
Talebpour and Mahmassani (2016)
Fagnant and Kockelman (2014)
Milakis et al. (2017)
Harper et al. (2016)
Meyer et al. (2017)
Diels and Bos (2016)
Aria et al. (2016)
Smolnicki and Sołtys (2016)
Data-analysis-based -
Alessandrini et al. (2015)
Fagnant and Kockelman (2015)
Wadud (2017)
Table 4: Case studies - Shared autonomous mobility
suggested that replacing traditional taxis by shared AVs in New York City could potentially reduce the fleet
size by 59% while maintaining the sale service quality. The study concluded that sharing AVs can increase
occupancy rate (from 1.2 to 3) and decrease system-wide vehicle distance (up to 55%) (case studies are
classified in Table 4).
Unlike for ridesharing systems, case studies on shared autonomous mobility systems have mainly fo-
cused on studying their expected outcomes and effects on people mobility and on existing transportation
systems. This may be due to the fact that the introduction of autonomous mobility services in transportation
systems is a new trend in transportation research, and so its potential impacts need to be studied and carefully
analyzed before it can be widely adopted. However, most of studies reviewed have used simulation-based
approaches, which is evidence that these new systems are still in an early stage of research. As a result,
more research studies will be directed towards studying their operational performance as soon as they are
widely deployed.
4. People and goods sharing rides
This section focuses on introducing shared mobility systems that combine both passenger and freight
transportation. First, we set the context by reviewing the most recent concepts and trends in city logistics
(Section 4.1). Then, the opportunities and challenges that can result from combining people and freight
flows are discussed and their modeling and solution approaches are investigated (Section 4.2). Table 5
summarizes the papers reviewed with their different characteristics and methods used.
4.1. Setting the context: planning city logistics
The demand for freight transportation basically results from the need to transport goods from producers
to consumers who are geographically apart. In general, this transportation chain consists of a pickup process
(pre-haul or first-mile), a transportation process (long haul), and a delivery process (end-haul or last-mile)
(Steadieseifi et al. (2014)). While freight transportation can take place in widespread geographical areas,
21
city logistics considers the transportation of goods and their potential effects on traffic flow and congestion in
urban areas (Savelsbergh and Van Woensel (2016)). However, both freight transportation and city logistics
aim to provide customers with the products they need at the right time and place and at low cost.
Increasing global population, and thus increasing demand for goods, together with digital revolution
and technological advances are creating both opportunities and challenges for planning and improving the
sustainability of urban freight systems. Given their fundamental role in providing for people’s daily needs,
efficient city logistics have the potential to improve quality of life for more and more people. Recent studies
in this direction have focused on anticipating the future opportunities and challenges facing city logistics.
In their recent review on city logistics, Savelsbergh and Van Woensel (2016) identified the trends driving
changes in city logistics: growing urban populations, increasing importance of e-commerce and swift supply
chains, and the rise of the sharing economy and sustainability aspects. They claimed that sharing assets
and capacities can enable higher capacity utilization, and thus reduce fleet sizes and numbers of freight
movements. Besides studying the impacts of the information revolution on city logistics, Taniguchi et al.
(2016) also described applications of big data and decision-support systems that can be used to enhance the
design and evaluation of city logistics schemes, and gave illustrations of the need for new innovations that
can help reduce the impact of freight in urban areas. One of the most common scenarios for reducing the
number of freight vehicles going into cities is to consolidate goods volume at urban distribution centers,
called consolidated distribution centers (CDC), which are usually located a city’s borders (Allen et al.
(2012)). In this scenario, cargo is delivered to a CDC by different supply chain operators, consolidated at
the center, and then shipped to final customers using clean and highly utilized vehicles (Alessandrini et al.
(2015); Figure 2). The main advantage of using CDCs is that shipments can be grouped by destination into
packages where every package will be transported using a vehicle. This way, the number of vehicle trips
and the need for parking bays can be reduced, this affording a more efficient delivery service.
Figure 2: Consolidated Distribution Center (CDC) with vehicle deliveries
In order to make it efficient, these vehicles have to be small, agile, have large enough loading capacity
and comply with the environmental requirements governing energy consumption, CO2emissions, and noise.
The problem of planning and optimizing itineraries of such a fleet, in which vehicles operate round trips,
is called the vehicle routing problem (VRP) (see Cattaruzza et al. (2015) for a review of VRPs for city
logistics and Koc¸ and Laporte (2018) for a review of VRPs with backhauls). Solving a VRP is about defining
routes that respect a number of constraints, including pickup and delivery locations, time windows, vehicle
22
capacity, narrow streets with limited accessibility, and other constraints. For example, Simoni et al. (2018)
proposed a heuristic approach for routing vehicles carrying parcels from CDCs to their final destinations
within an urban area, and identified the most efficient and environment-friendly strategies and regulations
for this delivery.
Another promising opportunity in city logistics is the deployment of autonomous mobility services.
With their potential application in future freight transportation systems, AVs might be used for refilling
shops from remote warehouses, performing last-mile deliveries to clients, and collecting and transporting
waste and products (CityMobil (2011);Alessandrini et al. (2015)). However, assessing the benefits of AVs
and their efficient employment and impacts on city logistics is an important topic in today’s research, and
still requires further investigation. Many freight transportation companies have started using unmanned
ground vehicles and unmanned aerial vehicles (drones) for small parcel deliveries (see Otto et al. (2018)
for a recent review). The idea is that these relatively small unmanned vehicles will depart from warehouses
or delivery trucks carrying small deliveries for individual customers. For example, Murray and Chu (2015)
studied a problem in which delivery trucks carrying drones depart from and return to a depot. In their
settings, customers are served either by delivery trucks or by drones that operate in coordination with the
delivery trucks. Depending on its flight endurance, a drone has to deliver the customer’s order and return to
either the truck or a depot, the aim being to minimize the time required to deliver all customer orders. Such a
system is thought to provide a more efficient delivery service at lower cost and with reduced environmental
impacts.
Furthermore, another important innovation is the potential for delivering customer orders to more con-
venient locations than the home (e.g. direct delivery to a customer’s car trunk (Savelsbergh and Van Woensel
(2016)). Thanks to new technologies, a one-time access to customer a car trunk can be granted during a
specific time-period and revoked as soon as the delivery is completed. In their recent paper, Reyes et al.
(2017) modeled this last-mile trunk delivery as a VRP with roaming delivery locations (VRPRDL). In their
approach, the delivery locations are first optimized for a fixed customer delivery sequence in order to gener-
ate an initial route. Then, the initial route is improved by switching a predecessor’s or successor’s delivery
location once a customer is inserted or deleted. Results reveal that trunk delivery could potentially cut
distance traveled in tests with realistic instances.
In such systems, some locations may change or move as the delivery process starts (like the location
of a delivery truck a drone is to return to in Murray and Chu (2015), and the roaming delivery locations in
Reyes et al. (2017)). Although this feature might lead to more flexible deliveries, it requires more complex
models and sophisticated heuristic approaches, due to the layer of complexity added by the synchronization
constraints required to adapt different departure and arrival times to these roaming locations. Reyes et al.
(2017), for example, proposed a neighborhood search heuristic with a set of insertion and deletion operators,
and considered the roaming delivery locations when building routes by enhancing the classical VRP inser-
tion and deletion operators by including customer shifts to different delivery locations and consequently
different time-windows within them. Thus, at each a time a new route is built, a set of alternative routes,
where precisely one customer delivery location is different to the original route, are generated. However,
due to the added complexity, generating these alternative routes requires additional computational effort.
Building on this review of the latest trends in city logistics, we focus in the following subsection on the
promising concept of integrating people and freight flows for future transportation systems development.
4.2. Combining people and freight transportation
Since both people and goods move in the urban environment, an efficient and effective transport network
that ensures smooth sharing of passengers and freights is an essential element in city life (Cochrane et al.
(2017)). There is ample literature on the problem of passengers or goods transport using dedicated networks,
23
but far less research on joint use of transport resources between passengers and goods flows. However,
combined transportation systems are starting to garner increasing attention.
A combined transportation system aims to use the underused assets in public mass-transport modes
such as urban rail, buses or in people private-car trips to bring loads to a central station or take loads from
that station to distribute it to the local neighborhood (Crainic and Montreuil (2016)). In such a system, we
have a set of passengers and parcels, each having an origin location from where it should be picked up,
and a destination location to where it should be carried and dropped off. We also have a transportation
system, having both private and public transportation modes, which is able to transport both passengers
and parcels simultaneously. Thus, the aim is to satisfy the demand of both passengers and parcels while
minimizing costs and distances traveled, and therefore reduce congestion and pollution levels in urban areas.
Of course, the transportation of goods must not disturb passenger trips. In other words, a passenger would
accept only small deviations and short extra times for transporting parcels in the same trip. Thus, trip times
that significantly exceed a passenger’s usual route times in order to load and deliver parcels would likely
be unacceptable. Although most problems dealing with passengers and goods transportation are NP-hard,
so very difficult to solve, many studies have attempted to tackle them with different models and solution
approaches. These models fall into two broad categories: single-tiered and two-tiered models (see Figure 3).
A single-tiered model considers a set of vehicles each having specific capacity. These vehicles are able to
transport passengers and goods to their destinations while accommodating certain like passenger and parcel
time windows and vehicle capacity and service time (see Li et al. (2014)). In a two-tiered model, combined
transport of passengers and goods is achieved via the contribution of a first-tier, generally composed of a
public transport line with a set of transfer points or stations, and a second tier, composed of a range of
vehicles being able to transport both passengers and goods (or only goods depending on the model studied)
from transfer points to their final destinations (see Trentini et al. (2015)). Strict synchronization between the
two tiers is therefore necessary. For planning and operating such combined systems, different models and
solution approaches have recently been proposed, most of which aim to minimize their operational costs,
or put differently, maximize their benefits. However, some studies have considered other objectives like
minimizing the number of vehicles required for making deliveries, minimizing total distance traveled, or
minimizing the wait time for deliveries. Below we take a deeper look at the existing models in the literature.
Li et al. (2014) extended the classical DARP formulation by introducing a new class of models called
the share-a-ride problem (SARP). The SARP refers to the fact that people and parcels are transported using
a set of taxis driving around in a city. The proposed model is therefore single-tiered. In this problem,
passenger requests are served by a fleet of taxis, and some parcels are delivered during these taxi trips as
long as delivery does not affect the passengers significantly. Passengers thus have priority over parcels.
Furthermore, the SARP assumes that a taxi cannot serve two passengers simultaneously and that a parcel
cannot be served by more than one taxi, i.e. it is either served by one taxi or not served at all. Another basic
assumption in SARP is that parcel transportation requests are known beforehand whereas passenger requests
arrive dynamically. In addition to the SARP, the authors propose a second model, which has similar settings
but with the assumption that the assignments of passengers to taxis and their delivery sequences are also
given. In this case, dubbed the freight insertion problem (FIP), the problem becomes static (see Figure 4).
Solving the FIP is about finding a way to insert parcel requests without significantly extending passenger
travel times. Since routing is given, the FIP has less complexity than the SARP, and can thus be solved
relatively fast, at least fast enough for solving real-life instances. To solve this problem, authors present
MILP formulations for both SARP and FIP and conduct a numerical study of both static and dynamic
scenarios. Given the complexity of the problem, the authors proposed an ALNS to solve it (Li et al. (2016a)).
The proposed approach was able to return solutions that are within 2.24% of the best results compared to
24
Figure 3: Single-tiered and two-tiered transportation systems
a mixed integer programming (MIP) solver and DARP test benchmarks from the literature. Beirigo et al.
(2018) introduced another SARP formulation where a fleet of SAVs is used to serve both passenger and
freight requests. The paper extends the original SARP formulation by allowing vehicles (in this case SAVs)
to carry one or more passengers and different-sized parcels in the same trip. To solve the extended problem,
the authors proposed a MILP formulation and analyzed it on a wide set of transportation scenarios. The
SARP study was further extended by considering two stochastic variants; one with stochastic travel times
and another with stochastic delivery locations (Li et al. (2016b)). In both cases, a two-stage stochastic
programming model with recourse is used with the ALNS heuristic and a scenario generator. Results
obtained from testing both stochastic models demonstrate that even though the convergence rate is faster,
the SARP is less sensitive to the stochastic delivery locations than the stochastic travel times. The study
thus concluded that considering stochastic information when modeling and planning real-life taxi-sharing
systems can dramatically improve their performance over deterministic solutions.
Arslan et al. (2016) proposed another single-tiered model in a study on the concept of crowd-sourcing
delivery, which aims to make parcel deliveries using excess capacity on trips that already take place (see
Mladenow et al. (2015); Goetting and Handover (2016) for recent reviews of the latest crowd-sourced
delivery models). For this purpose, the paper considered a decentralized model that automatically matches
parcel delivery requests to potential ad-hoc drivers. Parcel deliveries are made by self-employed drivers
who are willing to earn extra money on their way to home or work. The drivers indicate their origin and
destination locations, their vehicle capacity, and a time window. Likewise, parcel delivery requests also
have time windows that state when they should be picked up and delivered. Thus, a delivery is possible if
there is a feasible match between driver’s time window and parcel’s time window. A set of backup vehicles
is operated to cover parcel requests that cannot be delivered by ad-hoc drivers. Furthermore, the paper
presents an event-based rolling horizon framework that dynamically matches tasks to drivers at each time a
new delivery request or driver arrives throughout the day, as well as exact and heuristic recursive algorithms
for solving the routing subproblem. Results show that using ad-hoc drivers can potentially reduce last-mile
delivery costs as well as system-wide vehicle mileage. Archetti et al. (2016) considered a similar problem
to Arslan et al. (2016) but in their setting, a service provider uses not only a fleet of delivery vehicles
25
Figure 4: An illustrative example of the SARP and the FIP (Li et al. (2014))
and dedicated drivers but also a set of occasional drivers who are willing to make a single delivery using
their own vehicle. Making these deliveries should not significantly extend the trip time for the occasional
driver, who can then receive a small cost compensation for each delivery they perform.. To model the
problem, Archetti et al. (2016) introduced a new variant of the classical capacitated VRP called the ’vehicle
routing problem with occasional drivers’. The paper also presents a heuristic approach in which variable
neighborhood and tabu search strategies are combined to produce good quality solutions. Wang et al. (2016)
presented a crowd-tasking model in which last-mile deliveries are performed by a crowd of citizen workers,
and proposed to formulate the model as a network min-cost flow problem and use an iterative pruning
technique to make the network manageably small. Dayarian and Savelsbergh (2017) proposed another
crowd-sourced service in which customers can deliver some online orders. Potential customers express
an interest to participate in making deliveries on their way home, and thus supplement a set of dedicated
drivers performing the service, with vehicle routes generated using a tabu search heuristic. A number of
papers have considered transporting freight by the same rail network as passengers (see Steadieseifi et al.
(2014); Cochrane et al. (2017); Ozturk and Patrick (2017)), but they are outside our scope as the models do
not integrate passengers and freight in the same trip (i.e. rail is used during passenger off-hours).
Recent research has also focused on two-tiered models. Trentini et al. (2015) introduced a combined
system that uses the available capacity in a passenger bus line to transport parcels (also see Trentini et al.
(2013)). In their problem settings, all incoming goods are stored in CDCs, then loaded on buses operating
through the bus line when there is spare capacity, and finally unloaded at specific bus stops and delivered
to customers using a fleet of low-emission city freighters. The proposed problem is modeled as a VRP
with transfers, and a mathematical formulation is given, along with an ALNS to solve it (see Masson et al.
(2017), for a similar system). Fatnassi et al. (2015) proposed another two-tiered shared passengers and
goods model with a first tier (train, bus or truck line) transporting passengers or goods to connection points
where a second tier, consisting of a set of small electric and AVs moving on a specific guideway then
transport them to their final destinations. The paper uses a forward periodic-optimization approach to solve
26
this dynamic problem.
Behiri et al. (2018) studied the freight-rail transport scheduling problem in which existing urban rail is
used for transporting freight. In their model, one rail line is considered. On this line, there are several sta-
tions where freight can be loaded and unloaded. Freights are brought to these stations by truck at different
time windows in a day. To solve this problem, the paper proposes two heuristic approaches: a dispatch-
ing rule-based heuristic and a single-train decomposition-based heuristic. Similarly, Ghilas et al. (2016a)
considered a system in which freight requests are delivered by a set of vehicles such that a part of the trans-
portation process is carried out on a scheduled public transport line. To model this two-tiered system, the
paper introduces the pickup and delivery problem with time windows and scheduled lines (PDPTW-SL). In
this problem, two options are considered for transporting freight: direct and indirect shipments. In a direct
shipment, a freight request is picked up at its origin and delivered to its destination using one vehicle, i.e. the
scheduled line is not used. In an indirect shipment, a freight request is picked up by a vehicle, transferred
to a nearby transfer node, transported between two transfer nodes by a scheduled public transport line, and
finally picked up by another vehicle and delivered to its final destination. Thus, solving the problem is
about defining routes and schedules for both freight requests and delivery vehicles. In order to solve this
problem, the authors proposed a branch-and-price algorithm where the pricing problem is a variant of the
elementary shortest path problem with resource and precedence constraints (ESPPRPC). Due to the com-
plexity of the problem, an ALNS-based algorithm is also proposed (see Ghilas et al. (2016b)). Moreover,
a stochastic version of the problem in which the demand quantity of each freight request is only revealed
when the vehicle arrives at its pickup location was considered (Ghilas et al. (2016c)). To consider this
uncertainty, a scenario-based sample average approximation approach is introduced. Another two-tiered
crowd-sourced delivery system (Kafle et al. (2017)) suggested that a set of cyclists and pedestrians, called
crowd-sources, might be willing to deliver small-size parcels from a delivery truck to customers living in
the same neighborhood. A set of carrier trucks transport parcels to intermediate transfer points (first-tier)
and then potential crowd-sources perform the last-mile delivery. To solve this problem, the paper proposes
a tabu search algorithm. Results show that crowd-sourcing the service can lead to lower operational costs
compared with a pure-truck delivery service.
This review on systems that combine people and freight transportation shows that the topic is gaining
increasing interest (see Table 5for a summary). Models and algorithms for both single-tiered and two-tiered
systems have been explored. Although some papers have introduced exact approaches for solving this type
of problem, the bulk of the research has focused on developing heuristic approaches. This is due to the
complexity of such problems which require fast optimization approaches to tackle them in short compu-
tation times. We also find that most of the papers reviewed have focused on profitability. Nevertheless,
other objectives have also been considered, such as minimizing the number of vehicles needed to operate
the system and the distances covered. Thus, an useful direction for future research would be to also address
the environmental issues which have not yet been considered in the literature. We would prone the fol-
lowing broad areas for future research: (i) developing efficient solution algorithms (exact and heuristic) for
combined people-and-freight systems, (ii) extending the existing models by introducing multiple objectives
related to profit, operational costs, environmental impacts, etc., (iii) developing more flexible models and
efficient algorithms that consider the different sources of stochastic information (travel times, traffic jams,
freight demands etc.), (iv) improving the dynamic (real-time) framework of such systems by adding new
techniques and strategies (leading to shorter service times, strong synchronization between different tiers in
two-tiered models, etc.), (v) introducing new public policies to regulate the potential integration of goods
delivery in existing public transport systems, (vi) focusing more on increasing passenger satisfaction and
reducing the potential inconvenience that might arise in such systems, and finally (vii) studying the potential
27
Reference Problem variant Method Objective Constraints
RC TC CC OC SC
Characteristic
Trentini et al. (2013) Combined Del. H N, D X X X X Two-tiered
Trentini et al. (2015) Combined Del. H N, C X X X X Two-tiered
Fatnassi et al. (2015) Combined Del. H CX X X X Two-tiered
Li et al. (2016a) SARP H CXXX Single-tiered
Li et al. (2016b) Stochastic SARP H CXXX Single-tiered
Arslan et al. (2016) Crowd-sourcing E CXXX Single-tiered
Archetti et al. (2016) Crowd-sourcing H CXXX Single-tiered
Ghilas et al. (2016a) Combined Del. E, H CX X X X Two-tiered
Ghilas et al. (2016b) Combined Del. H CX X X X Two-tiered
Ghilas et al. (2016c) Combined Del. H CX X X X Two-tiered
Wang et al. (2016) Crowd-sourcing H CX X X X Single-tiered
Kafle et al. (2017) Crowd-sourcing H CX X X X Two-tiered
Dayarian and Savelsbergh (2017) Crowd-sourcing H WXXX Single-tiered
Masson et al. (2017) Combined Del. H N, C X X X X Two-tiered
Behiri et al. (2018) Combined Del. H TX X X X Two-tiered
Beirigo et al. (2018) SARP E CX X X X Single-tiered
Method: E: Exact approach, H: Heuristic approach.
Objectives: D: Min. Travel Distance, T: Min. Waiting Time, N: Min. Number of Vehi-
cles, C: Min. Operational Cost, W: Max. Collected Weight.
Table 5: Shared mobility - Combined people-and-freight systems
deployment of automated services and their impacts on the future development of such combined systems.
4.3. Case studies
Given its potential benefits, the integration of passenger and freight transportation streams has been
assessed in a number of studies in the last three years. Most of these studies have focused on analyzing
the performance of such integrated systems and evaluating their operational gains compared to the existing
transportation systems. A good example can be found in Fatnassi et al. (2015) which considers a case study
on the town of Corby (UK). The results demonstrate potential benefit of implementing a combined system
in terms of service time, energy consumed, noise and carbon emissions compared to classical transportation
systems. Ghilas et al. (2016c) suggested these combined systems can bring up to 16% savings on overall
operational cost. Considering real taxi trips in the San-Francesco area, Li et al. (2016a) showed that a
mixed-taxi service can outperform the other transportation systems available in the local urban area, but
also highlighted two key factors to help maximize the gain obtained by such a service: analysis of the
spatial characteristics of requests before implementing the service, and availability of a traditional freight
service to ensure that all requests are delivered. Gonzalez-Feliu and Mercier (2013) studied the potential
deployment of a combined people-freight approach in the city of Lyon (France) and found that it was crucial
to apply an accessibility analysis that shows the attractiveness of different urban zones before this combined
system can take place. Thus, the difficulty for households living at different city zones to reach their retailers
should be considered when deploying the system. Wang et al. (2016) evaluated their crowd-tasking model
using datasets from bus and taxi services in Singapore, and their results demonstrate that crowd-sourcing
can be efficiently used in large-scale problems with real-time deliveries where this kind of service can
be profitable to logistic companies as well as crowd-workers. More recently, Masson et al. (2017) led
28
a case study based on a dataset derived from the city of La Rochelle (France) and found that efficient
transshipment of freight from buses to city freighters is a major concern in a mixed system, as inefficient
transshipment of freight between the two tiers might delay deliveries and significantly affect passenger
trips. Although most case studies are ultimately optimistic over the future of combining passenger and
freight flows, some of the allied concerns and practical issues still need to be investigated. These issues
involve, among others, (i) security concerns, (ii) confidentiality and data privacy (like using only a barcode
with limited personal information to identify parcels), (iii) the redesign of parcels with different sizes to
fit in the shared transportation compartments, and (iv) uncertainties during deliveries (e.g. freight order
modifications and cancellations). We would thus advocate more studies to evaluate these issues and study
their impacts on the future deployment of these integrated systems.
5. Conclusions
This survey reviewed different variants of shared mobility systems along with their modeling choices
and solution approaches. The papers reviewed covered mobility systems where people share their rides
and mobility systems where people and goods are combined. We presented a set of case studies either
analyzing shared mobility system performances or studying their potential impacts on people’s lives and
future transportation systems.
New shared mobility systems for both people and freight transportation have the potential to provide
major societal, economic and environmental benefits. The development of algorithms for planning and
operating such systems is at the heart of the shared mobility concept. This survey highlighted a number
of promising optimization opportunities and challenges that arise when developing new systems to support
shared mobility. Relevant operations research models in this area have also been reviewed. Although
ridesharing is not a new concept, we have seen that the interest in enhancing dynamic ridesharing systems
and developing new systems for matching passengers on-the-fly continues to grow. More research is now
needed on systems that consider trip synchronization and traveler cost aspects, or more generally the quality
of the provided service. One of the latest big trends appears to be research on deploying new autonomous
mobility services. We now need more research on how these new services can operate and how they can
impact future transportation systems, and there is also a need to introduce new models and algorithms
that consider vehicle charging and maintenance operations. The potential integration of passenger and
freight transportation is another promising opportunity that is steadily gaining currency. As such, more
studies on developing realistic models and efficient algorithms that consider different objectives (including
environmental issues) and different sources of uncertainty are also needed, along with new public policies
to regulate this integration. We believe that these challenges and new innovations provide a rich vein of
research opportunities, and we anticipate that this review could spur more contributions in this emerging
area of transportation science.
Acknowledgment
This research work was carried out as part of the ANTHROPOLIS research project at the Technological
Research Institute SystemX, and was supported by public funding within the scope of the French Program
”Investissements d’Avenir”.
29
Bibliography
N. Agatz, A. Erera, M. Savelsbergh, and X. Wang. Optimization for dynamic ride-sharing: A review. European Jour-
nal of Operational Research, 223(2):295–303, dec 2012. ISSN 03772217. doi: 10.1016/j.ejor.2012.05.028. URL http:
//linkinghub.elsevier.com/retrieve/pii/S0377221712003864.
N. A. H. Agatz, A. L. Erera, M. W. P. Savelsbergh, and X. Wang. Dynamic ride-sharing: A simulation study in metro Atlanta.
Transportation Research Part B: Methodological, 45 ‘(9):1450–1464, 2011. ISSN 01912615. doi: 10.1016/j.trb.2011.05.017.
URL http://dx.doi.org/10.1016/j.trb.2011.05.017.
A. Alessandrini, A. Campagna, P. D. Site, F. Filippi, and L. Persia. Automated vehicles and the rethinking of mobility and
cities. Transportation Research Procedia, 5:145–160, 2015. ISSN 23521465. doi: 10.1016/j.trpro.2015.01.002. URL http:
//dx.doi.org/10.1016/j.trpro.2015.01.002.
J. Allen, M. Browne, A. Woodburn, and J. Leonardi. The Role of Urban Consolidation Centres in Sustainable Freight Transport.
Transport Reviews, 32(4):473–490, 2012. ISSN 01441647. doi: 10.1080/01441647.2012.688074.
J. Alonso-Mora, S. Samaranayake, A. Wallar, E. Frazzoli, and D. Rus. On-demand high-capacity ride-sharing via dynamic trip-
vehicle assignment. Proceedings of the National Academy of Sciences, page 201611675, 2017. ISSN 0027-8424. doi: 10.1073/
pnas.1611675114. URL http://www.pnas.org/lookup/doi/10.1073/pnas.1611675114.
C. Archetti, M. Savelsbergh, and M. G. Speranza. The Vehicle Routing Problem with Occasional Drivers. European Journal of
Operational Research, 254(2):472–480, 2016. ISSN 03772217. doi: 10.1016/j.ejor.2016.03.049. URL http://dx.doi.org/
10.1016/j.ejor.2016.03.049.
E. Aria, J. Olstam, and C. Schwietering. Investigation of Automated Vehicle Effects on Driver’s Behavior and Traffic Performance.
Transportation Research Procedia, 15:761–770, 2016. ISSN 23521465. doi: 10.1016/j.trpro.2016.06.063. URL http://dx.
doi.org/10.1016/j.trpro.2016.06.063.
A. Arslan, N. Agatz, L. G. Kroon, and R. A. Zuidwijk. Crowdsourced Delivery - A Pickup and Delivery Problem with Ad-hoc
Drivers. SSRN Electronic Journal, pages 1–29, 2016. ISSN 1556-5068. doi: 10.2139/ssrn.2726731. URL http://www.ssrn.
com/abstract=2726731.
A. Atahran, C. Lent´
e, and V. T’kindt. A Multicriteria Dial-a-Ride Problem with an Ecological Measure and Heterogeneous
Vehicles. Journal of Multi-Criteria Decision Analysis, 21(5-6):279–298, sep 2014. ISSN 10579214. doi: 10.1002/mcda.1518.
URL http://dx.doi.org/10.1002/mcda.442http://doi.wiley.com/10.1002/mcda.1518.
R. Baldacci, V. Maniezzo, and A. Mingozzi. An Exact Method for the Car Pooling Problem Based on Lagrangean Column
Generation. Operations Research, 52(3):422–439, 2004. ISSN 0030-364X. doi: 10.1287/opre.1030.0106. URL http:
//pubsonline.informs.org/doi/abs/10.1287/opre.1030.0106.
P. Bansal and K. M. Kockelman. Are we ready to embrace connected and self-driving vehicles? A case study of Texans. Trans-
portation, pages 1–35, 2016. ISSN 15729435. doi: 10.1007/s11116-016-9745- z.
P. Bansal, K. M. Kockelman, and A. Singh. Assessing public opinions of and interest in new vehicle technologies: An Austin
perspective. Transportation Research Part C: Emerging Technologies, 67:1–14, 2016. ISSN 0968090X. doi: 10.1016/j.trc.
2016.01.019. URL http://dx.doi.org/10.1016/j.trc.2016.01.019.
B. Barann, D. Beverungen, and O. M¨
uller. An open-data approach for quantifying the potential of taxi ridesharing. Decision
Support Systems, 99(2016):86–95, 2017. ISSN 01679236. doi: 10.1016/j.dss.2017.05.008. URL http://linkinghub.
elsevier.com/retrieve/pii/S0167923617300866.
W. Behiri, S. Belmokhtar-berraf, and C. Chu. Urban freight transport using passenger rail network : Scientific issues
and quantitative analysis. Transportation Research Part E, 115(February 2017):227–245, 2018. ISSN 1366-5545. doi:
10.1016/j.tre.2018.05.002. URL https://doi.org/10.1016/j.tre.2018.05.002.
B. A. Beirigo, F. Schulte, and R. R. Negenborn. Integrating People and Freight Transportation Using Shared Autonomous
Vehicles with Compartments. IFAC-PapersOnLine, 51(9):392–397, 2018. ISSN 24058963. doi: 10.1016/j.ifacol.2018.
07.064. URL https://doi.org/10.1016/j.ifacol.2018.07.064https://linkinghub.elsevier.com/retrieve/
pii/S2405896318307870.
G. Berbeglia, J.-f. Cordeau, and G. Laporte. A Hybrid Tabu Search and Constraint Programming Algorithm for the Dynamic Dial-
a-Ride Problem. INFORMS Journal on Computing, 24(3):343–355, aug 2012. ISSN 1091-9856. doi: 10.1287/ijoc.1110.0454.
URL http://pubsonline.informs.org/doi/abs/10.1287/ijoc.1110.0454.
J. Bischoffand M. Maciejewski. Autonomous taxicabs in Berlin – a spatiotemporal analysis of service performance. Interna-
tional Scientific Conference on Moiblity and Transport Transforming Urban Mobility, mobil.TUM, 00(June):11, 2016a. ISSN
23521465. doi: 10.1016/j.trpro.2016.12.078. URL http://dx.doi.org/10.1016/j.trpro.2016.12.078.
J. Bischoffand M. Maciejewski. Simulation of City-wide Replacement of Private Cars with Autonomous Taxis in Berlin. Procedia
Computer Science, 83(Ant):237–244, 2016b. ISSN 18770509. doi: 10.1016/j.procs.2016.04.121. URL http://dx.doi.org/
10.1016/j.procs.2016.04.121.
J. Bischoffand M. Maciejewski. Autonomous taxicabs in Berlin – a spatiotemporal analysis of service performance. Interna-
30
tional Scientific Conference on Moiblity and Transport Transforming Urban Mobility, mobil.TUM, 00(June):11, 2016c. ISSN
23521465. doi: 10.1016/j.trpro.2016.12.078. URL http://dx.doi.org/10.1016/j.trpro.2016.12.078.
K. Braekers and A. A. Kovacs. A multi-period dial-a-ride problem with driver consistency. Transportation Research Part B:
Methodological, 94:355–377, 2016. ISSN 01912615. doi: 10.1016/j.trb.2016.09.010. URL http://dx.doi.org/10.1016/
j.trb.2016.09.010.
K. Braekers, K. Ramaekers, and I. Van Nieuwenhuyse. The vehicle routing problem: State of the art classification and review.
Computers and Industrial Engineering, 99:300–313, 2016. ISSN 03608352. doi: 10.1016/j.cie.2015.12.007. URL http:
//dx.doi.org/10.1016/j.cie.2015.12.007.
D. Cattaruzza, N. Absi, D. Feillet, and J. Gonz´
alez-Feliu. Vehicle routing problems for city logistics. EURO Journal on Trans-
portation and Logistics, pages 1–29, 2015. ISSN 2192-4384. doi: 10.1007/s13676-014-0074- 0. URL http://dx.doi.org/
10.1007/s13676-014-0074-0.
N. D. Chan and S. A. Shaheen. Ridesharing in North America: Past, Present, and Future. Transport Reviews, 32(1):93–112, 2012.
ISSN 01441647. doi: 10.1080/01441647.2011.621557.
C. Chen, J. Ma, Y. Susilo, Y. Liu, and M. Wang. The promises of big data and small data for travel behavior (aka human
mobility) analysis. Transportation Research Part C: Emerging Technologies, 68:285–299, 2016a. ISSN 0968090X. doi:
10.1016/j.trc.2016.04.005. URL http://dx.doi.org/10.1016/j.trc.2016.04.005.
T. D. Chen and K. M. Kockelman. Management of a Shared Autonomous Electric Vehicle Fleet. Transportation Research Record:
Journal of the Transportation Research Board, 2572(2572):37–46, 2016. ISSN 0361-1981. doi: 10.3141/2572-05. URL
http://trrjournalonline.trb.org/doi/10.3141/2572-05.
T. D. Chen, K. M. Kockelman, and J. P. Hanna. Operations of a shared, autonomous, electric vehicle fleet: Implications of vehicle
& charging infrastructure decisions. Transportation Research Part A: Policy and Practice, 94:243–254, 2016b. ISSN 09658564.
doi: 10.1016/j.tra.2016.08.020. URL http://dx.doi.org/10.1016/j.tra.2016.08.020.
Z. Chen, F. He, Y. Yin, and Y. Du. Optimal design of autonomous vehicle zones in transportation networks. Transportation
Research Part B: Methodological, 99:44–61, 2017. ISSN 01912615. doi: 10.1016/j.trb.2016.12.021.
K. Cochrane, S. Saxe, M. J. Roorda, and A. Shalaby. Moving freight on public transit: Best practices, challenges, and opportuni-
ties. International Journal of Sustainable Transportation, 11(2):120–132, 2017. doi: 10.1080/15568318.2016.1197349. URL
https://doi.org/10.1080/15568318.2016.1197349.
J. F. Cordeau and G. Laporte. The dial-a-ride problem: Models and algorithms. Annals of Operations Research, 153(1):29–46,
2007. ISSN 02545330. doi: 10.1007/s10479-007- 0170-8.
G. H. d. A. Correia and B. van Arem. Solving the User Optimum Privately Owned Automated Vehicles Assignment Problem
(UO-POAVAP): A model to explore the impacts of self-driving vehicles on urban mobility. Transportation Research Part B:
Methodological, 87:64–88, 2016. ISSN 01912615. doi: 10.1016/j.trb.2016.03.002.
T. G. Crainic and B. Montreuil. Physical Internet Enabled Hyperconnected City Logistics. Transportation Research Procedia, 12
(June 2015):383–398, 2016. ISSN 23521465. doi: 10.1016/j.trpro.2016.02.074.
I. Dayarian and M. W. P. Savelsbergh. Crowdshipping and Same-day Delivery: Employing In-store Customers to Deliver Online
Orders. Optimization Online, 07-6142(Foundation), 2017.
R. A. Daziano, M. Sarrias, and B. Leard. Are consumers willing to pay to let cars drive for them? Analyzing response to
autonomous vehicles. Transportation Research Part C, 78:150–164, 2016. ISSN 0968090X. doi: 10.1016/j.trc.2017.03.003.
URL http://dx.doi.org/10.1016/j.trc.2017.03.003.
C. Diels and J. E. Bos. Self-driving carsickness. Applied Ergonomics, 53:374–382, 2016. ISSN 18729126. doi: 10.1016/j.apergo.
2015.09.009. URL http://dx.doi.org/10.1016/j.apergo.2015.09.009.
M. Drexl. Synchronization in Vehicle Routing–A Survey of VRPs with Multiple Synchronization Constraints. Transportation
Science, 46(3):297–316, 2012. ISSN 0041-1655. doi: 10.1287/trsc.1110.0400.
D. J. Fagnant and K. Kockelman. Preparing a nation for autonomous vehicles: Opportunities, barriers and policy recommendations.
Transportation Research Part A: Policy and Practice, 77:167–181, 2015. ISSN 09658564. doi: 10.1016/j.tra.2015.04.003. URL
http://dx.doi.org/10.1016/j.tra.2015.04.003.
D. J. Fagnant and K. M. Kockelman. The travel and environmental implications of shared autonomous vehicles, using agent-
based model scenarios. Transportation Research Part C: Emerging Technologies, 40:1–13, 2014. ISSN 0968090X. doi:
10.1016/j.trc.2013.12.001. URL http://dx.doi.org/10.1016/j.trc.2013.12.001.
D. J. Fagnant and K. M. Kockelman. Dynamic ride-sharing and fleet sizing for a system of shared autonomous vehicles in Austin,
Texas. Transportation, pages 1–16, 2016. ISSN 15729435. doi: 10.1007/s11116- 016-9729-z.
E. Fatnassi, J. Chaouachi, and W. Klibi. Planning and operating a shared goods and passengers on-demand rapid transit system
for sustainable city-logistics. Transportation Research Part B: Methodological, 81:440–460, 2015. ISSN 01912615. doi:
10.1016/j.trb.2015.07.016. URL http://dx.doi.org/10.1016/j.trb.2015.07.016.
M. Ferreira and P. M. D’Orey. Can ride-sharing become attractive? A case study of taxi-sharing employing a simulation modelling
approach. IET Intelligent Transport Systems, 9(2):210–220, 2015. ISSN 1751-956X. doi: 10.1049/iet-its.2013.0156. URL
31
http://digital-library.theiet.org/content/journals/10.1049/iet-its.2013.0156.
M. Fink, G. Desaulniers, M. Frey, F. Kiermaier, R. Kolisch, and F. Soumis. Column generation for vehicle routing problems with
multiple synchronization constraints. European Journal of Operational Research, 0:1–13, jul 2018. ISSN 03772217. doi:
10.1016/j.ejor.2018.06.046. URL https://doi.org/10.1016/j.ejor.2018.06.046https://linkinghub.elsevier.
com/retrieve/pii/S0377221718305988.
M. Furuhata, M. Dessouky, F. Ord´
o˜
nez, M. E. Brunet, X. Wang, and S. Koenig. Ridesharing: The state-of-the-art and future
directions. Transportation Research Part B: Methodological, 57:28–46, 2013. ISSN 01912615. doi: 10.1016/j.trb.2013.08.012.
URL http://dx.doi.org/10.1016/j.trb.2013.08.012.
T. Garaix, C. Artigues, D. Feillet, and D. Josselin. Optimization of occupancy rate in dial-a-ride problems via linear frac-
tional column generation. Computers &Operations Research, 38(10):1435–1442, oct 2011. ISSN 03050548. doi:
10.1016/j.cor.2010.12.014. URL http://dx.doi.org/10.1016/j.cor.2010.12.014http://linkinghub.elsevier.
com/retrieve/pii/S0305054810003102.
E. Gargiulo, R. Giannantonio, E. Guercio, C. Borean, and G. Zenezini. Dynamic Ride Sharing Service: Are Users Ready to
Adopt it? Procedia Manufacturing, 3(Ahfe):777–784, 2015. ISSN 23519789. doi: 10.1016/j.promfg.2015.07.329. URL
http://dx.doi.org/10.1016/j.promfg.2015.07.329.
V. Ghilas, J.-F. Cordeau, and E. Demir. The Pickup and Delivery Problem with Time Windows and Scheduled Lines: Models and
algorithms. Transportation Science, pages 1–30, 2016a. ISSN 19160615 03155986. doi: 10.1080/03155986.2016.1166793.
V. Ghilas, E. Demir, and T. Van Woensel. An adaptive large neighborhood search heuristic for the Pickup and Delivery Problem
with Time Windows and Scheduled Lines. Computers and Operations Research, 72:12–30, 2016b. ISSN 03050548. doi:
10.1016/j.cor.2016.01.018. URL http://dx.doi.org/10.1016/j.cor.2016.01.018.
V. Ghilas, E. Demir, and T. V. Woensel. A scenario-based planning for the pickup and delivery problem with time windows,
scheduled lines and stochastic demands. Transportation Research Part B: Methodological, 91:34–51, 2016c. ISSN 01912615.
doi: 10.1016/j.trb.2016.04.015. URL http://dx.doi.org/10.1016/j.trb.2016.04.015.
E. Goetting and W. N. Handover. Crowd-shipping: Is crowd-sourced the secret recipe for delivery in the future? (I), 2016.
J. Gonzalez-Feliu and A. Mercier. A combined people-freight accessibility approach for urban retailing and leisure planning at
strategic level. 5th International Conference on Urban Freight Transport, 2013. doi: 10.1128/CMR.15.3.506. URL https:
//halshs.archives-ouvertes.fr/halshs-00919537.
P. Grangier, M. Gendreau, F. Lehu ´
ed´
e, and L.-M. Rousseau. An adaptive large neighborhood search for the two-echelon multiple-
trip vehicle routing problem with satellite synchronization. European Journal of Operational Research, 254(1):80–91, oct
2016. ISSN 03772217. doi: 10.1016/j.ejor.2016.03.040. URL http://linkinghub.elsevier.com/retrieve/pii/
S0377221716301862.
W. Gruel and J. M. Stanford. Assessing the Long-term Effects of Autonomous Vehicles: A Speculative Approach. Transportation
Research Procedia, 13:18–29, 2016. ISSN 23521465. doi: 10.1016/j.trpro.2016.05.003. URL http://dx.doi.org/10.
1016/j.trpro.2016.05.003.
Q. P. Gu, J. L. Liang, and G. Zhang. Algorithmic analysis for ridesharing of personal vehicles. Lecture Notes in Computer Science
(including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 10043 LNCS(Cocoa 2016):
438–452, 2016. ISSN 16113349. doi: 10.1007/978- 3-319-48749- 6 32. URL http://dx.doi.org/10.1016/j.tcs.2017.
08.019.
F. Guerriero, M. E. Bruni, and F. Greco. A Hybrid Greedy Randomized Adaptive Search Heuritic to Solve the Dial-a-Ride Problem.
Asia-Pacific Journal of Operational Research, 30(01):1250046, feb 2013. ISSN 0217-5959. doi: 10.1142/S0217595912500467.
URL http://www.worldscientific.com/doi/abs/10.1142/S0217595912500467.
C. D. Harper, C. T. Hendrickson, S. Mangones, and C. Samaras. Estimating potential increases in travel with autonomous vehicles
for the non-driving, elderly and people with travel-restrictive medical conditions. Transportation Research Part C: Emerging
Technologies, 72:1–9, 2016. ISSN 0968090X. doi: 10.1016/j.trc.2016.09.003. URL http://dx.doi.org/10.1016/j.trc.
2016.09.003.
G. Heilporn, J.-F. Cordeau, and G. Laporte. An integer L-shaped algorithm for the Dial-a-Ride Problem with stochastic cus-
tomer delays. Discrete Applied Mathematics, 159(9):883–895, jun 2011. ISSN 0166218X. doi: 10.1016/j.dam.2011.01.
021. URL http://dx.doi.org/10.1016/j.dam.2011.01.021http://linkinghub.elsevier.com/retrieve/pii/
S0166218X11000473.
W. Herbawi and M. Weber. Evolutionary multiobjective route planning in dynamic multi-hop ridesharing. Lecture Notes in
Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 6622
LNCS:84–95, 2011. ISSN 03029743. doi: 10.1007/978-3-642- 20364-0 8.
W. Herbawi and M. Weber. The ridematching problem with time windows in dynamic ridesharing: A model and a genetic
algorithm. In 2012 IEEE Congress on Evolutionary Computation, pages 1–8. IEEE, jun 2012. ISBN 978-1-4673-1509-8. doi:
10.1109/CEC.2012.6253001. URL http://ieeexplore.ieee.org/document/6253001/.
A. Hevelke and J. Nida-R¨
umelin. Responsibility for Crashes of Autonomous Vehicles: An Ethical Analysis. Science and Engi-
32
neering Ethics, 21(3):619–630, 2015. ISSN 14715546. doi: 10.1007/s11948-014-9565- 5.
G. Hiermann, R. F. Hartl, J. Puchinger, and T. Vidal. Routing a mix of conventional, plug-in hybrid, and electric vehicles. European
Journal of Operational Research, 272(1):235–248, jan 2019. ISSN 03772217. doi: 10.1016/j.ejor.2018.06.025. URL https://
doi.org/10.1016/j.ejor.2018.06.025https://linkinghub.elsevier.com/retrieve/pii/S0377221718305605.
S. C. Ho, W. Szeto, Y.-H. Kuo, J. M. Leung, M. Petering, and T. W. Tou. A survey of dial-a-ride problems: Literature re-
view and recent developments. Transportation Research Part B: Methodological, 111:395–421, may 2018. ISSN 01912615.
doi: 10.1016/j.trb.2018.02.001. URL https://doi.org/10.1016/j.trb.2018.02.001http://linkinghub.elsevier.
com/retrieve/pii/S0191261517304484.
H. Hojabri, M. Gendreau, J.-Y. Potvin, and L.-M. Rousseau. Large neighborhood search with constraint programming for
a vehicle routing problem with synchronization constraints. Computers &Operations Research, 92:87–97, apr 2018.
ISSN 03050548. doi: 10.1016/j.cor.2017.11.011. URL https://doi.org/10.1016/j.cor.2017.11.011https://
linkinghub.elsevier.com/retrieve/pii/S0305054817302885.
H. Hosni, J. Naoum-Sawaya, and H. Artail. The shared-taxi problem: Formulation and solution methods. Transportation Research
Part B: Methodological, 70:303–318, 2014. ISSN 01912615. doi: 10.1016/j.trb.2014.09.011. URL http://dx.doi.org/10.
1016/j.trb.2014.09.011.
Z.-H. Hu and C. Wei. Synchronizing vehicles for multi-vehicle and one-cargo transportation. Computers &Industrial Engineering,
119(November 2017):36–49, may 2018. ISSN 03608352. doi: 10.1016/j.cie.2018.03.023. URL https://doi.org/10.1016/
j.cie.2018.03.023http://linkinghub.elsevier.com/retrieve/pii/S0360835218301086.
Y. Huang, R. Jin, F. Bastani, and X. Wang. Large Scale Real-time Ridesharing with Service Guarantee on Road Networks. arXiv
preprint arXiv:1302.6666, 7(14):2017–2028, 2013. ISSN 21508097. URL http://arxiv.org/abs/1302.6666.
M. F. Hyland and H. S. Mahmassani. Taxonomy of Shared Autonomous Vehicle Fleet Management Problems to Inform Future
Transportation Mobility. Transportation Research Record: Journal of the Transportation Research Board, 2653(2653):26–34,
2017. ISSN 0361-1981. doi: 10.3141/2653-04. URL http://trrjournalonline.trb.org/doi/10.3141/2653-04.
S. Jain and P. V. Hentenryck. Large neighborhood search for dial-a-ride problems. Lecture Notes in Computer Science (including
subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 6876 LNCS(m):400–413, 2011. ISSN
03029743. doi: 10.1007/978-3-642- 23786-7 31.
W. Jiang, C. R. Dominguez, P. Zhang, M. Shen, and L. Zhang. Large-Scale Nationwide Ridesharing System: A Case Study of
Chunyun. International Journal of Transportation Science and Technology, (October), 2017. ISSN 20460430. doi: 10.1016/j.
ijtst.2017.10.002. URL http://linkinghub.elsevier.com/retrieve/pii/S2046043017300552.
J. Jung, R. Jayakrishnan, and J. Y. Park. Dynamic Shared-Taxi Dispatch Algorithm with Hybrid-Simulated Annealing. Computer-
Aided Civil and Infrastructure Engineering, 31(4):275–291, 2016. ISSN 14678667. doi: 10.1111/mice.12157.
L. Kaan and E. V. Olinick. The Vanpool Assignment Problem: Optimization models and solution algorithms. Computers and
Industrial Engineering, 66(1):24–40, 2013. ISSN 03608352. doi: 10.1016/j.cie.2013.05.020. URL http://dx.doi.org/10.
1016/j.cie.2013.05.020.
N. Kafle, B. Zou, and J. Lin. Design and modeling of a crowdsource-enabled system for urban parcel relay and delivery. Trans-
portation Research Part B: Methodological, 99:62–82, may 2017. ISSN 01912615. doi: 10.1016/j.trb.2016.12.022. URL
http://linkinghub.elsevier.com/retrieve/pii/S019126151630265X.
P. Kalczynski and M. Miklas-Kalczynska. A decentralized solution to the car pooling problem. International Journal of Sustainable
Transportation, 8318:1–12, mar 2018. ISSN 1556-8318. doi: 10.1080/15568318.2018.1440674. URL https://doi.org/
10.1080/15568318.2018.1440674https://www.tandfonline.com/doi/full/10.1080/15568318.2018.1440674.
C. Katrakazas, M. Quddus, W.-H. Chen, and L. Deka. Real-time motion planning methods for autonomous on-road driv-
ing: State-of-the-art and future research directions. Transportation Research Part C: Emerging Technologies, 60:416–442,
2015. ISSN 0968090X. doi: 10.1016/j.trc.2015.09.011. URL http://linkinghub.elsevier.com/retrieve/pii/
S0968090X15003447.
D. Kirchler and R. Wolfler Calvo. A Granular Tabu Search algorithm for the Dial-a-Ride Problem. Transportation Research Part
B: Methodological, 56:120–135, oct 2013. ISSN 01912615. doi: 10.1016/j.trb.2013.07.014. URL http://dx.doi.org/10.
1016/j.trb.2013.07.014http://linkinghub.elsevier.com/retrieve/pii/S019126151300129X.
C¸ . Koc¸ and G. Laporte. Vehicle routing with backhauls: Review and research perspectives. Computers and Operations Research,
91:79–91, 2018. ISSN 03050548. doi: 10.1016/j.cor.2017.11.003.
R. Krueger, T. H. Rashidi, and J. M. Rose. Preferences for shared autonomous vehicles. Transportation Research Part C: Emerging
Technologies, 69:343–355, 2016. ISSN 0968090X. doi: 10.1016/j.trc.2016.06.015. URL http://dx.doi.org/10.1016/j.
trc.2016.06.015.
M. K¨
ummel, F. Busch, and D. Z. Wang. Framework for automated taxi operation: The family model. Transportation Research Pro-
cedia, 22(2016):529–540, 2017. ISSN 23521465. doi: 10.1016/j.trpro.2017.03.071. URL http://linkinghub.elsevier.
com/retrieve/pii/S2352146517302089.
S. Le Vine, A. Zolfaghari, and J. Polak. Autonomous cars: The tension between occupant experience and intersection capacity.
33
Transportation Research Part C: Emerging Technologies, 52:1–14, 2015. ISSN 0968090X. doi: 10.1016/j.trc.2015.01.002.
URL http://dx.doi.org/10.1016/j.trc.2015.01.002.
A. Lee and M. Savelsbergh. Dynamic ridesharing: Is there a role for dedicated drivers? Transportation Research Part B:
Methodological, 81:483–497, 2015. ISSN 01912615. doi: 10.1016/j.trb.2015.02.013. URL http://dx.doi.org/10.1016/
j.trb.2015.02.013.
F. Lehu´
ed´
e, R. Masson, S. N. Parragh, O. P´
eton, and F. Tricoire. A multi-criteria large neighbourhood search for the transportation
of disabled people. Journal of the Operational Research Society, 65(7):983–1000, jul 2014. ISSN 0160-5682. doi: 10.1057/
jors.2013.17. URL https://www.tandfonline.com/doi/full/10.1057/jors.2013.17.
M. W. Levin and S. D. Boyles. Effects of Autonomous Vehicle Ownership on Trip, Mode, and Route Choice. Transportation
Research Record: Journal of the Transportation Research Board, 2493:29–38, 2015. ISSN 0361-1981. doi: 10.3141/2493-04.
URL http://trrjournalonline.trb.org/doi/10.3141/2493-04.
M. W. Levin, T. Li, S. D. Boyles, and K. M. Kockelman. A general framework for modeling shared autonomous vehicles with
dynamic network-loading and dynamic ride-sharing application. 95th Annual Meeting of the Transportation Research Board,
64(January):1–23, 2016. ISSN 01989715. doi: 10.1016/j.compenvurbsys.2017.04.006. URL http://dx.doi.org/10.1016/
j.compenvurbsys.2017.04.006.
B. Li, D. Krushinsky, H. A. Reijers, and T. Van Woensel. The Share-A-Ride Problem: People and parcels sharing taxis. European
Journal of Operational Research, 238(1):31–40, 2014. ISSN 03772217. doi: 10.1016/j.ejor.2014.03.003. URL http://dx.
doi.org/10.1016/j.ejor.2014.03.003.
B. Li, D. Krushinsky, T. Van Woensel, and H. A. Reijers. An adaptive large neighborhood search heuristic for the share-a-ride
problem. Computers and Operations Research, 66:170–180, 2016a. ISSN 03050548. doi: 10.1016/j.cor.2015.08.008. URL
http://dx.doi.org/10.1016/j.cor.2015.08.008.
B. Li, D. Krushinsky, T. Van Woensel, and H. A. Reijers. The Share-a-Ride problem with stochastic travel times and stochastic
delivery locations. Transportation Research Part C: Emerging Technologies, 67:95–108, 2016b. ISSN 0968090X. doi: 10.
1016/j.trc.2016.01.014. URL http://dx.doi.org/10.1016/j.trc.2016.01.014.
A. Lim, Z. Zhang, and H. Qin. Pickup and Delivery Service with Manpower Planning in Hong Kong Public Hospitals. Trans-
portation Science, 51(2):688–705, may 2017. ISSN 0041-1655. doi: 10.1287/trsc.2015.0611. URL http://pubsonline.
informs.org/doi/10.1287/trsc.2015.0611.
M. P. Linares, J. Barcel´
o, C. Carmona, and L. Montero. Analysis and Operational Challenges of Dynamic Ride Sharing Demand
Responsive Transportation Models. Transportation Research Procedia, 21:110–129, 2017. ISSN 23521465. doi: 10.1016/j.
trpro.2017.03.082. URL http://dx.doi.org/10.1016/j.trpro.2017.03.082.
M. Liu, Z. Luo, and A. Lim. A branch-and-cut algorithm for a realistic dial-a-ride problem. Transportation Research Part B:
Methodological, 81:267–288, 2014. ISSN 01912615. doi: 10.1016/j.trb.2015.05.009. URL http://dx.doi.org/10.1016/
j.trb.2015.05.009.
Y. Liu and Y. Li. Pricing scheme design of ridesharing program in morning commute problem. Transportation Research Part C:
Emerging Technologies, 79:156–177, 2017. ISSN 0968090X. doi: 10.1016/j.trc.2017.02.020. URL http://dx.doi.org/
10.1016/j.trc.2017.02.020.
M. Lokhandwala and H. Cai. Dynamic ride sharing using traditional taxis and shared autonomous taxis: A case study of NYC.
Transportation Research Part C: Emerging Technologies, 97(September):45–60, 2018. ISSN 0968090X. doi: 10.1016/j.trc.
2018.10.007. URL https://linkinghub.elsevier.com/retrieve/pii/S0968090X18307551.
J. Ma, X. Li, F. Zhou, and W. Hao. Designing optimal autonomous vehicle sharing and reservation systems: A linear programming
approach. Transportation Research Part C: Emerging Technologies, 84:124–141, 2017. ISSN 0968090X. doi: 10.1016/j.trc.
2017.08.022. URL http://dx.doi.org/10.1016/j.trc.2017.08.022.
R. Ma and H. M. Zhang. The morning commute problem with ridesharing and dynamic parking charges. Transportation Research
Part B: Methodological, 106:345–374, 2017. ISSN 01912615. doi: 10.1016/j.trb.2017.07.002. URL https://doi.org/10.
1016/j.trb.2017.07.002.
S. Ma, Y. Zheng, and O. Wolfson. Real-time city-scale taxi ridesharing. 27(7):1782–1795, 2015. ISSN 1041-4347. doi: 10.1109/
TKDE.2014.2334313. URL http://ieeexplore.ieee.org/xpls/abs{_}all.jsp?arnumber=6847170.
M. Maciejewski, J. M. Salanova, J. Bischoff, and M. Estrada. Large-scale microscopic simulation of taxi services. Berlin and
Barcelona case studies. Journal of Ambient Intelligence and Humanized Computing, 7(3):385–393, 2016. ISSN 18685145. doi:
10.1007/s12652-016-0366-3.
L. M. Martinez. An Agent-based model to assess the impacts of introducing a shared-taxi system in Lisbon ( Portugal ). Journal
of Advanced Transportation, 49(3):475–495, 2015. ISSN 01976729. doi: 10.1002/atr.1283.
M. A. Masmoudi, M. Hosny, K. Braekers, and A. Dammak. Three effective metaheuristics to solve the multi-depot multi-trip
heterogeneous dial-a-ride problem. Transportation Research Part E: Logistics and Transportation Review, 96:60–80, 2016.
ISSN 13665545. doi: 10.1016/j.tre.2016.10.002.
M. A. Masmoudi, K. Braekers, M. Masmoudi, and A. Dammak. A hybrid Genetic Algorithm for the Heterogeneous Dial-A-
34
Ride Problem. Computers &Operations Research, 81:1–13, 2017. ISSN 03050548. doi: 10.1016/j.cor.2016.12.008. URL
http://linkinghub.elsevier.com/retrieve/pii/S0305054816303070.
N. Masoud, R. Lloret-Batlle, and R. Jayakrishnan. Using bilateral trading to increase ridership and user permanence in ridesharing
systems. Transportation Research Part E: Logistics and Transportation Review, 102:60–77, 2017. ISSN 13665545. doi:
10.1016/j.tre.2017.04.007. URL http://dx.doi.org/10.1016/j.tre.2017.04.007.
R. Masson, F. Lehu´
ed´
e, and O. P´
eton. The dial-a-ride problem with transfers. Computers and Operations Research, 41(1):12–23,
2014. ISSN 03050548. doi: 10.1016/j.cor.2013.07.020. URL http://dx.doi.org/10.1016/j.cor.2013.07.020.
R. Masson, A. Trentini, F. Lehu´
ed´
e, N. Malh´
en´
e, O. P´
eton, and H. Tlahig. Optimization of a city logistics transportation system
with mixed passengers and goods. EURO Journal on Transportation and Logistics, 6(1):81–109, mar 2017. ISSN 2192-4376.
doi: 10.1007/s13676-015-0085-5. URL http://link.springer.com/10.1007/s13676-015-0085-5.
J. Meyer, H. Becker, P. M. Bosch, and K. W. Axhausen. Autonomous vehicles: The next jump in accessibilities? Research in
Transportation Economics, pages 1–12, 2017. ISSN 07398859. doi: 10.1016/j.retrec.2017.03.005.
D. Milakis, M. Snelder, B. Van Arem, B. Van Wee, and G. H. De Almeida Correia. Development and transport implications
of automated vehicles in the Netherlands: Scenarios for 2030 and 2050. European Journal of Transport and Infrastructure
Research, 17(1):63–85, 2017. ISSN 15677141.
A. Mladenow, C. Bauer, and C. Strauss. Crowdsourcing in logistics. In Proceedings of the 17th International Conference on
Information Integration and Web-based Applications &Services - iiWAS ’15, number August 2016, pages 1–8, New York,
New York, USA, 2015. ACM Press. ISBN 9781450334914. doi: 10.1145/2837185.2837242. URL http://dl.acm.org/
citation.cfm?doid=2837185.2837242.
Y. Molenbruch, K. Braekers, and A. Caris. Typology and literature review for dial-a-ride problems. Annals of Operations Research,
259(1-2):295–325, dec 2017. ISSN 0254-5330. doi: 10.1007/s10479-017-2525-0. URL http://link.springer.com/10.
1007/s10479-017-2525-0.
C. C. Murray and A. G. Chu. The flying sidekick traveling salesman problem: Optimization of drone-assisted parcel delivery.
Transportation Research Part C: Emerging Technologies, 54:86–109, 2015. ISSN 0968090X. doi: 10.1016/j.trc.2015.03.005.
URL http://dx.doi.org/10.1016/j.trc.2015.03.005.
A. Otto, N. Agatz, J. Campbell, B. Golden, and E. Pesch. Optimization approaches for civil applications of unmanned aerial
vehicles (UAVs) or aerial drones: A survey. Networks, 72(4):411–458, dec 2018. ISSN 00283045. doi: 10.1002/net.21818.
URL http://doi.wiley.com/10.1002/net.21818.
O. Ozturk and J. Patrick. An optimization model for freight transport using urban rail transit. European Journal of Operational
Research, 267(3):1110–1121, 2017. ISSN 03772217. doi: 10.1016/j.ejor.2017.12.010. URL https://doi.org/10.1016/j.
ejor.2017.12.010.
J. Paquette, J.-F. Cordeau, G. Laporte, and M. M. Pascoal. Combining multicriteria analysis and tabu search for dial-a-ride prob-
lems. Transportation Research Part B: Methodological, 52:1–16, jun 2013. ISSN 01912615. doi: 10.1016/j.trb.2013.02.
007. URL http://dx.doi.org/10.1016/j.trb.2013.02.007http://linkinghub.elsevier.com/retrieve/pii/
S0191261513000337.
S. N. Parragh and V. Schmid. Hybrid column generation and large neighborhood search for the dial-a-ride problem. Computers
and Operations Research, 40(1):490–497, 2013. ISSN 03050548. doi: 10.1016/j.cor.2012.08.004. URL http://dx.doi.
org/10.1016/j.cor.2012.08.004.
S. N. Parragh, J. Pinho de Sousa, and B. Almada-Lobo. The Dial-a-Ride Problem with Split Requests and Profits. Transportation
Science, 49(2):311–334, may 2015. ISSN 0041-1655. doi: 10.1287/trsc.2014.0520. URL http://pubsonline.informs.
org/doi/10.1287/trsc.2014.0520.
V. Pimenta, A. Quilliot, H. Toussaint, and D. Vigo. Models and algorithms for reliability-oriented Dial-a-Ride with autonomous
electric vehicles. European Journal of Operational Research, 257(2):601–613, mar 2017. ISSN 03772217. doi: 10.1016/j.ejor.
2016.07.037. URL http://linkinghub.elsevier.com/retrieve/pii/S0377221716305744.
D. Reyes, M. Savelsbergh, and A. Toriello. Vehicle Routing with Roaming Delivery Locations. Transportation Research Part C:
Emerging Technologies, 80:71–91, 2017. doi: 10.1016/j.trc.2017.04.003. URL https://doi.org/10.1016/j.trc.2017.
04.003.
U. Ritzinger, J. Puchinger, and R. F. Hartl. A survey on dynamic and stochastic vehicle routing problems. International Journal of
Production Research, 54(1):215–231, 2016. ISSN 1366588X. doi: 10.1080/00207543.2015.1043403. URL http://dx.doi.
org/10.1080/00207543.2015.1043403.
A. H. Sampaio Oliveira, M. W. P. Savelsbergh, L. Veelenturf, and T. van Woensel. Crowd-based City Logis-
tics. SCL Report Series Georgia Institute of Technology, pages 1–14, 2017. URL http://repository.tue.nl/
46281f7c-72b1-4de8-848f- 97ae81f99de9{%}0Ahttp://www.optimization-online.org/DB{_}FILE/2017/11/
6346.pdf.
D. Sanchez, S. Martinez, and J. Domingo-Ferrer. Co-utile P2P ridesharing via decentralization and reputation management.
Transportation Research Part C: Emerging Technologies, 73:147–166, 2016. ISSN 0968090X. doi: 10.1016/j.trc.2016.10.017.
35
D. O. Santos and E. C. Xavier. Taxi and ride sharing: A dynamic dial-a-ride problem with money as an incentive. Expert Systems
with Applications, 42(19):6728–6737, 2015. ISSN 09574174. doi: 10.1016/j.eswa.2015.04.060.
M. Savelsbergh and T. Van Woensel. 50th Anniversary Invited Article—City Logistics: Challenges and Opportunities. Trans-
portation Science, 50(2):579–590, may 2016. ISSN 0041-1655. doi: 10.1287/trsc.2016.0675. URL http://pubsonline.
informs.org/doi/10.1287/trsc.2016.0675.
A. Scheltes and G. H. de Almeida Correia. Exploring the use of automated vehicles as last mile connection of train trips through
an agent-based simulation model: An application to Delft, Netherlands. International Journal of Transportation Science and
Technology, 6(1):28–41, 2017. ISSN 20460430. doi: 10.1016/j.ijtst.2017.05.004. URL http://linkinghub.elsevier.
com/retrieve/pii/S2046043016300296.
M. Schilde, K. Doerner, and R. Hartl. Integrating stochastic time-dependent travel speed in solution methods for the dy-
namic dial-a-ride problem. European Journal of Operational Research, 238(1):18–30, oct 2014. ISSN 03772217. doi:
10.1016/j.ejor.2014.03.005. URL http://dx.doi.org/10.1016/j.ejor.2014.03.005http://linkinghub.elsevier.
com/retrieve/pii/S0377221714002197.
S. A. Shaheen, N. D. Chan, and T. Gaynor. Casual carpooling in the San Francisco Bay Area: Understanding user characteristics,
behaviors, and motivations. Transport Policy, 51:165–173, 2016. ISSN 1879310X. doi: 10.1016/j.tranpol.2016.01.003. URL
http://dx.doi.org/10.1016/j.tranpol.2016.01.003.
M. D. Simoni, P. Bujanovic, S. D. Boyles, and E. Kutanoglu. Urban consolidation solutions for parcel delivery considering location,
fleet and route choice. Case Studies on Transport Policy, 6(1):112–124, 2018. ISSN 22136258. doi: 10.1016/j.cstp.2017.11.002.
URL https://doi.org/10.1016/j.cstp.2017.11.002.
P. M. Smolnicki and J. Sołtys. Driverless Mobility: The Impact on Metropolitan Spatial Structures. Procedia Engineering, 161:
2184–2190, 2016. ISSN 18777058. doi: 10.1016/j.proeng.2016.08.813. URL http://dx.doi.org/10.1016/j.proeng.
2016.08.813.
R. Sparrow and M. Howard. When human beings are like drunk robots: Driverless vehicles, ethics, and the future of transport.
Transportation Research Part C: Emerging Technologies, 80:206–215, 2017. ISSN 0968090X. doi: 10.1016/j.trc.2017.04.014.
URL http://dx.doi.org/10.1016/j.trc.2017.04.014.
M. Steadieseifi, N. P. Dellaert, W. Nuijten, T. Van Woensel, and R. Raoufi. Multimodal freight transportation planning: A literature
review. European Journal of Operational Research, 233(1):1–15, 2014. ISSN 03772217. doi: 10.1016/j.ejor.2013.06.055.
URL http://dx.doi.org/10.1016/j.ejor.2013.06.055.
M. Stiglic, N. Agatz, M. Savelsbergh, and M. Gradisar. The benefits of meeting points in ride-sharing systems. Transportation
Research Part B: Methodological, 82:36–53, 2015. ISSN 01912615. doi: 10.1016/j.trb.2015.07.025. URL http://dx.doi.
org/10.1016/j.trb.2015.07.025.
M. Stiglic, N. Agatz, M. Savelsbergh, and M. Gradisar. Enhancing Urban Mobility : Integrating Ride-sharing and Public Transit.
pages 1–20, 2016a.
M. Stiglic, N. Agatz, M. Savelsbergh, and M. Gradisar. Making dynamic ride-sharing work: The impact of driver and rider
flexibility. Transportation Research Part E: Logistics and Transportation Review, 91:190–207, 2016b. ISSN 13665545. doi:
10.1016/j.tre.2016.04.010. URL http://dx.doi.org/10.1016/j.tre.2016.04.010.
M. Stiglic, N. Agatz, M. Savelsbergh, and M. Gradisar. Enhancing urban mobility: Integrating ride-sharing and public transit.
Computers and Operations Research, 90:12–21, 2018. ISSN 03050548. doi: 10.1016/j.cor.2017.08.016. URL http://dx.
doi.org/10.1016/j.cor.2017.08.016.
R. Tachet, O. Sagarra, P. Santi, G. Resta, M. Szell, S. H. Strogatz, and C. Ratti. Scaling Law of Urban Ride Sharing. Nature
Publishing Group, pages 1–6, 2017. ISSN 2045-2322. doi: 10.1038/srep42868. URL http://dx.doi.org/10.1038/
srep42868.
A. Talebpour and H. S. Mahmassani. Influence of connected and autonomous vehicles on traffic flow stability and throughput.
Transportation Research Part C: Emerging Technologies, 71:143–163, 2016. ISSN 0968090X. doi: 10.1016/j.trc.2016.07.007.
URL http://dx.doi.org/10.1016/j.trc.2016.07.007.
E. Taniguchi, R. G. Thompson, and T. Yamada. New Opportunities and Challenges for City Logistics. Transportation Research
Procedia, 12(June 2015):5–13, 2016. ISSN 23521465. doi: 10.1016/j.trpro.2016.02.004. URL http://www.sciencedirect.
com/science/article/pii/S2352146516000053.
A. Trentini, R. Masson, F. Lehu´
ed´
e, N. Malh´
en´
e, O. P´
eton, and H. Tlahig. A shared ” passengers & goods ” city logistics system.
4th International Conference on Information Systems, 2013.
A. Trentini, R. Masson, F. Lehu´
ed´
e, and N. Malh´
en´
e. Optimization of a city logistics transportation system with mixed passengers
and goods. EURO J Transp Logist, pages 1–23, 2015. ISSN 2192-4376. doi: 10.1007/s13676- 015-0085-5. URL http:
//hal.archives-ouvertes.fr/hal-00859078/.
Z. Wadud. Fully automated vehicles: A cost of ownership analysis to inform early adoption. Transportation Research Part A:
Policy and Practice, 101:163–176, 2017. ISSN 09658564. doi: 10.1016/j.tra.2017.05.005. URL http://dx.doi.org/10.
1016/j.tra.2017.05.005.
36
X. Wang, N. Agatz, and A. Erera. Stable Matching for Dynamic Ride-Sharing Systems. Transportation Science, 52(4):850–
867, aug 2018. ISSN 0041-1655. doi: 10.1287/trsc.2017.0768. URL http://www.ssrn.com/abstract=2611353http:
//pubsonline.informs.org/doi/10.1287/trsc.2017.0768.
Y. Wang, D. Zhang, Q. Liu, F. Shen, and L. H. Lee. Towards enhancing the last-mile delivery: An effective crowd-tasking
model with scalable solutions. Transportation Research Part E: Logistics and Transportation Review, 93:279–293, sep
2016. ISSN 13665545. doi: 10.1016/j.tre.2016.06.002. URL http://dx.doi.org/10.1016/j.tre.2016.06.002https:
//linkinghub.elsevier.com/retrieve/pii/S1366554516300783.
Y. Wang, S. Winter, and N. Ronald. How much is trust: The cost and benefit of ridesharing with friends. Computers, Environment
and Urban Systems, 65:103–112, 2017. ISSN 01989715. doi: 10.1016/j.compenvurbsys.2017.06.002. URL http://dx.doi.
org/10.1016/j.compenvurbsys.2017.06.002.
R. Wolfler Calvo, F. de Luigi, P. Haastrup, and V. Maniezzo. A distributed geographic information system for the daily car pooling
problem. Computers and Operations Research, 31(13):2263–2278, 2004. ISSN 03050548. doi: 10.1016/S0305-0548(03)
00186-2.
Z. Xiang, C. Chu, and H. Chen. A fast heuristic for solving a large-scale static dial-a-ride problem under complex constraints.
European Journal of Operational Research, 174(2):1117–1139, 2006. ISSN 03772217. doi: 10.1016/j.ejor.2004.09.060.
M. D. Yap, G. Correia, and B. van Arem. Preferences of travellers for using automated vehicles as last mile public transport of
multimodal train trips. Transportation Research Part A: Policy and Practice, 94:1–16, dec 2016. ISSN 09658564. doi: 10.
1016/j.tra.2016.09.003. URL http://dx.doi.org/10.1016/j.tra.2016.09.003http://linkinghub.elsevier.com/
retrieve/pii/S0965856416307765.
B. Yu, Y. Ma, M. Xue, B. Tang, B. Wang, J. Yan, and Y. M. Wei. Environmental benefits from ridesharing: A case of Beijing.
Applied Energy, 191:141–152, 2017. ISSN 03062619. doi: 10.1016/j.apenergy.2017.01.052.
W. Zhang, S. Guhathakurta, J. Fang, and G. Zhang. Exploring the impact of shared autonomous vehicles on urban parking demand:
An agent-based simulation approach. Sustainable Cities and Society, 19:34–45, 2015. ISSN 22106707. doi: 10.1016/j.scs.
2015.07.006. URL http://dx.doi.org/10.1016/j.scs.2015.07.006.
J. P. Zmud and I. N. Sener. Towards an Understanding of the Travel Behavior Impact of Autonomous Vehicles. Transportation
Research Procedia, 25:2504–2523, 2017. ISSN 23521465. doi: 10.1016/j.trpro.2017.05.281. URL http://dx.doi.org/10.
1016/j.trpro.2017.05.281.
37
