Integrated people-and-goods transportation systems: from a literature review to a general framework for future research

Abstract

The promotion of urban mobility by integrating people-and-goods transportation has attracted increasing attention in recent years. Within this framework, diversified forms such as co-modality, freight on transit, and crowdshipping have been proposed, piloted or implemented. The success of the implementation and market penetration depends on not only the novelties of the concept but also the planning and operational efficiency. Thus, a comprehensive review focusing on the operation of integrated people-and-goods transportation systems and associated critical decisions and subproblems is performed. Different practical forms in which people and goods are transported in an integrated manner are identified. The critical decisions associated with each form and subproblem are discussed, along with corresponding models and solution approaches. Notably, because integrated transportation systems are in the early exploration stage at present, new forms are expected to emerge. Therefore, this paper proposes a general framework to realise the planning and operation of new forms in the future. The decisions and subproblems identified from existing forms are fed to the proposed general framework to identify two key research opportunities: to improve or extend existing research and to conduct pioneering research to fill the gaps in the frameworks for operating potential forms of integrated people-and-goods transportation.

Full text

1
Integrated people-and-goods transportation systems: from a literature review to a
1
general framework for future research
2
Rong Chenga, Yu Jianga*, Otto Anker Nielsena
3
a DTU Management, Department of Technology, Management, and Economics, Technical University
4
of Denmark, 2800 Kgs. Lyngby, Denmark
5
Corresponding author:
6
Yu Jiang yujiang@dtu.dk,
7
Bygningstorvet, 116, 110A 2800 Kgs. Lyngby, Denmark
8
9
10
11
12
13
14
15
16
17
18
19

2
Integrated people-and-goods transportation systems: from a literature review to a
1
general framework for future research
2
Abstract:
3
The promotion of urban mobility by integrating people-and-goods transportation has attracted
4
increasing attention in recent years. Within this framework, diversified forms such as co-
5
modality, freight on transit, and crowdshipping have been proposed, piloted or implemented. The
6
success of the implementation and market penetration depends on not only the novelties of the
7
concept but also the planning and operational efficiency. Thus, a comprehensive review focusing
8
on the operation of integrated people-and-goods transportation systems and associated critical
9
decisions and subproblems is performed. Different practical forms in which people and goods
10
are transported in an integrated manner are identified. The critical decisions associated with each
11
form and subproblem are discussed, along with corresponding models and solution approaches.
12
Notably, because integrated transportation systems are in the early exploration stage at present,
13
new forms are expected to emerge. Therefore, this paper proposes a general framework to realise
14
the planning and operation of new forms in the future. The decisions and subproblems identified
15
from existing forms are fed to the proposed general framework to identify two key research
16
opportunities: to improve or extend existing research and to conduct pioneering research to fill
17
the gaps in the frameworks for operating potential forms of integrated people-and-goods
18
transportation.
19
Keywords: Integrated people-and-goods transportation; shared mobility; share-a-ride; freight on
20
transit; crowdshipping
21
22
23

3
1 Introduction
1
Urban mobility faces increasing challenges with population growth, urbanisation, e-commerce and
2
varying land-use patterns. Many daily tasks require transporting people or goods. While transport
3
services enhance the convenience of daily life, they also have adverse effects, such as greenhouse gas
4
emissions, local air pollution, traffic accidents, and congestion (European Commission, 2019). These
5
negative externalities can be mitigated by establishing shared and integrated transportation systems
6
(Mourad et al., 2019). Although moving people and goods together has been successfully implemented
7
in long-haul transportation modes such as aircraft and ferries, passenger and goods movements in
8
urban transport systems are typically planned and performed separately. Since the transportation of
9
people and goods is mutually affected by sharing and competition for road space and infrastructures,
10
a separate implementation may underutilise the existing infrastructure and vehicle capacity. Thus, a
11
promising solution, integrating people-and-goods transportation systems, has attracted increasing
12
attention in recent years.
13
The idea of transporting people and goods together in an urban transportation context was
14
highlighted by the European Commission, stating that “Local authorities need to consider all urban
15
logistics related to passenger and freight transport together as a single logistics system” (European
16
Commission, 2007). Since then, several researchers have focused on integrated people-and-goods
17
transportation systems (hereinafter referred to as integrated transportation systems). Diverse novel
18
terms such as co-modality, freight on transit (FOT), crowdshipping, cohabitation of passengers and
19
goods, and cargo hitching have been proposed, coupled with various methodological developments.
20

4
There have been several remarkable reviews focusing on particular forms of integrated
1
transportation systems, like crowdshipping and FOT (Alnaggar et al., 2021; Le et al., 2019; Elber &
2
Rentschler, 2021), discussing it within broader topics such as shared mobility (Mourad et al., 2019),
3
collaborative urban transportation (Cleophas et al., 2019), and city logistics (Savelsbergh & Van
4
Woensel, 2016), or conducting bibliometric analysis (Cavallaro & Nocera, 2022). This paper offers a
5
systematic review to complement existing reviews with the following objectives: 1) categorising
6
different forms of integrated transportation systems; 2) identifying the key issues for different forms
7
and discussing corresponding solutions; 3) proposing a general framework to describe the operation
8
of integrated transportation systems; and 4) giving recommendations for future development and
9
research.
10
The remaining paper is organised as follows. Section 2 describes different forms of integrated
11
transportation systems. Section 3 specifies the research problems in existing studies. These problems
12
are incorporated in a general framework proposed in Section 4. Section 5 highlights the research gaps
13
and future research directions.
14
2 Forms of integrated people-and-goods transportation
15
We define an integrated people-and-goods transportation system as a system in which the resources
16
for transporting people and goods are jointly utilised such that people and goods are transported in the
17
same vehicle, either private or public, or share the same infrastructure, such as railways, stations, and
18
platforms. We then categorise three forms: people and goods share-a-ride (SAR), FOT, and
19
crowdshipping. In what follows, we will first introduce the definition and characteristics of each form
20

5
(see Table 1), then comment on real-world applications.
1
2.1 Definition
2
(1) People and goods share-a-ride
3
In this form, a shared vehicle, e.g., a taxi or shared autonomous vehicle (SAV), provides door-
4
to-door service for both passengers and goods. The vehicle with passengers can simultaneously
5
transport small parcels such as mail, documents, and takeaway meals. The primary research problem
6
involved is the routing problem, known as the people and parcel SAR problem (SARP, Li et al., 2014).
7
(2) Freight on transit
8
Elbert & Rentschler (2022) defined FOT as “the integrated and organised transportation of
9
passengers and goods within urban areas using a system of vehicles such as buses and trains that
10
operate at regular times on fixed routes and are used by the public.” We hereby extend this definition
11
by considering 1) emerging flexible public transport, particularly, demand-responsive services such as
12
personal rapid transit and freight rapid transit; 2) urban-suburban and urban-rural transit. Depending
13
on the public transport vehicles used, goods and passengers could share three resources in FOT:
14
carriage, vehicle, and tracks.
15
(3) Crowdshipping
16
Buldeo Rai (2017) defined crowdshipping as “an information connectivity enabled
17
marketplace concept that matches supply and demand for logistics services with an undefined and
18

6
external crowd that has free capacity with regards to time and/or space, participates on a voluntary
1
basis and is compensated accordingly”. Crowdshippers are categorised into dedicated and ad-hoc
2
crowdshippers. Dedicated crowdshippers devoted their available time to perform deliveries using
3
dedicated trips. In contrast, ad-hoc crowdshippers utilise their already planned trips with extra
4
capacities. The two modes of crowdshipping have their own advantages and disadvantages. For
5
example, crowdshipping with dedicated crowdshippers usually provides more efficient crowd logistics
6
than ad-hoc crowdshippers, but it leads to much longer travel distances than crowdshipping with ad-
7
hoc drivers (Buldeo Rai et al., 2018). Both types of crowdshipping are likely to be functional in the
8
future. Nonetheless, in this study, we do not consider dedicated crowdshippers because, although
9
people and goods move simultaneously, they do not integrate peoples’ existing travel demands but
10
induce new ones. Without further specification, the crowdshippers in the rest of this paper refer to as
11
ad-hoc crowdshippers. Primarily, crowdshippers perform crowdsourced delivery through a single
12
transportation mode, e.g., taxis, public transit, or their vehicles. With the emerging concept of mobility
13
as a service (MaaS), He & Csiszár (2021) and Le Pira et al. (2021) proposed utilising multiple
14
transportation modes.
15
2.2 Applications and barriers
16
2.2.1 Applications
17
Table 2 lists the applications of different forms. To the best of the authors’ knowledge, there are no
18
real-world applications of people-and-goods SAR yet. Hence, we only present FOT and
19
crowdshipping.
20

7
(1) FOT
1
(a) Bus-based FOT. This is the most widely implemented FOT system worldwide, e.g.,
2
Bussgods in Sweden, Matkahuolto in Finland, Greyhound Freight in Australia, Maritime Bus in
3
Canada, and Greyhound Package Express US. It usually operates on existing long-distance transit
4
routes connecting regional centres and rural areas. Goods utilise the available space on passenger
5
vehicles, e.g., the luggage compartment or a dedicated goods compartment of a bus. Most above-
6
mentioned systems are still in operation, except Greyhound Package Express US, which ends on
7
September 30, 2022, for concentrating on passenger services.
8
(b) Tram-based FOT. Most of the tram-based FOT are implemented in Europe in the form that
9
dedicated freight trams share tracks with passenger trams connecting urban and suburban areas. Three
10
projects (CarGo Tram of Volkswagen, Cargo-Trams/E-Trams in Zurich, and Recycling Trams in Iasi)
11
succeed under specific conditions. The success of the first one is attributed to its low cost of building
12
additional connection tracks, as the factory is only about three miles away from the logistics centre.
13
The other two are provided as public service and avoid additional infrastructure investment and
14
interference with passenger traffic by carefully selecting stop stations. Three projects are short-lived.
15
City Cargo in Amsterdam was abandoned because it failed to acquire adequate finance for investments
16
in trams, electric last-mile delivery vehicles, new tracks, and distribution centres, and there were
17
conflicting objectives among stakeholders (Arvidsson & Browne, 2013). The other two projects,
18
GüterBim in Vienna and TramFret in Saint-Etienne were discontinued due to a lack of customer
19
interest.
20

8
(c) Train-based FOT. One successful example of the train-based FOT is the Dabbawala food
1
delivery system in Mumbai. It links kitchens in local villages to people working in metropolitan areas
2
using a hub-and-spoke transport system with passenger trains and bicycles.
3
(d) Metro-based FOT. We only found one trial in Sapporo that tested using the metro to
4
transport parcels from the suburbs to the city centre. The trial was successful, but the project ceased
5
due to poor demand and the high cost of retrofitting metro stations to handle goods.
6
(2) Crowdshipping.
7
Alnaggar et al. (2021) reviewed crowdsourced delivery platforms operated by E-retailers (e.g.,
8
Amazon Flex) and couriers (e.g., DHL), among which four are with ad-hoc crowdshippers, i.e., DHL
9
Myways, Hitch, Nimber, and Roadie. Hitch mainly supports local deliveries, while others allow for
10
both local and long-haul deliveries. Notably, public transport-based crowdshipping has emerged in
11
recent years. It allows passengers to bring a parcel from a parcel locker located in a public transport
12
station to another along their ride. A “Crowd ship” trial was conducted in the Greater Copenhagen
13
Area in 2020 to analyse people’s preference for public transport-based crowdshipping. A similar
14
project, “Öffi-Packerl” in Vienna, is expected to make the first test in Vienne in 2024.
15
2.2.2 Barriers
16
We summarise the main challenges for deploying the integrated system from five aspects, policy,
17
economics, society, organisation, and technology.
18
First, passengers and goods transportation are usually regulated by different authorities with
19
separate rules and policies (Bruzzone et al., 2021). Passengers can carry goods on their trip, but taxi
20

9
drivers and privately hired vehicles are forbidden to be couriers if no passenger is on board. This could
1
explain why taxi companies or technology companies (e.g., Uber, Grab, etc.) do not offer integrated
2
people and goods transportation services. However, in practice, there is a grey area where a passenger
3
hires a taxi while the “real passenger” is a package. The good news is that the Land Transport Authority
4
of Singapore is monitoring recent trends to see if these regulations need to be reviewed, and a
5
temporary relaxation of this rule was extended for a third time
1
.
6
Secondly, economic viability is important for the success of a project. Many FOT projects were
7
terminated due to a lack of money or customer interests and conflicting objectives among stakeholders.
8
These issues could be partially avoided by carefully identifying suitable markets and optimising the
9
organisation/operation/revenue allocation of the integrated transport service.
10
Thirdly, people may have psychological barriers. For example, passengers may feel unsafe or
11
reluctant to share a ride with goods; the conflicts between freight operators and passengers at transit
12
stations may increase passengers’ discomfort level; crowdshippers may be concerned about privacy.
13
These problems could be mitigated by regulating the type of goods, proper planning of integrated
14
transportation services, and tightening regulations on privacy and data security.
15
Fourthly, organisational challenges include finding initial capital investment, coordinating
16
various stakeholders and entities, dealing with resistance from passengers, transit agencies, workers
17
from logistics service providers, ensuring the safety of passengers and goods, etc (Cochrane et al.,
18
2017).
19
1
https://tnp.straitstimes.com/news/singapore/cabbies-private-hire-drivers-can-make-deliveries-until-march

10
Lastly, technical challenges include searching underutilised capacity, designing the routes and
1
schedules of shared vehicles, selecting routes for FOT, integrating freight delivery and passenger
2
schedules, coordinating last-mile delivery with FOT, matching crowdshippers with parcels, designing
3
an optimal price for the integrated transportation systems, etc.
4
Overall, the technical challenges are easier to overcome than the challenges on other
5
dimensions. Besides, solving technical challenges could contribute to resolving other challenges. For
6
example, deploying advanced techniques aiming at operating integrated transportation systems cost-
7
effectively could enhance the economic viability of the integrated system, which contributes to
8
attracting investors; well-planned freight hub location, route and schedules of freight vehicles could
9
reduce unnecessary conflicts between passengers and goods or inconvenience to passengers.
10

11
Table 1 Forms of integrated transportation systems in the literature
Integration form
Transportation
means
Shared resource
Goods operators
Practical
application
Vehicles
Infrastructure
Dedicated
workers*
Crowd-
shippers**
Senders and
receivers
Share-a-ride
Taxis
√
√
SAVs
√
√
Freight on transit
Buses
√
√
√
Metros
√
√
√
√
Trains
√
√
√
√
Trams
√
√
√
Personal rapid transit
√
√
√
Drones
√
√
Robots
√
√
Crowdshipping
Private cars/bikes/
cargo bikes
√
√
√
Public transport
√
√
√
*Dedicated workers: Staff at stations or distribution centres are in charge of loading and unloading goods.
**Crowdshippers: Ordinary people assist in picking up and delivering goods.

12
Table 2 Application of integrated people-and-goods transportation
Integration form
Transportation
means
Projects
Status
Comment / Failure reason
Freight on transit
Bus
Bussgods (Sweden)
On going
Matkahuolto (Finland)
On going
Greyhound Freight
(Australia)
On going
Maritime Bus (Canada)
On going
Greyhound Package
Express (US)
Closed in September
2022
Concentrate on passenger services
Metro
Subway-integrated city
logistics system (Japan)
September 2–15, 2010
Lack of money
Train
Dabbawalas (India)
On going
Tram
CarGo Tram of Volkswagen
(Dresden)
November 2000 –
December 2020
End of producing
Cargo-Trams/E-Trams
(Zurich)
On going
Recycling Tram (Iasi)
On going
GuterBim (Vienna)
May 2005 – June 2007
Lack of customer interest
City Cargo (Amsterdam)
March 2007 – April
2007
Lack of money
TramFret (Saint-Etienne)
June 2017 – July 2017
Lack of customer interest
Crowdshipping
Private vehicles
DHL Myways (Stockholm)
September 2013 –
Unknown
Unknown

13
Hitch (US)
Unknown
Unknown
Nimber (London, Athens,
Oslo)
On going
Roadie (US)
On going
Public transport
Crowd ship (Denmark)
September – October
2020
Öffi-Packerl (Austria)
Planning
Sources:
Alnaggar et al., 2021; Arvidsson et al., 2016; Arvidsson and Browne, 2013; Cochrane et al, 2017; Fessler et al., 2022; Kikuta et al., 2012; Qu et
al., 2022;
https://www.railjournal.com/passenger/metros/tokyo-metro-to-test-parcel-operation/;
https://en.wikipedia.org/wiki/CarGoTram;
https://industriemagazin.at/artikel/die-wiener-gueterbim-das-kurze-gastspiel-der-transport-strassenbahn/;
http://www.tautonline.com/zurichs-cargo-tram/;
https://aqtr.com/association/actualites/freight-transit-new-concept-city-logistics;
https://brutkasten.com/oeffi-packerl-entwicklung-startet/.

14
3 Literature review
1
As discussed in the previous section, solving technical problems contributes to mitigating the
2
barriers to implementing integrated transportation systems. This section reviews the technical
3
problems examined in the existing studies for each form listed in Table 1 (see Figure 1).
4
5
6
Figure 1 Technical problems that have been studied for different integration forms
7
3.1 People-and-goods share-a-ride
8
3.1.1 Route planning for share-a-ride vehicles
9
Li et al. (2014) first defined the routing problem for integrated transportation using taxis
10
as SARP. Several regulations were introduced to ensure high-quality services: (R1) Passengers
11

15
must have a maximum ride time. (R2) An upper limit exists on the number of parcels served
1
during one passenger service. (R3) Two passengers cannot be served simultaneously by one
2
taxi. R1 and R2 aim to decrease the influence of parcel delivery on passenger services, and R3
3
aims to ensure personal security and convenience (e.g., gender or smoking preferences). These
4
regulations were gradually relaxed to present a more general SARP, leading to more profits.
5
Beirigo et al. (2018) and Tholen et al. (2021) eliminated R2 and R3. Yu et al. (2018, 2022b)
6
relaxed all these regulations.
7
Many features have been added to the original SARP to accommodate various
8
application scenarios. Yu et al. (2021a) allowed passenger compartments to store parcels,
9
which utilises the total vehicle capacity more flexibly and efficiently. Yu et al. (2022b)
10
extended a single depot to multiple depots, given that it is challenging to serve scattered
11
transportation demands from one depot. To make the models closer to real life, Li et al. (2016b)
12
took stochastic travel times and delivery locations into account. Ren et al. (2021) described the
13
dynamics in SAR by updating parcel delivery information and reoptimizing routes when
14
vehicles arrive at distribution centres (the origins of parcels). Considering the trends of
15
electrification and automation in transportation, Lu et al. (2022) investigated a system with a
16
mixed fleet of electric and gasoline taxis, while Beirigo et al. (2018), Tholen et al. (2021), and
17
Zhang et al. (2022) envisioned a system with SAVs.
18
To solve these problems, scholars have developed different models and solution
19
approaches. MIP models and two-stage stochastic programming models are commonly used
20
for deterministic problems and problems with uncertainty, respectively. Regarding solution
21

16
approaches, commercial solvers such as CPLEX and Gurobi can solve small instances (Li et
1
al., 2014; Beirigo et al., 2018; Tholen et al., 2021). Metaheuristics, e.g., genetic algorithm (Ren
2
et al., 2021), adaptive large neighbourhood search (ALNS) (Li et al., 2016a, 2016b), simulated
3
annealing (Yu et al., 2018, 2021a, 2022b), are widely applied to solve large-scale instances
4
since SARP is an NP-hard problem. Other solution methods include the Lagrangian dual
5
decomposition method (Zhang et al., 2022), math-heuristic (Lu et al., 2022), and model-free
6
deep reinforcement learning (Manchella et al., 2021a, 2021b).
7
3.1.2 Pricing
8
The above-mentioned studies were based on a given pricing strategy. Specifically, the initial
9
price for each passenger and parcel was considered to remain unchanged or increase based on
10
travel distance. However, a passenger may obtain a discount depending on the degree of
11
deviation from his/her direct route (Li et al., 2014; Ren et al., 2021; Yu et al., 2021a). A more
12
interesting strategy developed by Manchella et al. (2021a) allows drivers and passengers to
13
negotiate for the best price.
14
3.2 Freight on transit
15
3.2.1 Freight hub location
16
The freight hub location problem aims to choose among existing passenger public
17
transport stations as distribution centres for delivering or transhipping goods. Most papers on
18
this topic rely on the metro system as the backbone, indicating that researchers are optimistic
19

17
that freight can be successfully integrated with the metro system. We divide these studies into
1
two groups based on whether all selected freight hubs have the function of connecting
2
underground and ground networks. In the first group, goods can enter the metro system from
3
the ground and leave from the metro system to the ground at any selected freight hubs (Zhao
4
et al., 2018; Ji et al., 2020; Kizil & Yildiz, 2022). The decision variables are the locations of
5
freight hubs. In the second group, there are two types of freight hubs (Dong et al., 2018; Sun
6
et al., 2022). Freight hubs of the first type are similar to the freight hubs in the first group,
7
where goods have access and egress to the underground and ground networks. Freight hubs of
8
the second type can only be used for transhipping freight between different metro lines, not
9
connected to the ground network. In addition to deciding the locations of freight hubs, their
10
functions are decided as well.
11
Azcuy et al. (2021) considered a more general urban delivery system using public
12
transport where the public transport could be the bus, metro, tram, etc. Freight can be
13
transferred from public transport to last-mile delivery vehicles at any selected station.
14
3.2.2 Route planning
15
3.2.2.1 Route planning for public transport vehicles
16
Two types of public transport services have been studied in FOT, scheduled public transport
17
(with fixed lines and schedules) and demand-responsive public transport (with flexible routes
18
and schedules). When scheduled public transport is used in FOT, it is usually assumed that the
19

18
capacity of the public transport system is underutilised, and the existing routes and schedules
1
of public transport vehicles are treated as exogenous model parameters. Only Li et al. (2021)
2
designed the stations where added freight trains should stop, categorised as a route planning
3
problem.
4
When FOT is based on demand-responsive public transport, passengers and freight
5
could be on a shared network or in the same vehicle. For the former scenario, Fatnassi et al.
6
(2015) devised two routing strategies: a reactive dynamic matching strategy and a proactive
7
one for passenger rapid transit (PRT) and freight rapid transit (FRT). For the latter, Chebbi &
8
Chatuachi (2016) studied an empty vehicle redistribution problem that minimizes the empty
9
movement and the number of used vehicles while reducing the wasted capacity of PRT. Peng
10
et al. (2021) explored a bus-pooling service at a railway station, where demand-responsive
11
buses pick up passengers and parcels and deliver them to their destinations. Parcels with similar
12
itineraries and departure times to passengers were matched and inserted into bus routes
13
following the shortest road route.
14
3.2.2.2 Route planning for supportive vehicles
15
As conventional public transport modes with fixed routes cannot provide door-to-door services,
16
support vehicles (e.g., small trucks or electric vehicles run by logistic companies) are typically
17
used to realise the first/last-mile transportation.
18
Route planning for supportive vehicles is usually formulated as variants of pickup and
19
delivery problem (PDP) to accommodate operation modes. Masson et al. (2017) modelled a
20
PDP with transfers in a setting where all goods originate from warehouses known as
21

19
consolidation and distribution centres (CDCs). Buses start from a CDC and travel to bus stops
1
where goods are unloaded and transhipped to support vehicles for the last-mile delivery.
2
Similar work was performed by Ye et al. (2021) for a metro-based FOT. The difference is that
3
the supportive vehicles also perform the first-mile delivery from the CDC to metro stations.
4
Another variant is the PDP with scheduled lines (PDP-SL). Different from the PDP
5
with transfers where goods must take public transport, the PDP-SL allows goods to be either
6
delivered directly to customers by support vehicles or first collected by a support vehicle,
7
transported via scheduled lines (SLs) such as bus, train, metro, etc., and then delivered to
8
customers by another support vehicle (Ghilas et al., 2016a). Ghilas et al. (2016c) extended the
9
deterministic PDP-SL problem proposed by Ghilas et al. (2016a) into a stochastic PDP-SL
10
problem by considering uncertain freight demand. People and parcels only share public
11
transport vehicles in the two studies, whereas they share supportive vehicles in Ghilas et al.
12
(2013). All three studies assume that the freight capacity is fixed and not influenced by
13
passenger flows. This assumption is relaxed by Mourad et al. (2021), in which robots function
14
as support vehicles. When the delivery robots travel on a bus, they share the same bus capacity
15
with passengers but have a lower priority. In other words, at some stations, the robots may be
16
unable to board a bus or be required to deboard to make space for passengers.
17
As an alternative to using ground vehicles as support vehicles, Huang et al. (2020)
18
introduced an innovative integrated transportation system that involves trains with given routes
19
and timetables for passenger transportation and drones for parcel delivery.
20

20
In terms of the solution approaches, most studies apply metaheuristics to solve large-
1
scale instances, e.g., ALNS (Ghilas et al., 2016b, 2016c; Masson et al., 2017; Mourad et al.
2
2021) and variable neighbourhood search (Ye et al., 2021). Ghilas et al. (2018) developed a
3
branch-and-price algorithm to solve medium-sized instances.
4
3.2.3 Timetabling
5
The design of timetables for scheduled public transport vehicles in the context of FOT has been
6
considered only in rail-based FOT, i.e., trams and trains. When people and freight share the
7
vehicles/carriage, the timetables of passenger vehicles were designed from scratch, aiming to
8
transport more freight in less time without influencing passenger transport (Li et al., 2022).
9
When people and freight share the rail infrastructure, two strategies are found to design
10
timetables for added dedicated freight trains: 1) Timetables for freight trains are created while
11
the timetables of passenger trains remain unchanged (Ozturk & Patrick, 2018); 2) Schedules
12
for both passenger and freight trains are constructed from scratch (Li et al., 2021; Hörsting &
13
Cleophas, 2022). Hörsting & Cleophas (2022) compared the two transportation modes and
14
concluded that sharing vehicles/carriages is more robust towards fluctuating demand while
15
sharing infrastructures allows higher dwell time for dedicated freight trains/trams.
16
3.2.4 Freight flow assignment
17
The freight flow assignment problem determines where, when, and on which vehicle a request
18
takes a public transport ride. The flow assignment can be obtained either as key decision
19

21
variables in a model that exclusively determines the flow given public transport routes and
1
schedules of vehicles that can be used for freight transportation or as auxiliary variables in a
2
model that designs the hub location, route, and timetable (e.g., Ji et al., 2020; Ozturk & Patrick,
3
2018; Li et al., 2021). This section focuses on the former case.
4
For bus-based FOT, Pimentel & Alvelos (2018) developed an MIP model to determine
5
the freight flow that minimises the delivery time. Their system allows goods to be unloaded at
6
any stop but only loaded at specific bus stops. This situation was relaxed by Cheng et al. (2018)
7
by permitting goods to be loaded and unloaded at any stop. If the goods capacity on the part of
8
the selected route is not adequate, the goods are unloaded at intermediate stops and wait for the
9
next vehicle along the same route.
10
For rail-based FOT, Behiri et al. (2018) hypothesised that physical components of a rail
11
network, i.e., stations, railways, and trains, are shared by participants, and freights can be
12
loaded and unloaded at any station. The objective of their model is to minimise the total waiting
13
time of each demand, defined as the difference between the time at which demand is loaded
14
into the train and that at which it arrives at the departure station. In this manner, the turnover
15
of goods in stations can be maximised. Sahli et al. (2022) simplified the model and improved
16
the heuristic solution algorithm proposed by Behiri et al. (2018). Di et al. (2022) considered a
17
system where freight and passengers are allowed to share each service train. In addition to
18
optimising the flow assignment, they also optimised the carriage arrangement.
19

22
3.2.5 Pricing
1
Most studies on FOT focus on solving operation management problems to reduce operation
2
costs and total delivery time. Only a few consider the price charged to customers and the profits
3
operators gain. We only found two papers mentioning the price. The first one is Li et al. (2021),
4
which set a parameter representing the price of transporting a container. The other is Ma et al.
5
(2022), which jointly optimised a logistic company’s modal split strategy and a metro
6
company’s pricing strategy based on non-cooperative and cooperative game theoretical
7
models. The results showed that metro-based FOT could generate Pareto-improving outcomes
8
for the metro and logistics companies.
9
3.3 Crowdshipping
10
3.3.1 Demand and supply prediction
11
Two approaches can be used to predict the demand and supply in crowdshipping. The first one
12
is to use historical data (Shen & Lin, 2020), while the second one is to identify the factors that
13
influence the demand and supply through a survey (Le et al., 2019; Gatta et al., 2019; Le &
14
Ukkusuri, 2019; Rechavi & Toch, 2022; Ermagun et al., 2020). As crowdshipping is a new
15
service, only a limited amount of historical data is available. Thus, most are based on the second
16
approach. The key factors influencing the demand and supply of crowdshipping are listed as
17
follows.
18

23
(1) Demand: Dry cleaning, groceries, and home-delivered foods are favoured categories of
1
goods in crowdshipping. Factors influencing people’s acceptance rate of
2
crowdshipping include personal attributes (including socio-demographic
3
characteristics), built environments, crowd types, and driver performance. Specifically,
4
younger people, online shoppers, and people with a strong sense of community and
5
environmental concern are more likely to accept crowdshipping; areas with high
6
population density but low job accessibility are suitable for crowdshipping
7
development (Buldeo Rai et al., 2021; Le et al., 2019).
8
(2) Supply: Young individuals and students are more likely to work as crowdshippers. The
9
supply of crowdshipping is enhanced by lower additional travel time spent on
10
crowdsourced tasks, higher remuneration, and higher levels of crowdshipping
11
experience (Ermagun et al., 2020; Fessler et al., 2022; Gatta et al., 2019; Le &
12
Ukkusuri, 2019; Rechavi & Toch, 2022).
13
3.3.2 Matching strategy
14
As defined in section 2.1, in this review, we focus on the case in which crowdshippers are
15
matched with delivery requests on their way to a pre-planned trip, known as en route matching
16
(Alnaggar et al., 2021).
17
The key component in the en route matching problem is the criteria for an acceptable
18
matching. Most studies set a maximum percentage by which crowdshippers can deviate from
19
their normal trip in terms of the distance or travel time (Al Hla et al., 2019; Archetti et al.,
20

24
2016; Martín-Santamaría et al., 2021; Zehtabian et al., 2022). Additionally, the crowdshipper’s
1
earliest departure time at his/her origin and latest arrival time at the destination can be confined
2
(Chen et al., 2018; Macrina et al., 2020; Arslan et al., 2019). Other than time-specified criteria,
3
the maximum number of parcels or stops that crowdshippers accept are used by Arslan et al.
4
(2019), Wang et al. (2016), Voigt & Kuhn (2022), and Zehtabian et al. (2022). Instead of a
5
single match, Ausseil et al. (2022) and Mancini & Gansterer (2022) provided several options
6
for a crowdshipper to choose from.
7
3.3.3 Facility location
8
Most studies on crowdshipping have focused on operational-level decisions and considered
9
point-to-point deliveries in which the origin-destination pairs of crowdshippers are close to
10
those of the parcels to be delivered. This may cause a lower success delivery rate compared
11
with crowdshipping allowing relays. To overcome this challenge, facilities such as parcel
12
lockers could be established to connect multiple crowdshippers for the same task, leading to a
13
facility location problem. Ghaderi et al. (2022) developed a two-phase algorithm to locate the
14
parcel lockers to maximise total profits and delivery rate. Considering the stochastic crowd
15
capacity and demands, Nieto-Isaza et al. (2022) developed a two-stage stochastic programming
16
model to determine the locations of mini depots to minimise total expected installation and
17
transportation costs.
18

25
3.3.4 Route planning for vehicles and goods
1
The route planning problem includes routing for a fleet owned by an operator and occasional
2
drivers. Archetti et al. (2016) initially modelled this problem as a VRP with occasional drivers
3
(VRPOD). This framework involves a single depot from where goods, dedicated vehicles, and
4
occasional drivers start. Many variants of this simple setting have been studied. For example,
5
Al Hla et al. (2019) considered the behaviours of both regular and occasional drivers. Triki
6
(2021) allowed occasional drivers to bid for delivery tasks. Macrina et al. (2017) considered
7
the time windows of customers. Macrina et al. (2020), Yu et at. (2021b, 2022a), and Lan et al.
8
(2022) introduced transhipment nodes. Besides, the stochasticity and dynamics of the
9
crowdshipping system have been addressed by Archetti et al. (2021), Dahle et al. (2017),
10
Mousavi et al. (2022), Dayarian & Savelsbergh (2020), Santini et al. (2022), and Silva &
11
Pedroso (2022).
12
In the above-mentioned VRPOD framework, all parcels originate from the depot, and
13
a single crowdshipper fulfils the crowdsourced delivery. This overlooks the possibility that
14
relaying crowdsourced tasks between crowdshippers could attract more participants to work as
15
crowdshippers and increase the delivery success rate, addressed by Chen et al. (2018) and Voigt
16
& Kuhn (2022). They extended the VRPOD to a pickup and delivery problem with occasional
17
drivers, where each parcel has a different origin, vehicles not only deliver parcels but also
18
collect parcels, and relays between crowdshippers are allowed.
19
Unlike the VRPOD that determines the operator-scheduled vehicle routes, Yildiz
20
(2021a, 2021b) explored the express package routing problem to determine the combinations
21

26
of self-scheduled trips of crowdshippers to fulfil transportation requests. They designed a
1
system involving service points at which pickup and drop-off operations occur. Senders drop
2
off goods at their selected service points, and receivers pick up goods from other service points.
3
Crowdshippers are responsible for transferring goods between the selected service points. Each
4
crowdshipper only performs one single trip between service points, but one task may have
5
multi-leg trips.
6
Crowdshippers are typically ordinary people. In addition, taxi drivers can serve as
7
crowdshippers to transport parcels without influencing the passenger service (Chen et al., 2016;
8
Chen et al., 2017; Cheng et al., 2022). Elsewhere, Boysen et al. (2022) considered voluntary
9
employees of distribution centres as crowdshippers and optimised the delivery routes for these
10
employees to maximise the number of parcels assigned to them.
11
3.3.5 Pricing and compensation
12
Crowdshippers usually receive compensation from retailers such as Walmart and logistics
13
companies. Five compensation schemes have been proposed in the literature: 1) Customer-
14
dependent compensation: The compensation for crowdshippers depends on the customer’s
15
location. A larger distance between the customer and depot corresponds to a higher
16
compensation for crowdshippers (Archetti et al., 2016, 2021; Dahle et al., 2019; Macrina et al.,
17
2017); 2) Crowdshipper dependent compensation: The compensation for crowdshippers
18
depends on the amount of additional distance travelled or additional time spent compared with
19
normal travel (Dahle et al., 2019; Yu et al., 2021b, 2022a ); 3) Fixed compensation for each
20

27
delivery: The compensation for crowdshippers depends on the number of deliveries performed
1
(Boysen et al., 2022; Dahle et al., 2019; Lan et al., 2022; Santini et al., 2022; Yildiz,
2
2021a,2021b); 4) Combined compensation: The compensation for crowdshippers consists of
3
two elements. Specifically, a fixed compensation is provided when the crowdshippers fulfil at
4
least one delivery, and variable compensation is provided depending on additional efforts (extra
5
travel distance or travel time) made for crowdsourced deliveries (Dahle et al., 2019; Dayarian
6
& Savelsbergh, 2020; Mousavi et al., 2022); 5) Auction-based compensation: Crowdshippers
7
bid for crowdsourced tasks and are paid their bidding price if they win (Triki, 2021).
8
In terms of optimising the compensation provided to crowdshippers and the price
9
charged for requesters, we found three studies in which a third-party platform controls the
10
crowdshipping service. Le et al. (2021) optimised the price and compensation from the
11
platform’s perspective, aiming to maximise the platform’s profits. Zhou et al. (2021) proposed
12
a pricing strategy considering the varying package–driver ratio in a local region to maximise
13
the number of stable matches such that both the requester and crowdshipper have strong
14
incentives to be matched. These two studies neglect attributes associated with the parcel, e.g.,
15
weight and size, which may influence the behaviours of the receivers and crowdshippers. This
16
aspect is addressed by Xiao et al. (2021), wherein a multi-unit multi-attribute auction for
17
crowdsourced delivery to maximise social welfare is designed.
18
4 General framework
19
Based on the above review of the existing forms of integrated transportation systems, we
20

28
propose a general framework for planning and operating such systems, as shown in Figure 2.
1
First, we divide an integrated transportation system into three core components, each
2
comprising various elements worthy of investigation.
3
(1) Passenger and goods demand. Demand is generated by the requirement of people to
4
move from origins to destinations for a certain purpose, such as work, while goods must
5
be delivered from senders to recipients to fulfil customers’ requirements.
6
(2) Transport supply. Integrated people-and-goods transportation operators may be public
7
transport operators (e.g., bus, metro, and train companies), private transportation
8
companies, retailers with their own fleets and dedicated drivers, or third-party
9
companies that employ occasional drivers who use their vehicles to perform tasks.
10
(3) Infrastructure and technology. The infrastructure includes all materials that support the
11
integrated transportation of people and goods, e.g., roads, railways, and information
12
and communications technology.
13
In this context, each operator must address the following three main problems, which
14
contain two or three subproblems.
15
(1) Demand management. This problem includes i) demand prediction to understand how
16
travellers and senders make transportation decisions; ii) pricing strategies to control the
17
spatiotemporal distribution of transportation demand.
18

29
(2) Supply management. This problem includes predicting and planning the capacity that
1
can be used to fulfil the transportation demand and designing compensation strategies
2
to control the supply.
3
(3) Demand and supply matching. This problem aims at matching a specific request with
4
a vehicle, which typically involves the design of the routes and schedules of vehicles
5
or the assignment of requests to a vehicle, depending on the integration form.
6
Figure 2 also shows the subproblems examined in the existing studies. Several gaps
7
remain, considered promising research directions, as described in Section 5.
8
The performance of the integrated transportation system could be evaluated from
9
different perspectives. On the demand side, passengers care mostly about the travel time,
10
waiting time, and travel cost, while shippers are more concerned with the transportation costs
11
and whether the goods are delivered in a satisfactory condition and timely manner. On the
12
supply side, operators focus mostly on profit, operation cost, and demand satisfaction rate.
13
Solving the operation management problems will directly influence the indicators on the two
14
sides, which will further influence the demand and supply. The demand and supply, in turn,
15
influence the operational strategies and decisions and the resulting efficiency of the integrated
16
people-and-goods transportation system. The application of integrated people-and-goods
17
transportation will inevitably impact various aspects of sustainability, like environmental (e.g.,
18
air pollutants) and social (e.g., employment, equity). Nevertheless, they are well beyond the
19
scope of this review and are left for future work.
20

30
1
Figure 2 A general framework for integrated people-and-goods transportation systems
2
5 Research gaps and future directions
3
Based on the framework proposed in Section 4, we specify future research directions from
4
three aspects. For each aspect, we categorise two types of future research aimed at, respectively,
5
filling the research gap and enhancing the existing research. Notably, research gaps may exist
6
owing to the different implementation methods of different forms in practice. Then, we
7
presented some research opportunities in the era of technology.
8

31
5.1 Demand management
1
5.1.1. Pioneering research
2
As shown in Figure 2, demand prediction for SAR and FOT has not been extensively studied,
3
despite its importance for service operators in providing supply that matches the demand.
4
Currently, both SAR and FOT are in the early implementation stage, and the amount of
5
historical data is inadequate. In this context, the demand can be predicted by identifying the
6
factors influencing people’s choices. To this end, unique features/phenomena associated with
7
the system must be understood. For example, in terms of passenger demand, some passengers
8
may transfer from separated to integrated modes due to decreased travel costs. In contrast,
9
passengers with a high value of time are less likely to accept a detour for the delivery of goods
10
even if a discount is available. Moreover, some passengers may refuse to be transported with
11
goods owing to safety and comfort concerns when the goods are placed in the same vehicle (or
12
carriage). In terms of goods demands, the incomes of senders and recipients, environmental
13
conscientiousness, and requirement for time windows determine whether the senders and
14
recipients choose an SAR vehicle or public transport for delivery. Additionally, the goods’
15
attributes (type, size, weight, volume, value, etc.) determine whether they can be transported
16
with people and the mode suitable for transporting them. For example, dangerous goods cannot
17
be transported with passengers, while groceries, which typically involve a large volume and
18
number of goods, can be transported by trains or metros instead of taxis. In the future, with the
19
development of SAR and FOT and the availability of adequate data, historical data may be
20

32
used to predict the spatiotemporal transportation demand distribution.
1
5.1.2. Research for improvement
2
Most studies have considered the price charged to customers as a parameter under different
3
pricing schemes. Because the influence of pricing on demand and supply is complex, the
4
pricing can instead be considered a decision variable and determined using an optimization
5
model.
6
5.2 Supply management
7
5.2.1. Pioneer research
8
Supply management in SAR frameworks has not been extensively investigated. This is
9
probably because SAR can be implemented in multiple ways, and a common supply
10
management strategy cannot be applied to all methods. In practice, the supply management
11
subproblem to be addressed depends on the type of operator. Operators are classified in terms
12
of the possession of the fleet and drivers. The supply management subproblem for operators
13
with their own fleet and drivers is focused on resource planning because the supply is
14
determined by the fleet size, the dedicated drivers must follow the routes designed by the
15
operators to service customers, and the drivers are paid salaries by the operators. In this
16
framework, the supply distribution does not need to be predicted, and no compensation exists.
17
The supply management subproblem for operators without their own fleet and drivers is
18
focused on supply prediction and compensation because the supply of drivers is affected by
19

33
various factors such as age, income, and compensation for each request.
1
The supply prediction for FOT has not been extensively studied. Compensation for
2
drivers might not be considered a problem that needs to be practically addressed, as drivers of
3
public transport vehicles typically receive a regular monthly salary. For public transport
4
operators, supply prediction is focused on predicting the available capacity that can be used for
5
transporting passengers and goods. This supply decides whether goods and passengers can be
6
transported simultaneously in specific periods. This aspect can be considered a counterpart to
7
predicting the passengers’ route choice behaviour given the transit capacities. In the case of
8
underused capacity, companies can enter the integrated people-and-goods transport market.
9
Otherwise, the companies can simply serve passengers as usual.
10
5.2.2. Research for improvement
11
Scheduled-public-transport-based FOT consists of two parts: a public transport mode for
12
backbone transportation and support vehicles for first/last-mile transportation. Although
13
several researchers have studied FOT based on various transportation means, the backbone
14
transportation consisted of a single public transport mode. In practice, each public transport
15
mode has its advantages and disadvantages. For example, metros and trains are faster and more
16
punctual than buses, while buses have a wider service area. Different public transportation
17
modes can be combined to fully exploit the advantages of all parties to provide a more time-
18
efficient or cost-efficient service. Because hubs for transferring passengers from one public
19
transport mode to another already exist, goods can also be transhipped at these nodes. In
20

34
addition, as described in Section 3.2.1, most studies on freight hub selection have focused on
1
metro-based FOT. For bus-based FOT, it is typically assumed that all bus stations can be used
2
to handle and tranship goods. However, considering the goods transportation demand and cost
3
of reforming a passenger platform to an enhanced platform suitable for goods storage and
4
transhipment, it is not reasonable to set all passenger boarding/alighting points as goods
5
loading/unloading points. Therefore, the hub location problem must also be addressed in bus-
6
based FOT. In addition, the fleet size of scheduled and demand-responsive public transport
7
vehicles considerably influences the supply and must be further investigated.
8
For crowdshipping, research can be performed to understand peoples’ attitudes toward
9
public-transport-based crowdshipping. Moreover, as in the case of the pricing problem in
10
demand management, the compensation provided to crowdshippers affects the supply of
11
crowdshippers. Therefore, the compensation of crowdshippers must be optimised. In addition,
12
researchers can optimise the location and capacity of the support infrastructure. In this way,
13
similar origin-destination pairs between crowdshippers and parcels do not need to be identified,
14
making the system more flexible and efficient.
15
5.3 Matching
16
5.3.1. Pioneer research
17
The matching problem for SAR has not been studied, probably because the existing studies
18
have assumed that the SAR operators belong to the first type of operator, as described in
19
Section 5.2.1. In this scenario, the dedicated drivers working for the operator cannot reject
20

35
requests assigned to them. If the operators belong to the second type, the occasional drivers
1
can reject requests assigned to them. In this scenario, an optimal matching problem must be
2
solved to increase the successful matching rate. Different matching strategies can be explored,
3
e.g., en route matching and negotiation between occasional drivers and requesters.
4
5.3.2. Research for improvement
5
The following research directions can be considered for improving research on matching
6
transportation demands with supply.
7
First, for SAR with the second type of operators, occasional drivers can be guided to
8
reposition their vehicles after finishing their tasks to increase their chance of accepting another
9
task when they do not have a personal trip planned after the delivery.
10
Second, for scheduled-public-transport-based FOT, matching between goods requests
11
with a specific public transport vehicle can be examined in the context of multimodal urban
12
transport so that a goods request can be matched to a multimodal trip chain of public transport
13
services.
14
Third, we recommend developing models to solve the matching problem in public-
15
transport-based crowdshipping. This matching is different from that in private-vehicle-based
16
crowdshipping because a parcel request suitable for a private-vehicle-based crowdshipper
17
might not be accepted by a public-transport-based crowdshipper.
18

36
Fourth, researchers can attempt to solve the subproblems associated with all three forms
1
of integrated systems in a stochastic and dynamic setting, as this is more realistic and research
2
in this domain is limited.
3
5.4 Opportunities in the era of technology
4
Rapid developments of new technologies, e.g., 5G technology, artificial intelligence
5
(AI), Internet of things (IoT), autonomous vehicles (AVs), digital twins, etc., will revolutionize
6
the transportation industry, which brings opportunities for the development of integrated
7
transportation system. First, the potential deployment of AVs, drones, and robots and their
8
impacts on integrated transportation systems should be studied before they are widely applied.
9
Second, technologies such as ICT and intelligent transportation systems enable the
10
synchromodality, which aims to provide efficient, reliable, and flexible transportation services
11
using real-time information. This strengthens the need for fast online algorithms to support
12
real-time re-optimisation. Third, driven by AI, IoT, etc., digital twins could be used to simulate
13
different activities in the integrated transportation system, which enables planners to manage
14
transportation dynamically, react to unexpected events appropriately, etc. Moreover, digital
15
twins could be used to analyse the potential impacts of new concepts before real
16
implementations. As Arvidsson & Browne (2013) recommended, it is better to try a new
17
concept in a small-scale fashion and gradually scale up, especially for big projects requiring
18
high investment or new infrastructure. This could be achieved by digital twins in a time and
19
cost-efficient way to explore the economic viability and scalability to meet the exploding
20

37
delivery demand and the need for infrastructure and equipment investment for the integrated
1
transportation system before real application. We recommend that researchers apply more
2
advanced methods, e.g., digital twins, in the era of technologies to assess the feasibility of an
3
integrated system, in addition to using traditional methods such as simulation.
4
Declaration of interest statement
5
The authors declare that they have no known competing financial interests or personal
6
relationships that could have appeared to influence the work reported in this paper.
7
References
8
Al Hla, Y. A., Othman, M., & Saleh, Y. (2019). Optimizing an eco-friendly vehicle routing
9
problem model using regular and occasional drivers integrated with driver behaviour
10
control. Journal of Cleaner Production, 234, 984-1001.
11
Alnaggar, A., Gzara, F., & Bookbinder, J. H. (2021). Crowdsourced delivery: A review of
12
platforms and academic literature. Omega, 98, 102139.
13
Archetti, C., Savelsbergh, M., & Speranza, M. G. (2016). The vehicle routing problem with
14
occasional drivers. European Journal of Operational Research, 254(2), 472-480.
15
Archetti, C., Guerriero, F., & Macrina, G. (2021). The online vehicle routing problem with
16
occasional drivers. Computers & Operations Research, 127, 105144.
17
Arslan, A. M., Agatz, N., Kroon, L., & Zuidwijk, R. (2019). Crowdsourced delivery—A
18
dynamic pickup and delivery problem with ad hoc drivers. Transportation Science,
19
53(1), 222-235.
20
Arvidsson, N., & Browne, M. (2013). A review of the success and failure of tram systems to
21
carry urban freight: the implications for a low emission intermodal solution using
22
electric vehicles on trams.
23

38
Arvidsson, N., Givoni, M., & Woxenius, J. (2016). Exploring last mile synergies in passenger
1
and freight transport. Built Environment, 42(4), 523-538.
2
Ausseil, R., Pazour, J. A., & Ulmer, M. W. (2022). Supplier menus for dynamic matching in
3
peer-to-peer transportation platforms. Transportation Science, 56(5), 1304-1326.
4
Azcuy, I., Agatz, N., & Giesen, R. (2021). Designing integrated urban delivery systems using
5
public transport. Transportation Research Part E: Logistics and Transportation Review,
6
156, 102525.
7
Behiri, W., Belmokhtar-Berraf, S., & Chu, C. (2018). Urban freight transport using passenger
8
rail network: Scientific issues and quantitative analysis. Transportation Research Part
9
E: Logistics and Transportation Review, 115, 227-245.
10
Beirigo, B. A., Schulte, F., & Negenborn, R. R. (2018). Integrating people and freight
11
transportation using shared autonomous vehicles with compartments. IFAC-
12
PapersOnLine, 51(9), 392-397.
13
Boysen, N., Emde, S., & Schwerdfeger, S. (2022). Crowdshipping by employees of distribution
14
centres: Optimization approaches for matching supply and demand. European Journal
15
of Operational Research, 296(2), 539-556.
16
Bruzzone, F., Cavallaro, F., & Nocera, S. (2021). The integration of passenger and freight
17
transport for first-last mile operations. Transport policy, 100, 31-48.
18
Buldeo Rai, H., Verlinde, S., Merckx, J., & Macharis, C. (2017). Crowd logistics: an
19
opportunity for more sustainable urban freight transport?. European Transport Research
20
Review, 9(3), 1-13.
21
Buldeo Rai, H., Verlinde, S., & Macharis, C. (2018). Shipping outside the box. Environmental
22
impact and stakeholder analysis of a crowd logistics platform in Belgium. Journal of
23
Cleaner Production, 202, 806-816.
24
Buldeo Rai, H., Verlinde, S., & Macharis, C. (2021). Who is interested in a crowdsourced last
25
mile? A segmentation of attitudinal profiles. Travel Behaviour and Society, 22, 22-31.
26

39
Cavallaro, F., & Nocera, S. (2022). Integration of passenger and freight transport: A concept-
1
centric literature review. Research in Transportation Business & Management, 43,
2
100718.
3
Chebbi, O., & Chaouachi, J. (2016). Reducing the wasted transportation capacity of personal
4
rapid transit systems: an integrated model and multi-objective optimization approach.
5
Transportation research part E: logistics and transportation review, 89, 236-258.
6
Chen, C., Pan, S., Wang, Z., & Zhong, R. Y. (2017). Using taxis to collect citywide E-
7
commerce reverse flows: a crowdsourcing solution. International Journal of Production
8
Research, 55(7), 1833-1844.
9
Chen, C., Zhang, D., Ma, X., Guo, B., Wang, L., Wang, Y., & Sha, E. (2016). Crowddeliver:
10
Planning city-wide package delivery paths leveraging the crowd of taxis. IEEE
11
Transactions on Intelligent Transportation Systems, 18(6), 1478-1496.
12
Chen, W., Mes, M., & Schutten, M. (2018). Multi-hop driver-parcel matching problem with
13
time windows. Flexible services and manufacturing journal, 30, 517-553.
14
Cheng, G., Guo, D., Shi, J., & Qin, Y. (2018, December). When packages ride a bus: Towards
15
efficient city-wide package distribution. In 2018 IEEE 24th International Conference
16
on Parallel and Distributed Systems (ICPADS) (pp. 259-266). IEEE.
17
Cheng, S., Chen, C., Pan, S., Huang, H., Zhang, W., & Feng, Y. (2022). Citywide package
18
deliveries via crowdshipping: minimizing the efforts from crowdsourcers. Frontiers of
19
Computer Science, 16, 1-13.
20
Cleophas, C., Cottrill, C., Ehmke, J. F., & Tierney, K. (2019). Collaborative urban
21
transportation: Recent advances in theory and practice. European Journal of
22
Operational Research, 273(3), 801-816.
23
Cochrane, K., Saxe, S., Roorda, M. J., & Shalaby, A. (2017). Moving freight on public transit:
24
Best practices, challenges, and opportunities. International Journal of Sustainable
25
Transportation, 11(2), 120-132.
26

40
Dahle, L., Andersson, H., Christiansen, M., & Speranza, M. G. (2019). The pickup and delivery
1
problem with time windows and occasional drivers. Computers & Operations Research,
2
109, 122-133.
3
Dahle, L., Andersson, H., & Christiansen, M. (2017). The vehicle routing problem with
4
dynamic occasional drivers. In Computational Logistics: 8th International Conference,
5
ICCL 2017, Southampton, UK, October 18-20, 2017, Proceedings 8 (pp. 49-63).
6
Springer International Publishing.
7
Dayarian, I., & Savelsbergh, M. (2020). Crowdshipping and same‐day delivery: Employing in‐
8
store customers to deliver online orders. Production and Operations Management,
9
29(9), 2153-2174.
10
Di, Z., Yang, L., Shi, J., Zhou, H., Yang, K., & Gao, Z. (2022). Joint optimization of carriage
11
arrangement and flow control in a metro-based underground logistics system.
12
Transportation Research Part B: Methodological, 159, 1-23.
13
Dong, J., Hu, W., Yan, S., Ren, R., & Zhao, X. (2018). Network planning method for
14
capacitated metro-based underground logistics system. Advances in Civil Engineering,
15
2018.
16
Elbert, R., & Rentschler, J. (2022). Freight on urban public transportation: A systematic
17
literature review. Research in Transportation Business & Management, 45, 100679.
18
Ermagun, A., Punel, A., & Stathopoulos, A. (2020). Shipment status prediction in online
19
crowdsourced shipping platforms. Sustainable Cities and Society, 53, 101950.
20
European Commission. (2007). Green paper, towards a new culture for urban mobility.
21
European Union, Brussels.
22
European Commission. (2019). Transport in the European Union – current trends and issues.
23
https://ec.europa.eu/transport/sites/transport/files/2019-transport-in-theeu-current-
24
trends-and-issues.pdf. (Accessed 10 May 2021)
25
Fatnassi, E., Chaouachi, J., & Klibi, W. (2015). Planning and operating a shared goods and
26
passengers on-demand rapid transit system for sustainable city-logistics. Transportation
27
Research Part B: Methodological, 81, 440-460.
28

41
Fessler, A., Thorhauge, M., Mabit, S., & Haustein, S. (2022). A public transport-based
1
crowdshipping concept as a sustainable last-mile solution: Assessing user preferences
2
with a stated choice experiment. Transportation Research Part A: Policy and Practice,
3
158, 210-223.
4
Gatta, V., Marcucci, E., Nigro, M., & Serafini, S. (2019). Sustainable urban freight transport
5
adopting public transport-based crowdshipping for B2C deliveries. European Transport
6
Research Review, 11(1), 1-14.
7
Ghaderi, H., Zhang, L., Tsai, P. W., & Woo, J. (2022). Crowdsourced last-mile delivery with
8
parcel lockers. International Journal of Production Economics, 251, 108549.
9
Ghilas, V., Cordeau, J. F., Demir, E., & Woensel, T. V. (2018). Branch-and-price for the pickup
10
and delivery problem with time windows and scheduled lines. Transportation Science,
11
52(5), 1191-1210.
12
Ghilas, V., Demir, E., & Van Woensel, T. (2013). Integrating passenger and freight
13
transportation: Model formulation and insights. Proceedings of the 2013 Beta Working
14
Papers; Technische Universiteit Eindhoven: Eindhoven, The Netherlands, 441, 1-23.
15
Ghilas, V., Demir, E., & Van Woensel, T. (2016a). The pickup and delivery problem with time
16
windows and scheduled lines. INFOR: Information Systems and Operational Research,
17
54(2), 147-167.
18
Ghilas, V., Demir, E., & Van Woensel, T. (2016b). An adaptive large neighbourhood search
19
heuristic for the pickup and delivery problem with time windows and scheduled lines.
20
Computers & Operations Research, 72, 12-30.
21
Ghilas, V., Demir, E., & Van Woensel, T. (2016c). A scenario-based planning for the pickup
22
and delivery problem with time windows, scheduled lines and stochastic demands.
23
Transportation Research Part B: Methodological, 91, 34-51.
24
He, Y., & Csiszár, C. (2021). Model for crowdsourced parcel delivery embedded into mobility
25
as a service based on autonomous electric vehicles. Energies, 14(11), 3042.
26
Hörsting, L., & Cleophas, C. (2022). Scheduling shared passenger and freight transport on a
27
fixed infrastructure. European Journal of Operational Research.
28

42
Huang, H., Savkin, A. V., & Huang, C. (2020). A new parcel delivery system with drones and
1
a public train. Journal of Intelligent & Robotic Systems, 100, 1341-1354.
2
Ji, Y., Zheng, Y., Zhao, J., Shen, Y., & Du, Y. (2020). A multimodal passenger-and-package
3
sharing network for urban logistics. Journal of Advanced Transportation, 2020, 1-16.
4
Kikuta, J., Ito, T., Tomiyama, I., Yamamoto, S., & Yamada, T. (2012). New subway-integrated
5
city logistics system. Procedia-Social and Behavioral Sciences, 39, 476-489.
6
Kizil, K. U., & Yildiz, B. (2022). Public Transport-based Crowd-shipping with Backup
7
Transfers. Transportation Science.
8
Lan, Y. L., Liu, F., Ng, W. W., Gui, M., & Lai, C. (2022). Multi-objective two-echelon city
9
dispatching problem with mobile satellites and crowd-shipping. IEEE Transactions on
10
Intelligent Transportation Systems, 23(9), 15340-15353.
11
Le, T. V., Stathopoulos, A., Van Woensel, T., & Ukkusuri, S. V. (2019). Supply, demand,
12
operations, and management of crowd-shipping services: A review and empirical
13
evidence. Transportation Research Part C: Emerging Technologies, 103, 83-103.
14
Le, T. V., & Ukkusuri, S. V. (2019). Modelling the willingness to work as crowd-shippers and
15
travel time tolerance in emerging logistics services. Travel Behaviour and Society, 15,
16
123-132.
17
Le, T. V., Ukkusuri, S. V., Xue, J., & Van Woensel, T. (2021). Designing pricing and
18
compensation schemes by integrating matching and routing models for crowd-shipping
19
systems. Transportation Research Part E: Logistics and Transportation Review, 149,
20
102209.
21
Le Pira, M., Tavasszy, L. A., de Almeida Correia, G. H., Ignaccolo, M., & Inturri, G. (2021).
22
Opportunities for integration between mobility as a service (MaaS) and freight
23
transport: A conceptual model. Sustainable Cities and Society, 74, 103212.
24
Li, B., Krushinsky, D., Reijers, H. A., & Van Woensel, T. (2014). The share-a-ride problem:
25
People and parcels sharing taxis. European Journal of Operational Research, 238(1),
26
31-40.
27

43
Li, B., Krushinsky, D., Van Woensel, T., & Reijers, H. A. (2016a). An adaptive large
1
neighbourhood search heuristic for the share-a-ride problem. Computers & Operations
2
Research, 66, 170-180.
3
Li, B., Krushinsky, D., Van Woensel, T., & Reijers, H. A. (2016b). The share-a-ride problem
4
with stochastic travel times and stochastic delivery locations. Transportation Research
5
Part C: Emerging Technologies, 67, 95-108.
6
Li, F., Guo, X., Zhou, L., Wu, J., & Li, T. (2022). A capacity matching model in a collaborative
7
urban public transport system: integrating passenger and freight transportation.
8
International Journal of Production Research, 60(20), 6303-6328.
9
Li, Z., Shalaby, A., Roorda, M. J., & Mao, B. (2021). Urban rail service design for collaborative
10
passenger and freight transport. Transportation Research Part E: Logistics and
11
Transportation Review, 147, 102205.
12
Lu, C. C., Diabat, A., Li, Y. T., & Yang, Y. M. (2022). Combined passenger and parcel
13
transportation using a mixed fleet of electric and gasoline vehicles. Transportation
14
Research Part E: Logistics and Transportation Review, 157, 102546.
15
Ma, M., Zhang, F., Liu, W., & Dixit, V. (2022). A game theoretical analysis of metro-integrated
16
city logistics systems. Transportation Research Part B: Methodological, 156, 14-27.
17
Macrina, G., Pugliese, L. D. P., Guerriero, F., & Laganà, D. (2017). The vehicle routing
18
problem with occasional drivers and time windows. In Optimization and Decision
19
Science: Methodologies and Applications: ODS, Sorrento, Italy, September 4-7, 2017
20
47 (pp. 577-587). Springer International Publishing.
21
Macrina, G., Pugliese, L. D. P., Guerriero, F., & Laporte, G. (2020). Crowd-shipping with time
22
windows and transshipment nodes. Computers & Operations Research, 113, 104806.
23
Manchella, K., Haliem, M., Aggarwal, V., & Bhargava, B. (2021a). Passgoodpool: Joint
24
passengers and goods fleet management with reinforcement learning aided pricing,
25
matching, and route planning. IEEE Transactions on Intelligent Transportation
26
Systems, 23(4), 3866-3877.
27

44
Manchella, K., Umrawal, A. K., & Aggarwal, V. (2021b). FlexPool: A distributed model-free
1
deep reinforcement learning algorithm for joint passengers & goods transportation.
2
IEEE Transactions on Intelligent Transportation Systems, 22(4), 2035-2047.
3
Mancini, S., & Gansterer, M. (2022). Bundle generation for last-mile delivery with occasional
4
drivers. Omega, 108, 102582.
5
Martín-Santamaría, R., López-Sánchez, A. D., Delgado-Jalón, M. L., & Colmenar, J. M.
6
(2021). An efficient algorithm for crowd logistics optimization. Mathematics, 9(5),
7
509.
8
Masson, R., Trentini, A., Lehuédé, F., Malhéné, N., Péton, O., & Tlahig, H. (2017).
9
Optimization of a city logistics transportation system with mixed passengers and goods.
10
EURO Journal on Transportation and Logistics, 6(1), 81-109.
11
Mourad, A., Puchinger, J., & Chu, C. (2019). A survey of models and algorithms for optimizing
12
shared mobility. Transportation Research Part B: Methodological, 123, 323-346.
13
Mourad, A., Puchinger, J., & Van Woensel, T. (2021). Integrating autonomous delivery service
14
into a passenger transportation system. International Journal of Production Research,
15
59(7), 2116-2139.
16
Mousavi, K., Bodur, M., & Roorda, M. J. (2022). Stochastic last-mile delivery with crowd-
17
shipping and mobile depots. Transportation Science, 56(3), 612-630.
18
Nieto-Isaza, S., Fontaine, P., & Minner, S. (2022). The value of stochastic crowd resources and
19
strategic location of mini-depots for last-mile delivery: a Benders decomposition
20
approach. Transportation Research Part B: Methodological, 157, 62-79.
21
Ozturk, O., & Patrick, J. (2018). An optimization model for freight transport using urban rail
22
transit. European Journal of Operational Research, 267(3), 1110-1121.
23
Peng, Z., Feng, R., Wang, C., Jiang, Y., & Yao, B. (2021). Online bus-pooling service at the
24
railway station for passengers and parcels sharing buses: A case in Dalian. Expert
25
Systems with Applications, 169, 114354.
26

45
Pimentel, C., & Alvelos, F. (2018). Integrated urban freight logistics combining passenger and
1
freight flows–mathematical model proposal. Transportation research procedia, 30, 80-
2
89.
3
Qu, X., Wang, S., & Niemeier, D. (2022). On the urban-rural bus transit system with passenger-
4
freight mixed flow. Communications in Transportation Research, 2, 100054.
5
Rechavi, A., & Toch, E. (2022). Crowd logistics: Understanding auction-based pricing and
6
couriers’ strategies in crowdsourcing package delivery. Journal of Intelligent
7
Transportation Systems, 26(2), 129-144.
8
Ren, T., Jiang, Z., Cai, X., Yu, Y., Xing, L., Zhuang, Y., & Li, Z. (2021). A dynamic routing
9
optimization problem considering joint delivery of passengers and parcels. Neural
10
Computing and Applications, 33, 10323-10334.
11
Sahli, A., Behiri, W., Belmokhtar-Berraf, S., & Chu, C. (2022). An effective and robust genetic
12
algorithm for urban freight transport scheduling using passenger rail network.
13
Computers & Industrial Engineering, 173, 108645.
14
Santini, A., Viana, A., Klimentova, X., & Pedroso, J. P. (2022). The probabilistic travelling
15
salesman problem with crowdsourcing. Computers & Operations Research, 142,
16
105722.
17
Savelsbergh, M., & Van Woensel, T. (2016). 50th anniversary invited article—city logistics:
18
Challenges and opportunities. Transportation Science, 50(2), 579-590.
19
Shen, H., & Lin, J. (2020). Investigation of crowdshipping delivery trip production with real-
20
world data. Transportation Research Part E: Logistics and Transportation Review, 143,
21
102106.
22
Silva, M., & Pedroso, J. P. (2022). Deep reinforcement learning for crowdshipping last-mile
23
delivery with endogenous uncertainty. Mathematics, 10(20), 3902.
24
Sun, X., Hu, W., Xue, X., & Dong, J. (2022). Multi-objective optimization model for planning
25
metro-based underground logistics system network: Nanjing case study. Journal of
26
Industrial and Management Optimization, 19(1), 170-196.
27

46
Tholen, M. V. D., Beirigo, B. A., Jovanova, J., & Schulte, F. (2021, September). The share-a-
1
ride problem with integrated routing and design decisions: the case of mixed-purpose
2
shared autonomous vehicles. In International Conference on Computational Logistics
3
(pp. 347-361). Springer, Cham.
4
Triki, C. (2021). Using combinatorial auctions for the procurement of occasional drivers in the
5
freight transportation: A case study. Journal of Cleaner Production, 304, 127057.
6
Voigt, S., & Kuhn, H. (2022). Crowdsourced logistics: The pickup and delivery problem with
7
transshipments and occasional drivers. Networks, 79(3), 403-426.
8
Wang, Y., Zhang, D., Liu, Q., Shen, F., & Lee, L. H. (2016). Towards enhancing the last-mile
9
delivery: An effective crowd-tasking model with scalable solutions. Transportation
10
Research Part E: Logistics and Transportation Review, 93, 279–293.
11
Xiao, F., Wang, H., Guo, S., Guan, X., & Liu, B. (2021). Efficient and truthful multi-attribute
12
auctions for crowdsourced delivery. International Journal of Production Economics,
13
240, 108233.
14
Ye, Y., Guo, J., & Yan, L. (2021). A mixed decision strategy for freight and passenger
15
transportation in metro systems. Computational intelligence and neuroscience, 2021, 1-
16
22.
17
Yildiz, B. (2021a). Express package routing problem with occasional couriers. Transportation
18
Research Part C: Emerging Technologies, 123, 102994.
19
Yildiz, B. (2021b). Package routing problem with registered couriers and stochastic demand.
20
Transportation Research Part E: Logistics and Transportation Review, 147, 102248.
21
Yu, V. F., Indrakarna, P. A., Redi, A. A. N. P., & Lin, S. W. (2021a). Simulated annealing with
22
mutation strategy for the share-a-ride problem with flexible compartments.
23
Mathematics, 9(18), 2320.
24
Yu, V. F., Purwanti, S. S., Redi, A. P., Lu, C. C., Suprayogi, S., & Jewpanya, P. (2018).
25
Simulated annealing heuristic for the general share-a-ride problem. Engineering
26
Optimization, 50(7), 1178-1197.
27

47
Yu, V. F., Jodiawan, P., Hou, M. L., & Gunawan, A. (2021b). Design of a two-echelon freight
1
distribution system in last-mile logistics considering covering locations and occasional
2
drivers. Transportation Research Part E: Logistics and Transportation Review, 154,
3
102461.
4
Yu, V. F., Jodiawan, P., & Redi, A. P. (2022a). Crowd-shipping problem with time windows,
5
transhipment nodes, and delivery options. Transportation Research Part E: Logistics
6
and Transportation Review, 157, 102545.
7
Yu, V. F. V., Zegeye, M. M., Gebeyehu, S. G., Indrakarna, P. A., & Jodiawan, P. (2022b). The
8
multi-depot general share-a-ride problem. Expert Systems with Applications, 213,
9
119044.
10
Zehtabian, S., Larsen, C., & Wøhlk, S. (2022). Estimation of the arrival time of deliveries by
11
occasional drivers in a crowd-shipping setting. European Journal of Operational
12
Research, 303(2), 616-632.
13
Zhang, S., Markos, C., & James, J. Q. (2022). Autonomous vehicle intelligent system: joint
14
ride-sharing and parcel delivery strategy. IEEE Transactions on Intelligent
15
Transportation Systems, 23(10), 18466-18477.
16
Zhao, L., Li, H., Li, M., Sun, Y., Hu, Q., Mao, S., & Xue, J. (2018). Location selection of intra-
17
city distribution hubs in the metro-integrated logistics system. Tunnelling and
18
Underground Space Technology, 80, 246-256.
19
Zhou, Z., Chen, R., & Guo, S. (2021). A domain‐of‐influence based pricing strategy for task
20
assignment in crowd-sourcing package delivery. IET Intelligent Transport Systems,
21
15(6), 808-823.
22
23
Table 1. Forms of integrated transportation systems in the literature.
24
Table 2. Application of integrated people-and-goods transportation.
25
Figure 1. Technical problems that have been studied for different integration forms.
26

48
Figure 2. A general framework for integrated people-and-goods transportation systems.
1