Archives of Computational Methods in Engineering

https://doi.org/10.1007/s11831-022-09874-w

SURVEY ARTICLE

Emerging Research Fields in Vehicle Routing Problem: A Short Review

M. Salehi Sarbijan1 · J. Behnamian1

Received: 30 April 2021 / Accepted: 15 December 2022

© The Author(s) under exclusive licence to International Center for Numerical Methods in Engineering (CIMNE) 2022

Abstract
Vehicle routing problems (VRPs) have been the subject of intense and fast-growing research for the last six decades. These
problems have received much attention in combinatorial optimization, especially in operations research and transportation.
This paper aims to systematically review and analyze recent research on the VRP by covering 89 papers between 2001 and
2022. The focus of this study is on the new and emerging topics existing in the VRP, including feeder VRP (FVRP), real-
time VRP (RTVRP), and collaborative VRP (CVRP) for urban transportation. To this end, recent findings are reviewed to
recognize novel research gaps for the problem. Based on reviews, we suggest applying multi-agent solving approaches and
novel metaheuristic algorithms in the FVRP. Furthermore, studying the real-time methodology in various routing problems
such as electric VRP (EVRP), two-echelon VRP (2E-VRP), rich VRP (RVRP), agro-food supply chains, biomass supply
chain, and healthcare supply chains are other suggestions of this research. Finally, integrating new topics in the CVRP (e.g.,
blockchain, big data, and data mining) are among the research gaps identified in this review.

1 Introduction and environmental aspects are some of these developments

for the VRP in recent years. Regarding the limited capacity
Vehicle routing problem (VRP) is among the most exten- of the vehicle in VRP, it must return to the central depot
sive problems in operations research. Danzig and Ramser at specific intervals for reloading. As a result, if the total
[1] proposed for the first time the optimum routing of a fleet customer demand exceeds the truck capacity, the number
of gasoline delivery trucks between a bulk terminal and a of return trips and the corresponding travel costs would be
large number of service stations supplied by the terminal. increased. Two major strategies are considered for reducing
The VRP consists of a fleet of limited-capacity vehicles, the total travel costs spent for reloading: (1) building larger
some customers in different locations, and a central depot. depots in the proximity of the demand market and (2) dis-
VRP aims to identify the best routes for vehicles to deliver patching larger vehicles with greater capacity. The first strat-
services to all customers with minimum costs (i.e., travel egy is only possible when the land price, construction costs,
time and travel distance). So far, researchers have developed and depot operation costs are low enough. This strategy is
more complex mathematical models that are more realistic difficult to implement, as land for building a new depot is
and operational. In addition, they have introduced various scarce and expensive, especially when the depot is situated
problem-solving approaches along with the advances in the around highly developed urban areas. The second strategy
world of technology and the processing power of computer seems permissible but may face problems, particularly when
systems. Considering the capacity limitation for the fleet the streets are very narrow or when large vehicles are not
of vehicles, multiple depots for the distribution of freight, allowed to enter the streets for safety concerns. Access limi-
time windows for visiting customers, heterogeneity of tations are usually observed in different urban areas. To deal
the vehicle fleet, the simultaneous and non-simultaneous with the problems in these two strategies, researchers have
pickup and delivery of freight, the delivery of several parts come up with a third approach, i.e., feeder vehicle routing
of freight to customers, inventory decisions, depot location, problem (FVRP), to reduce the total cost of reloading [2].
In the urban transportation distribution based on the classic
VRP problem, when the customer demand is slightly more
* J. Behnamian than the capacity of one vehicle, another vehicle is required.
[email protected]
As a result, using an extra vehicle increases the costs. Unlike
1
Department of Industrial Engineering, Faculty VRP, in the FVRP, small vehicles with low capacity can be
of Engineering, Bu-Ali Sina University, Hamedan, Iran

13
Vol.:(0123456789)
 M. Salehi Sarbijan, J. Behnamian

used as a second vehicle, which involves lower costs and alliances benefits suppliers and customers by reducing
saves money. In addition, the traffic congestion in urban costs and distances traveled. Thus, the collaborative VRP
areas and narrow roads block access to some customers has been developed as a new problem based on collabora-
with large vehicles. However, in FVRP, small vehicles allow tion between distribution centers to share vehicle capacity
reaching customers, increasing their satisfaction level, and and customer demands.
lowering traffic and pollution. Also, using small vehicles in The main contributions of this review are as follows.
FVRP leads to improved routes, eliminating unnecessary First, this survey is a systematic, up-to-date review of the
trips, and creating alternative short routes. classification of existing studies on the three variants of
In the static version of VRP, all route information (i.e., the VRP, emphasizing the recent literature. Second, the
traffic, customers, and vehicles’ data) is already known and papers between 2001 and 2022 are reviewed to identify
does not change along the route. Nevertheless, in several research gaps and introduce the latest topics and upcom-
real-world cases, route information involves some degree ing trends in the subject of vehicle routing, including feeder
of uncertainty and may vary dynamically along the route. VRP (FVRP), real-time VRP (RTVRP), and collaborative
Therefore, the optimization process must be adjusted (CVRP) for urban transportation. Third, to introduce future
dynamically. A real-time VRP (RT-VRP) is a dynamic VRP research opportunities, we tried to motivate researchers to
in which demands are produced dynamically over the time move purposefully by focusing on new topics.
horizon of operations without prior knowledge. In this prob- The remainder of our survey is organized as follows.
lem, the received orders should be responded to quickly and After introducing the theoretical background in Sect. 1, the
then dispatched to a service vehicle. RTVRPs play a pivotal research methodology and classifications are described in
role in urban distribution areas as unexpected events (e.g., Sect. 2. The definition and literature review are provided in
changes in customer request (increased request or eliminated Sect. 3. Section 4 discusses the analysis and interpretation of
order), changes in the time window (progress or delay), road research literature. Section 5 presents future research on this
blockade due to accidents or traffic congestion, and vehicle work and gives a brief outlook. Finally, Sect. 6 concludes
breakdowns) may disrupt the distribution plan. In RTVRP, this research.
initial planning is done for the previous customers by solving
a static VRP. However, in this problem, vehicles whose lim-
ited capacity has not yet expired will not return to the depot.
In the next step, the received new orders must be planned. In 2 Methodology and Classification
this case, the final location of the vehicle in the initial plan
acts as a secondary depot. In other words, secondary plan- There are many review papers on VRPs. For example, Vidal
ning is done from the end location where each vehicle has et al. [5] presented an overview of existing and emerging
traveled. It is noteworthy that if a customer’s order exceeds VRPs, and Koç et al. [6] reviewed simultaneous pickup and
the capacity of the vehicle, the customer must wait until the delivery for the VRPs. In the field of dynamic VRP and
vehicle returns to the depot the next time. RTVRP, Pillac et al. [7] reviewed dynamic vehicle rout-
In the classical distribution scenario, suppliers offer ser- ing problems, Abbatecola et al. [8] and Adacher [9] sur-
vices to their activity area’s customers by creating their veyed sources of dynamism and optimization heuristics for
distribution network centers. In such a scenario, distribu- dynamic VRPs, respectively. Ghiani et al. [10] investigated
tors distribute their freight independently by themselves. solution concepts, algorithms, and parallel computing strate-
Therefore, sharing logistics resources is not an easy task. gies for real-time VRP (RTVRP). Also, for CVRP, Guajardo
Moreover, such a distribution leads to an imbalance in and Rönnqvist [11] considered cost allocation methods in
the capacity of the vehicles’, freight distribution demand collaborative transportation, Gansterer and Hartl [12] sur-
in a given time and location, and issues such as drivers’ veyed operations research models and solution techniques in
unemployment and empty load routes [3, 4]. Hence, it is CVRP, Cleophas et al. [13] examined theory and practice in
vital to find a way to lower transportation and distribution collaborative urban transportation, and Pan et al. [14] stud-
expenses and enhance the satisfaction level of custom- ied horizontal collaborative transport.
ers. To this end, companies must analyze new ideas or
adjust their logistics plans to reduce costs and delivery
time unless they would not stay competitive. Other than 2.1 Systematic Review Protocol
the city logistic challenges, VRP has found new research
areas. In the meantime, several transport companies are According to [15], a review study is conducted to identify,
trying to optimize transport operations by establishing evaluate, and interpret research in a particular area of inter-
alliances to share the capacity of vehicles and customer est [16]. Figure 1 indicates the flowchart explaining the pro-
demands. Practical analyses have also proven that forming posed methodology.

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

Fig. 1  Flow chart demonstrating the systematic review methodology

2.2 Research Questions titles of papers in two databases of the web of science

website published from 2001 to 2022. Accordingly, sev-
In this study, three special research questions were formu- eral papers were found on the VRP. The second step was
lated to accomplish emerging research fields in the VRP: (i) using the “feeder vehicle routing problem” in the title of
What are the models and algorithms used in FVRP, RTVRP, the papers. As a result, 47 papers were chosen by search-
and CVRP? (ii) What are the most emerging research chal- ing over different sources. In the third step, the “real-time
lenges and research fields in VRP, RTVRP, and CVRP in vehicle routing problem” was searched, which yielded
urban transportation? (iii) What are the research directions 59 papers. Finally, by adding further words such as “col-
in the VRP? laboration”, “collaborative”, “cooperation”, and “horizon-
tal cooperation” to a vehicle routing problem, 82 papers
2.3 Search Strategy and Selection Criteria were selected for the reviewing process. Since routing
was not used in some papers, such papers were excluded.
Initially, relevant papers were retrieved using an advanced As a result, 12 papers directly related to the FVRP were
search of “vehicle routing problem” as the keyword in the selected. Also, considering the source of dynamics in

13
 M. Salehi Sarbijan, J. Behnamian

VRP, 38 papers were selected from 59 papers. Besides, 3 Literature Review

some duplicate and unrelated studies on the CVRP were
discarded. Finally, 39 papers were selected. This section considers the definitions and research literature
In this study, about 89 papers about FVRPs, RTVRPs, of the three classes of VRP, including FVRP, RTVRP, and
and CVRPs in urban transportation were surveyed. Fig- CVRP.
ure 2 provides an overview of the number of published
papers per year between 2001 and 2022. 3.1 The Feeder Vehicle Routing Problem

The FVRP consists of a physical depot, a group of customers

2.4 Classification with different characteristics in terms of parking space, and a
heterogeneous fleet of vehicles. The heterogeneous fleet uses
In this study, the papers on VRP were classified into three two classes of small and large vehicles, which are dispatched
main types: FVRP, RTVRP, and CVRP. In Fig. 3, the from the physical depot. The small vehicle class can be rep-
research literature topics are summarized. resented as a vehicle, a motorcycle, or a bicycle. Compared
to the small vehicle class, the large vehicle class rarely has

Fig. 2  The number of published 8 8 8

papers per year 7
6 6
5 5 5
4 4 4 4
3 3 3
2
1 1 1 1
0

Fig. 3  Summarize the classification of the problem

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

a capacity limit (e.g., truck or trailer). In the case of custom- A small empty vehicle (such as a motorcycle or a bicy-
ers, they are classified according to the parking space avail- cle) may deviate from its assigned route to reload from
ability. Some customers (type 1) provide sufficient space a large vehicle (such as a truck) at a virtual depot (VD)
and are visitable via both vehicle classes. However, most nearby or a physical depot (PD). Chang [19] further
customers (type 2) live in crowded and hardly reachable extended this research and renamed the LFVRP. Inflex-
areas and only use the small vehicle class. Consider a graph ibility caused by pre-specified VDs is a well-documented
such as G = (N, A), where N shows the nodes and A shows problem with LFVRP and LFVRP-PVD. Chen et al. [20]
the arcs. Set N includes a depot and the remaining nodes of solved this problem by introducing the LFVRP virtual
the customer set. The beginning and end of the routes for depot (LFVRP-VD) as an extension of the LFVRP-PVD.
different vehicles in the FVRP are defined from the depot. Similarly, Chen et al. [21] presented an LFVRP with VD
All customers must be serviced and visited by at least one and time windows (LFVRP-VDTW). Elsewhere, Chen
vehicle (truck or motorcycle). Figure 4 shows an example and Wang [22] introduced a generalized problem of the
of an FVRP with a truck (bold line) and a motorcycle (dot- extended LFVRPTW (ELFVRPTW). In another effort,
ted line). Both vehicles exit from a depot and, for reloading, Chen [23] investigated four cases of reloading services,
meet each other at three feeder and joint nodes (points a, b, small vehicle feeding, time window (TW), and the num-
and c), and return to the depot after providing services to ber of VD candidate point constraints on the LFVRPTW
all customers. for 15 instances. Brandstätter and Reimann [24] studied
Here, the literature on the FVRP is classified based on the a new extension of the LFVRP. This problem is a VRP
problem features and solution approaches. with synchronization constraints in which two customer
types are visited by two vehicle types. Huang et al. [2]
3.1.1 Problem Features introduced an FVRP in which each small (motorcycle) or
large vehicle (truck) visits the customer. In this delivery
Tu et al. [17] and Chang et al. [18] were the first who type, motorcycles and trucks must set off the depot, visit
proposed the Line-haul feeder vehicle routing problem the customers, and finally return to the depot. Recently,
with a pre-specified virtual depot (LFVRP-PVD) for Salehi Sarbijan and Behnamian [24] modified the FVRP
lunchbox delivery in Taiwan. LFVRP-VPD is established into a minimum of two motorcycles and trucks, remaining
based on two vehicle types (namely, small and large) to the pair of motorcycles and trucks the same throughout the
serve two various consumer classes (i.e., types I and II). delivery process.

Fig. 4  An example of an FVRP

route with a truck and motor-
cycle [2]

13
 M. Salehi Sarbijan, J. Behnamian

3.1.2 Solution Approaches develop LFVRP TW: local search, multiple solution genera-
tion, metaheuristic, and mathematical model-based heuristic.
The FVRP studies can be examined using some solutions. Furthermore, Huang et al. [2] applied the ant colony opti-
In this regard, Chang [19] offered a two-phase heuristic for mization (ACO) algorithm to solve the FVRP. Also, Salehi
the LFVRP. In the first phase, initial routes are built using Sarbijan and Behnamian [27] compared the hybrid PSO-SA
the nearest neighbor search (NNS). On the other hand, the with PSO, VNS, and ACO in small, medium, and large-sized
initial solution is improved in the second phase using the instances for the Multi-Fleet FVRP. All the papers of FVRP
node exchange technique. Chen et al. [20] offered threshold are shown in Table 1.
methods and cost-sharing as two heuristics for LFVRP-VD
to minimize waiting costs and vehicle travel time. Further- 3.2 The Real‑Time Vehicle Routing Problem
more, Chen et al. [21] and Chen [23] employed a two-stage
solution heuristic for solving the LFVRPTW problem. RTVRPs play a critical role in urban distribution because
This problem includes a Tabu search (TS) and a modified unexpected events such as changes in customer demand,
sequential insertion heuristic. Brandstätter and Reimann changes in the delivery time window, road blockades due
[24] presented linkage and split as two heuristic techniques to accidents or heavy traffic, and vehicle breakdowns con-
to obtain near-optimal solutions to the LFVRP. This study stantly lead to disruption of the distribution program. To
was extended by Brandstätter and Reimann [25] by propos- address these issues, planners need to fix the disruption of
ing two techniques to improve splitting and linkage in the vehicle routes in real-time, thereby improving improve vehi-
LFVRP. Brandstätter [26] applied the following methods to cle operation efficiency and increasing service quality. The
RTVRP is a dynamic routing problem in which requests
are generated dynamically over the time horizon of opera-
Table 1  Literature summary of FVRP tions without prior knowledge. Received requests should be
Reference Problem Method responded to as quickly as possible and then assigned to a
vehicle for service providing. Figure 5 presents the imple-
[17] LFVRP-PVD Simulation
mentation of a dynamic VRP. Before the vehicle leaves
[18] LFVRP Heuristic
the depot (at t0 ), an initial route planning to meet customer
[19] LFVRP Heuristic
demand includes ABCD, i.e., the starting point for plan-
[20] LFVRP-VD Heuristic
ning. When the vehicle performs its route, a new request
[21] LFVRP-VDTW Metaheuristic
X appears at tr , and the initial route is reconfigured to meet
[22] ELFVRPTW Heuristic
new demand X. In this case, the total route is divided into
[23] LFVRPTW Heuristic
three parts: (i) the part of the route that has already been
[24] LFVRP Metaheuristic
performed and is not changeable, (ii) the current position of
[25] LFVRP Heuristic
the vehicle to reach the next customer, and (iii) the rest of
[26] FVRPTW Heuristic
the route, which will probably be implemented in the future
[2] FVRP Metaheuristic
and is changeable. Finally, the implemented route is defined
[27] Multi-Fleet FVRP Metaheuristic
as ABCXD at tT.

Fig. 5  An example of dynamic VRP (DVRP) [28]

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

The research literature on the problem of RTVRP is clas- while implementing a routing scheme. Naoum-Sawaya and
sified based on the source of existing dynamics, including Elhedhli [43] offered a two-phase stochastic optimization
changes in customer demand, travel times, disruption and model for the ambulance relocation problem. In another
traffic in movement, and the combination of dynamics. paper, Jagtenberg et al. [44] considered the problem of
dynamic ambulance relocation, where a polynomial-time
3.2.1 Changes in Customer Request heuristic was proposed that individuates itself by needing
neither assumption. De Armas and Melian-Batista [45] pre-
Yang et al. [29] considered a general real-time truckload sented a variable neighborhood search (VNS) algorithm for
pickup and delivery problem. This problem includes vari- solving the dynamic rich vehicle routing problem with time
ous costs related to job rejection, empty travel of trucks, windows. Euchi et al. [46] proposed an ant colony optimi-
and time delay in job completions. In this study, the authors zation based on a 2opt local search (AAC-2opt) to solve a
compared five rolling horizon approaches for the RTVRP, DVRP with pickup and delivery. Mandziuk and Zychowski
which is solved whenever a new request arrives. Monte- [47] established a memetic algorithm for 21 DVRP bench-
manni et al. [30] offered an ant colony system (ACS) algo- mark instances. Sarasola et al. [48] offered a VNS algorithm
rithm for solving the DVRPs. These researchers validated for the DVRP with stochastic demands. Abdallah et al. [49]
the efficiency of this algorithm in solving DVRPs through solved a DVRP using the GA. Okulewicz and Mandziuk [50]
empirical data. Potvin et al. [31] studied a dynamic vehi- offered a two-stage multi-SPO (2MPSO) to solve the DVRP.
cle routing and scheduling problem with time windows and According to their empirical results, the algorithm outper-
proposed real-time customer requests and dynamic travel formed several similar PSO algorithms. Chen et al. [51] pre-
times. Du et al. [32] considered a real-time vehicle-dis- sented an ALNS for solving a DVRP in which the routing
patching system for milk runs. These researchers applied could vary using real-time data. The numerical outcomes
seven modules to the real-time system. Jemai and Mellouli revealed that this ALNS could quickly solve the problem and
[33] proposed a neural tabu search algorithm solving for provide a high-quality solution. Okulewicz and Mandziuk
the RTVRP. Hsueh et al. [34] considered the DVRP with [52] developed the PSO and DE metaheuristic algorithms for
pickup and delivery demands and travel times. These authors 21 data sets of DVRPs in continuous search space (CSS) and
aimed to minimize travel time and the total penalty due to advanced discrete coding using genetic algorithms (GAs).
late arrivals. De oliveira et al. [35] studied architectures for In another study, Archetti et al. [53] studied the VNS algo-
solving the dynamic VRP with time windows and used the rithm for the online VRP with occasional drivers. Guiqin
ant colony system metaheuristic for building the routes of et al. [54] offered a hybrid genetic algorithm-tabu search
the vehicles. Okhrin and Richter [36] proposed a VRP by (GA-TS) algorithm for a two-echelon DVRP with proactive
considering real-time variations in travel times with time satellite stations (2E-DVRP-PSSs). Moreover, Ge and Jin
windows. The paper formulated a mixed-integer linear pro- [55] provided a novel hybrid simulated annealing-GA (SA-
gramming (MILP) model and extended a genetic algorithm GA) for DVRPs.
(GA) to solve it. Liao and Hu [37] presented a two-phase
approach for solving the dynamic VRP. In the first phase, 3.2.2 Travel time
the sweep procedure was used in the vehicle assignment.
In the second phase, a tabu search algorithm (TSA) was Fu [56] considered the problem of routing in which travel
accomplished to improve routes under real-time information. time on each link can be modeled as a random variable. The
Khouadjia et al. [38] studied a VRP with dynamic requests study formulated the closed-loop adaptive shortest path rout-
and proposed some solving methods based on the variable ing problem. Chang et al. [57] proposed the RTVRP with
neighborhood search (VNS) algorithm and dynamic adapted time windows and synchronous delivery/pickup demands.
particle swarm optimization. Wohlgemuth et al. [39] pre- In this study, a MILP model is considered that is frequently
sented dynamic pickup and delivery requests that may both solved in the rolling time horizon. Haghani and Jung [58]
obtain and dispatch goods and modify travel times. These proposed a hybrid genetic algorithm to solve the DVRP with
researchers offered a multi-phase MILP capable of working time-dependent travel times. Ferrucci et al. [59] proposed a
under variable demands. Hong [40] considered the dynamic novel pro-active real-time control approach for DVRPs with
VRP with hard time windows. The advanced large neigh- a soft time window.
borhood search algorithm was proposed to solve the static
problem. Furthermore, Barkaoui and Gendreau [41] studied 3.2.3 Traffic Congestion
dynamic routing problem with time windows and presented
an adaptive evolutionary approach. Woensel et al. [60] proposed a tabu search to solve the
Hu et al. [42] proposed modeling based on a knowledge capacitated VRP with time-dependent travel times due to
approach that integrates policies for handling disconnection traffic congestion using a stochastic procedure. This study

13
 M. Salehi Sarbijan, J. Behnamian

defined travel speed, approximations based on queuing increase vehicle use and reduce empty routes. This type of
theory, and the mass of vehicles in a link. Zhao et al. [61] collaboration is called horizontal cooperation. The second
considered the DVRP by time windows with real-time traf- category is based on the extent the companies share their
fic information, travel time, and travel speed. Li et al. [62] information and equipment. This category includes meth-
developed a Lagrangian relaxation algorithm for the vehi- ods such as allocating demand and capacity to companies.
cle rescheduling problem. Also, Li et al. [63] studied the The motivation and purpose of papers reveal another type of
real-time vehicle rerouting problem with time windows and categorization that can be considered. Collaboration in trans-
proposed a Lagrangian relaxation algorithm for the primal portation can be categorized in terms of the type of partner
problem. companies. In this respect, some studies have also examined
the impact of collaborative transportation on reducing envi-
3.2.4 Combining Dynamic Resources ronmental pollution.

Chen et al. [64] presented vehicle routing with time-depend- 3.3.1 Horizontal Cooperation
ent travel, real-time, and time windows. In another study,
Grzybowska and Barceló [65] offered a decision support Cruijssen et al. [67] considered the concept of joint route
system for real-time freight management. This system was planning wherein out-wide experiences are used to measure
designed by integrating the chosen VRP with pickup and the association of the cost savings on several specifications
delivery (VRPPD) and dynamic traffic simulation mod- of the VRP with time window. Liu et al. [68] considered
els. Messaoud and Alaoui [66] studied the RT-VRP with a two-stage greedy algorithm for solving real larger-scale
dynamic customers and traffic factors information. To solve multi-depot capacitated VRPs. In the first stage, a collec-
this problem, they also employed a new strategy based on tion of directed cycles is built to fulfill the transportation
the compound genetic algorithm. The papers based on com- orders. In the second stage, chains that are made of cycles
bining dynamic resources are summarized in Table 2. are produced. Nadarajah and Bookbinder [69] proposed a
two-phase procedure for less-than-truckload carrier col-
3.3 Collaborative Vehicle Routing Problem (CVRP) laboration. The first phase considers collaboration between
multiple carriers at the city’s entrance. The second phase
In recent years, horizontal collaboration in the road transpor- considered collaboration between carriers at transshipment
tation industry has become increasingly common. The main conveniences to conclude the initial routes. Stenger et al.
idea of this model is that distribution companies can signifi- [70] studied multi-depot VRP (MDVRP) with particular
cantly increase the efficiency of their operations by joining shipping. These researchers used VNS as an improvement
a coalition and partnering in a horizontal logistics coalition. that embedded cyclic exchange neighborhoods and an adap-
This efficiency can be enhanced by solving the CVRP, i.e., tive algorithm. Adenso-Díaz et al. [71] presented an opti-
the VRP in which customers are routinely visited by differ- mization approach for addressing backhauling routes and
ent transport companies and assigned to common vehicle solved it using the GRASP algorithm. Muñoz-Villamizar
routes. In collaborative logistics, several transport compa- et al. [72] studied a method based on mathematical mod-
nies form some coalitions by sharing their vehicle capacity eling for horizontal collaboration in the urban distribution
and demand to optimize transport operations. In traditional of goods with stochastic demand. In this paper, coopera-
VRP, transportation companies optimize their vehicle routes tive and non-cooperative scenarios were compared. Kimms
individually. In this case, the efficiency is reduced, and a and Kozeletskyi [73] investigated a cost allocation outline
small profit will be achievable. Thus, researchers are increas- for horizontal cooperation of traveling salesmen problem
ingly looking for optimization opportunities outside the using multi-depots and time-limited restrictions. A method
traditional realm of optimization. One promising research based on the concept was determined for the cost allocation.
direction is to optimize transport companies’ operational In another paper, Kimms and Kozeletskyi [74] examined
activities jointly. Consider a simple example of a company to cooperation in traveling salesman problem with the rolling
understand the potential of horizontal collaboration by con- horizon. They applied stochastic dynamic programming to
sidering a truck with a full capacity of goods moving from determine a route.
A to B. Then, instead of returning empty, it gathers another Yang et al. [75] considered the impacts of collabora-
company’s goods and transfers them from B to A. Collabora- tive distribution between two logistics service providers
tive transportation is a relatively new field in transportation on delivery cost and time. Fernández et al. [76] presented
and logistics studies. The research literature in this field can the shared customer collaboration VRP (SCC-VRP). In
be categorized from several general perspectives. In the first this problem, some customers needed services from more
category, several carrier companies may form a union by than one carrier, and many carriers acted. These carriers
sharing their transportation requests and vehicle capacity to transferred a part of their demand to their union partners.

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

Table 2  Literature summary of RTVRP

Reference Problem Source of dynamism Method

[29] General real-time truckload pickup and delivery Customer request and travel time Simulation
[30] DVRP Customer request Metaheuristic
[31] RTVRP and scheduling problem Customer request and travel time Metaheuristic
[32] RTVRP dispatching in Milk runs Customer request Simulation, Heuristic
[33] RTVRP Customer request Metaheuristic
[34] RTVRP in natural disasters Travel time Metaheuristic
[35] RTVRPTW Customer request Metaheuristic
[36] VRP with real-time travel times Customer request and travel time Metaheuristic
[37] DVRP Customer request and travel time Metaheuristic
[38] DVRP Customer request Metaheuristic
[39] DVRP in disaster relief Customer request Simulation
[40] DVRP with hard time windows. Customer request Metaheuristic
[41] RTVRP and dispatching Customer request Metaheuristic
[42] RTVRP Customer request Exact
[43] RT ambulance redeployment Customer request Simulation
[45] Dynamic rich VRP Customer request Metaheuristic
[44] RT ambulance redeployment Customer request Metaheuristic
[46] DVRP with pickup and delivery Customer request Metaheuristic
[47] DVRP Customer request Metaheuristic
[48] DVRP Customer request Metaheuristic
[49] DVRP Customer request Metaheuristic
[50] DVRP Customer request Metaheuristic
[51] DVRP Customer request Metaheuristic
[52] DVRP Customer request Metaheuristic
[53] Online VRP Customer request Metaheuristic
[54] Two-echelon DVRP Customer request Metaheuristic
[55] DVRP Customer request Metaheuristic
[56] RTVRP Travel time Exact
[57] RTVRPTWDP Travel time Metaheuristic
[58] DVRP with time-dependent travel times Travel time Metaheuristic
[59] DVRP for urgent goods Travel time Metaheuristic
[60] Capacitated VRP with time-dependent travel times Traffic congestion, travel time Heuristic, Simulation
[61] DVRPTW Traffic congestion Simulation
[62] RTVSRP Traffic congestion Metaheuristic
[63] RTVRPTWPD Traffic congestion Metaheuristic
[64] RTTDVRPTW Time-dependent, travel time, customer request Metaheuristic
[65] RTVRP Traffic congestion, travel time, customer request Heuristic, Simulation
[66] RTVRP Traffic congestion, travel time, customer request Metaheuristic

The paper proposed a mathematical programming model 3.3.2 Centralized Collaboration and Revenue Sharing
for the problem and solved it using a branch-and-cut algo-
rithm. Vaziri et al. [77] presented a VRP impartial carrier Dai and Chen [78] studied a carrier collaboration problem
collaboration with multi-commodity demand allocations with pickup and delivery under a centralized collaboration
and multi pickup and delivery services. They developed approach. These authors considered three benefit alloca-
a MILP model to maximize the total benefit. Moreover, tion mechanisms, including benefit allocation by the Shap-
they proposed a GA to solve the problem and developed ley value, the adequate allocation notion, and an alloca-
the VNS algorithm for better solutions. tion mechanism for the quota of every single carrier to the

13
 M. Salehi Sarbijan, J. Behnamian

integration. Zakharov and Shchegryaev [79] investigated combined the k-means clustering algorithm and an improved
resistant cooperation in DVRPs on large-scale transport NSGA-II (Im-NSGA-II). Defryn and Sörensen [91] stud-
networks using the Shapley value. The authors proposed ied horizontal cooperation and traveling salesman problem
a heuristic repetitious algorithm to build the characteristic with soft time windows. To this end, they employed two
subadditive of the dynamic routing game. Defryn et al. [80] approaches: the partner efficiency approach and the coali-
studied a selective multiple VRP in which customers depend tion efficiency approach. Wang et al. [92] considered a col-
on various partners. This study combined the allocation cost laborative 2E-MCVRP based on a state-space-time using
according to the Shapley value and collaborative routing. dynamic programming, k-means clustering, and Im-NSGA-
II. Wang et al. [93] studied a collaborative MDVRP with a
3.3.3 Demand Allocation and Capacity Method time window assignment. This study used a combination
of k-means clustering, the Clarke-Wright saving algorithm,
Krajewska et al. [81] investigated the VRP with multi-depot and an extended NSGA-II (E-NSGA-II). Zhang et al. [94]
pickup and delivery with time windows in a cooperative proposed a multi-objective optimization for CVRP by con-
condition using the value of Shapley and solved by an adap- sidering carriers and depot sharing. They also offered an
tive large neighborhood search heuristic. Dahl and Derigs extended variable neighborhood search algorithm based on
[82] formulated an experimental analysis on the suitability three matrices: carrier collaboration, depot collaboration,
of a collaborative decision support system for a peculiar and transportation sequence. Wang et al. [95] introduced
express carrier network. In this study, each carrier had a the K-means genetic hybrid algorithm and clustering for
set of vehicles with different characteristics, requirements, collaborative multi-depot pickup and delivery logistics
and time windows. Li [83] studied the CVRP with pickup networks with (CMDPDLN) split time windows and loads.
and delivery and multiple distribution centers. The three- Wang et al. [96] studied the collaborative mechanism for
objective modeling of this study included minimizing the the two-echelon collaborative multi-depot multi-period VRP
number of employed vehicles, the travel distance, and the (2E-CMDPVRP) with resource sharing.
standard deviation. This study proposed a TSA with the
Lagrangian relaxation method. In another paper, Vornhusen 3.3.5 Shipper Collaboration
et al. [84] proposed a method to integrate transshipment into
the CoVRP with pickup and delivery, and presented a MILP Ergun et al. [97] studied the shipper collaboration prob-
model for the problem. lem by considering the time and driver constraints. They
applied the greedy algorithm to solve the problem. Ergun
3.3.4 Coalition type and Capacity Sharing et al. [98] introduced the truckload shipper collaboration
problem. Özener and Ergun [99] examined the shipper col-
Hernández et al. [85] presented the deterministic dynamic laboration problem by presenting and comparing the propor-
single carrier collaboration problem for road transportation. tional method and two new methods. Yilmaz and Savasaneril
They modeled the problem as a binary (0–1) multi-commod- [100] presented the collaboration between small shippers in
ity and solved it by applying a branch and cut algorithm. a transportation market by uncertain requests. In this study,
Sprenger and Mönch [86] considered a collaborative trans- two naive and myopic mechanisms were used. Sun et al.
portation planning problem with simulation for the German [101] proposed a contribution constrained packing model
food industry. They applied the powerful ant colony opti- for the shipper collaboration problem. They presented fair-
mization (ACO) algorithm to improve the greedy heuristic ness criteria to measure variant cost allocation procedures.
algorithm. Dai et al.[87] studied the multiple-carrier col-
laboration problem with pickup and delivery requirements 3.3.6 Cooperation and Environmental Problems
and proposed a multi-round based on the price-setting
and Lagrangian relaxation approach. Wang et al. [88] pre- Pradenas et al. [102] studied a mathematical model for
sented the MCVRP based on vehicle routing optimization CVRP with backhauls, time windows, and environmental
methods and profit distribution methods. These research- criteria. Juan et al. [103] presented horizontal cooperation
ers proposed a multi-stage linked approach with clustering, in road transportation with ­CO2 emissions and backhaul-
dynamic programming, and a heuristic algorithm. Wang ing. These authors considered both cooperative and non-
et al. [89] presented a CMCVRP with a simultaneous pickup cooperative approaches to reduce transportation distance
and delivery approach. They combined the k-means clus- and environmental effects. Perez-Bernabeu et al. [104]
tering algorithm with the non-dominated sorting genetic proposed horizontal cooperation in road transportation
algorithm-II (NSGA-II) to find optimal paths for coalition with different VRP scenarios for costs and greenhouse gas
coordinators. Wang et al. [90] considered the two-echelon emissions. Munoz-Villamizar et al. [105] applied multi-
collaborative multiple-center VRP (2E-CMCVRP). They objective mathematical programming optimization to the

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

effect of the use of electric vehicles in collaborative urban 4 Analysis of the Literature
transportation. These researchers proposed a two-stage
heuristic approach, including allocating delivery points Figure 6 shows an overview of the number of publications
to network members and determining the optimal routes. conducted in each application field during the assessment
All these papers are summarized in Table 3. period (2001–2022). In recent years, VRP research in coop-
eration has attracted interest among researchers; 31 of all 59
review papers were published in the last decade. The number

Table 3  Literature summary of CVRP

Reference Problem Type of cooperation Method

[67] VRPTW Horizontal cooperation Metaheuristic

[68] MDCARPFL Horizontal cooperation Metaheuristic
[69] VRPTW with LTL carrier collaboration Horizontal cooperation Simulation, Heuristic
[70] MDVRPPC Horizontal cooperation Metaheuristic
[71] VRP Horizontal cooperation Metaheuristic
[72] CVRP with last-mile delivery Horizontal cooperation Simulation
[73] TSP Horizontal cooperation Metaheuristic
[74] TSP with multi-depots Horizontal cooperation Simulation
[75] Collaborative distribution between two LSPs Horizontal cooperation Exact
[76] The shared customer collaboration VRP Horizontal cooperation Exact
[77] Carrier collaboration VRP Horizontal cooperation Metaheuristic
[78] CCPPD Centralized collaboration Simulation
[79] VRP and carriers cooperation Centralized collaboration Metaheuristic
[80] SVRP Centralized collaboration Metaheuristic
[81] VRPTWDP Demand allocation Metaheuristic
[82] CVRP Capacity sharing Simulation
[83] CVRPSPD Demand allocation Metaheuristic
[84] CVRPSPD Capacity sharing Simulation
[85] Single carrier collaboration problem Capacity sharing Exact
[86] VRP Capacity sharing Metaheuristic
[87] MCCPPD Coalition. auction Metaheuristic
[88] CMCVRP Collaborative multiple DCs Metaheuristic
[89] CMCVRPSDP Capacity sharing Metaheuristic
[90] 2E-CMCVRP Capacity sharing Metaheuristic
[91] COLTSPSTW Capacity sharing, Horizontal cooperation Metaheuristic
[92] CTMCVRP-SST Capacity sharing Metaheuristic
[93] CMDVRPTWA​ Capacity sharing Metaheuristic
[94] CVRP-SCD Capacity sharing Metaheuristic
[95] CMDPDLN with split loads and time Capacity sharing Metaheuristic
[96] 2E-CMDPVRP Capacity sharing Metaheuristic
[97] Shipper collaboration problem considering the time and Shipper collaboration Metaheuristic
driver constraints
[98] Truckload shipper collaboration problem Shipper collaboration Metaheuristic
[99] Shipper collaboration problem Shipper collaboration Simulation
[100] Collaboration in small shippers with uncertain demand Shipper collaboration Simulation
[101] Shipper collaboration problem Shipper collaboration Metaheuristic
[102] VRPTW with backhauls Horizontal cooperation and environmental criteria Metaheuristic
[103] VRPTW with backhauls Horizontal cooperation and environmental criteria Simulation
[104] Multi depot VRP Horizontal cooperation and environmental criteria Metaheuristic
[105] Electric vehicles in VRP Horizontal cooperation and environmental criteria Metaheuristic

13
 M. Salehi Sarbijan, J. Behnamian

9 FVRP RTVRP CVRP

8
No. of publications 7
6
5
4
3
2
1
0

Year

Fig. 6  The number of papers with respect to their publication year and application fields

Table 4  The number of published papers per journal Fig. 7. As can be seen, the highest share is related to cus-
Journal #Papers
tomer requests (57.89). Also, travel time, combined cus-
tomer request and travel time, traffic congestion, combined
European Journal of Operational Research 11 travel time with traffic congestion, and time-dependent
Computers & Operations Research 10 were used in 13.16%, 10.53%, 7.89%, and 5.26% of the
Transportation Research Part E 7 papers, respectively.
Transportation Science 5 Table 6 shows the breakdown of the number of CVRP
Computers & Industrial Engineering 5 papers by type of cooperation. As can be seen, horizontal
Journal of Cleaner Production 5 cooperation and capacity-sharing methods were used in
Expert Systems with Applications 4 56% of the papers.
Applied Soft Computing 4 As shown in Fig. 8, shipper collaboration, horizontal
Procedia-Social and Behavioral Sciences 3 cooperation and environmental criteria, centralized col-
laboration, and demand allocation were used in 12.82%,
10.26%, 7.69%, and 5.13% of the papers, respectively.
Table 5  The number of classified papers in the field Source of dyna- Also, each of the coalition auctions, collaborative mul-
mism in RTVRP tiple DCs, and combined capacity sharing and horizontal
Source of dynamism Number of Percentage cooperation were considered in 2.56% of the papers.
papers The papers were further classified based on the prob-
lem attributes, as shown in Table 7. This classification
Customer request 22 57.89
helps identify trends of the VRP variants in all review
Travel time 5 13.16
papers and the percentage per publication year. The overall
Customer request and travel time 4 10.53
percentages of the VRP attributes in the study period are
Traffic congestion 3 7.89
illustrated in Fig. 9. As can be seen, the VRP variants have
Traffic congestion, travel time, cus- 2 5.26
tomer request been considered the most active topic in the research lit-
Traffic congestion, travel time 1 2.63 erature. According to the results, the capacitated VRP was
Time-dependent, travel time 1 2.63 considered in the majority of papers (21.78%). About 19%
of the papers consider DVRP. The VRP with time win-
dows was considered in 15.84%, and PDP and its variants
of review papers published in academic journals is shown in were considered in 9.9% of the studies. The CMCVRP and
Table 4. Our search results demonstrate that nine journals time-dependent VRP were considered in 7.92 and 5.94%
contain more than half (53) of the entire 89 review papers. of the papers, respectively. Other variants, such as shipper
Table 5 summarizes the source of dynamism of review collaboration problem, green VRP, multiple depots VRP,
papers published in RTVRP. Also, the percentage of the stochastic travel times VRP, VRP with backhauls, FVRP,
RTVRP attributes in the review papers is presented in

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

57.89

Attribute Presence (%)

13.16
10.53
7.89
5.26
2.63 2.63

Customer request Travel time Customer request Traffic congestion Traffic congestion, Traffic congestion, Time-dependent,
and travel time travel time, travel time travel time
customer request

Fig. 7  Percentage of RTVRP attributes in the review papers

Table 6  The number of classified papers on the field Type of coop- papers used heuristic and metaheuristic methods to solve the
eration in CVRP problem. Also, 15 and 9% of the papers used simulation and
Type of cooperation Number Percentage exact methods, respectively. Figure 11 summarizes the per-
of papers centage of heuristic and metaheuristic algorithms developed
in the review papers. From the results, 17.78% of papers
Horizontal cooperation 11 28.21
considered the VNS, LNS, and ALNS algorithms; 15.56%
Capacity sharing 11 28.21
of the papers are based on the TSA; the ACO and local
Shipper collaboration 5 12.82
search algorithms were included in 13.33% of the papers,
Horizontal cooperation and environmental 4 10.26
criteria and the NSGA-II and GA algorithms were used in 11.11%
Centralized collaboration 3 7.69 of the papers. Among these studies, 6.67% used time roll-
Demand allocation 2 5.13 ing horizon, nearest neighbor, Greedy algorithm, and PSO
Coalition auction 1 2.56 algorithms. Also, about 8.89 and 4.44% of them considered
Collaborative multiple DCs 1 2.56 Lagrangian relaxation and MOPSO algorithms, respectively.
Capacity sharing, horizontal cooperation 1 2.56

5 Future Research
and split delivery VRP, are considered in 4.95, 3.96, 2.97,
2.97, 1.98, 1.98, and 0.99% of the studies, respectively. The VRP is among the most widespread problems in opera-
Survey papers have been conducted using simulation, tions research, such that extensive studies on this problem
exact, heuristic, and metaheuristic solution methodologies. and its expansions have been presented so far. Recently, sev-
An overview of these methodologies is given in Table 8. eral papers were published on the VRP in 2020, mostly on
The overall percentages of simulation, exact, and heuristic solution methods such as Logic-based benders decomposi-
solution methodologies are illustrated in Fig. 10. Since the tion [106, 107], a parallelized large neighborhood search
VRP is NP-hard, according to Fig. 10, about 76% of the [108], An adaptive memetic approach [109], a constant

28.21 28.21
Attribute Presence (%)

12.82
10.26
7.69
5.13
2.56 2.56 2.56

Horizontal Capacity sharing Shipper Horizontal Centralized Demand Coalition. Collaborative Capacity sharing
cooperation collaboration cooperation and collaboration allocation auction multiple DCs , Horizontal
environmental cooperation
criteria

Fig. 8  Percentage of CVRP in the review papers

13
 M. Salehi Sarbijan, J. Behnamian

Table 7  Percentage of the VRP variants in the review papers approximation algorithm [110], the exponential multi-
VRP variants Number of Percentage (%) insertion neighborhood [111], memory-based iterated local
models search algorithm [112], and discrete scenario-based opti-
mization [113]. Some investigations have focused on the
Capacitated VRP 22 21.78
electric VRP and the related methods to solve it [114–117].
DVRP 19 18.81
Also, some papers have addressed the VRP with environ-
VRP with time windows 16 15.84
mental criteria [118, 119] and some others published papers
Pickup and delivery problem 10 9.90
in 2020 have considered the VRP in the cold supply chain
CMCVRP 8 7.92
[120–122]. The latest published papers were in the field of
Time-dependent VRP 6 5.94
the 2E-VRP [123–126]. In the following, future research
Shipper collaboration problem 5 4.95
directions are presented in the fields of FVRP, RTVRP, and
Green VRP 4 3.96
CVRP.
Multiple depots VRP 3 2.97
Stochastic travel times VRP 3 2.97
5.1 Future Research in FVRP
VRP with backhauls 2 1.98
FVRP 2 1.98
The FVRP is a new type of VRP applicable in urban trans-
Split delivery VRP 1 0.99
portation. First, there are fewer research works on the

21.78
18.81
Attribute Presence (%)

15.84

9.90
7.92
5.94 4.95
3.96 2.97 2.97 1.98 1.98 0.99

Fig. 9  Percentage of the VRP attributes in the review papers

Table 8  An overview of solution methodologies

Method Paper

Simulation Simulation: [17], Rolling horizon approach: [29], Best fit algorithm: [32],Clustering: [39], Stochastic optimization: [43], Sce-
nario based: [61, 72, 82, 84], Developed core computation algorithm: [74], Numerical experiments, Shapley value: [78, 103],
Cost-allocation mechanism, cooperative game theory:[99], Naive and myopic mechanisms:[100], Monte carlo: [103]
Exact DSS, A knowledge-based modeling approach, PAM: [42, 65, 82], Label correcting algorithm, shortest path: [56], Model based:
[75], Branch-and-cut: [76, 85], Dynamic programming: [88, 92]
Heuristic and Nearest neighbor: [18, 19, 80], Cost sharing and threshold: [20], Tabu search: [21, 33, 37, 57, 59, 60, 65, 83], Two-stage solu-
metaheuris- tion heuristics: [22], Sequential insertion heuristic: [23], Linkage and split approach: [24], Aco: [25, 2, 35, 86, 30, 46], Local
tic search: [26, 31, 34, 68, 69, 91], GA: [36, 49, 58, 66, 77], VNS, LNS, ALNS: [38, 70, 77, 81, 94, 45, 51, 53] PSO: [38, 27,
50, 52], An adaptive evolutionary approach: [41], Polynomial-time heuristic: [44], Lagrangian relaxation: [62, 63, 83, 87],
Time rolling horizon: [64, 73], Clarke and wright’s savings heuristic: [67, 93], GRASP: [71], Tailor-made: [73], The coalition
induction algorithm: [79], NSGA-II: [89, 90, 92, 93, 96], MOPSO: [89, 90], Greedy algorithm: [68, 97, 98], Nelder–mead
algorithm: [101], Scatter search: [102], Iterated local search: [104], A heuristic approach based in MILP: [105], memetic [47],
GA-TS: [54, 55, 95]

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

76.00 problem. Some cases, such as combining FVRP with hetero-

geneous fleets, pickup and delivery of goods, multi-depot,
Attribute Presence (%)

routing with backhaul, periodic, open vehicle routing, inven-

tory routing problem, inventory-location-routing, and green
vehicle routing, can be implemented as future developments.

15.00 9.00 5.2 Future Research in RTVRP

The real-time approach to vehicle routing is considered a

Heuristic and meta Simulation Exact re-optimization. Therefore, considering this approach in
heuristic
the VRP variants—e.g., E-VRP, the 2E-VRP, the rich VRP
(RVRP), green VRP, the heterogeneous VRP, the peri-
Fig. 10  Percentage of publications in terms of methods
odic VRP, MD-VRP, and muti-trip VRP (MT-VRP)—can
be considered a future development. Other developments
modeling aspect of FVRP. As a development of these prob- considered the real-time approach in various supply chain
lems, it is suggested to model the FVRP, especially with the types, such as closed-loop supply chains, multi-item multi-
increase in the number of vehicles. Second, the multi-agent echelon supply chains, agro-food supply chains, cold supply
system includes a group of sagacious independent agents chains, humanitarian supply chains, biomass supply chains,
that communicate with each other to achieve a special pur- and healthcare supply chains. Blocked roads are caused by
pose. As another development, the multi-agent system is accidents and vehicle breakdowns as sources of dynamism.
proposed for the FVRP based on the agents’ behavior and Furthermore, the effect of real-time on the sustainability of
the coordination between the agents. Third, according to the routing problems is suggested.
research literature on the FVRP, the researchers have focused
mostly on economic optimization. Therefore, it is suggested 5.3 Future Research in CVRP
to consider aspects of sustainability such as environmen-
tal and social development associated with this problem. The collaborative vehicle routing problem is an active
Fourth, it is necessary to consider reverse logistics models, research subject with high applied importance. Suggesting
closed-loop supply chain models, models with a perishable future works are as follows:
product, emission caps, and carbon prices in the FVRP.
Fifth, a humanitarian supply chain deals with delivering the • combing of horizontal and vertical cooperation.
resources needed by people in areas affected by natural dis- • considering operational, tactical and strategic planning
asters. Since it is difficult to reach some places during natu- horizons in decisions related to the collaborative vehicle
ral disasters, the development of the FVRP in the humanitar- routing problem.
ian supply chain is proposed. Sixth, as a future line of work, • Incorporating new approaches such as data mining, big
the extension of the population-based metaheuristics such data, and blockchain in the CVRP.
as artificial bee colony [120, 127], discrete firefly algorithm • using social aspects in the CVRP modeling.
[128], discrete paper swarm optimization [129], and robust • applying the multi-period and multi-item in CVRP.
optimization approach [130] can be used. The FVRP can • collaborating in multi-echelon vehicle routing optimiza-
also be extended to the scope of vehicle routing as a routing tion.

17.78
15.56
13.33 13.33
Attribute Presence (%)

11.11 11.11
8.89
6.67 6.67 6.67 6.67
4.44

VNS, LNS, Tabu ACO Local NSGA-II GA Lagrangian Time Nearest Greedy PSO MOPSO
ALNS search search relaxation rolling neighbor algorithm
horizon

Fig. 11  Percentage of heuristic and metaheuristic in the review papers

13
 M. Salehi Sarbijan, J. Behnamian

• studying pickup and delivery based on resource sharing 7. Pillac V et al (2013) A review of dynamic vehicle routing prob-
in CVRP. lems. Eur J Oper Res 225(1):1–11
8. Abbatecola L, Fanti MP, Ukovich W (2016) A review of new
approaches for dynamic vehicle routing problem. In: IEEE
international conference on automation science and engineer-
ing (CASE). IEEE
6 Conclusion 9. Adacher L (2019) Optimization heuristics for dynamic vehicle
routing problems. Int J Simul-Syst Sci Technol 20(1):1–7
The VRP is one of the major problems in operations research 10. Ghiani G et al (2003) Real-time vehicle routing: solution con-
cepts, algorithms and parallel computing strategies. Eur J Oper
related to transportation science. A systematic review of the Res 151(1):1–11
literature on FVRP, RTVRP, and Co-VRP for urban trans- 11. Guajardo M, Rönnqvist M (2016) A review on cost allocation
portation was conducted based on 89 published papers methods in collaborative transportation. Int Trans Oper Res
between 2001 and 2022. Among these papers, 12, 39, and 23(3):371–392
12. Gansterer M, Hartl RF (2018) Collaborative vehicle routing: a
38 papers were devoted to FVRP, CoVRP, and RTVRP top- survey. Eur J Oper Res 268(1):1–12
ics, respectively. Based on our search, 31 papers from all 13. Cleophas C et al (2019) Collaborative urban transporta-
59 published papers in the last decade were associated with tion: recent advances in theory and practice. Eur J Oper Res
CVRP. This review also showed that travel time, combined 273(3):801–816
14. Pan S et al (2019) Horizontal collaborative transport: survey of
customer request and travel time, traffic congestion, com- solutions and practical implementation issues. Int J Prod Res
bined travel time with traffic congestion, and time-dependent 57(15–16):5340–5361
were used in 13.16%, 10.53%, 7.89%, and 5.26% of papers, 15. Abdallah M et al (2020) Artificial intelligence applications in
respectively. Also, shipper collaboration, horizontal coop- solid waste management: a systematic research review. Waste
Manage 109:231–246
eration and environmental criteria, centralized collaboration, 16. Keele S (2007) Guidelines for performing systematic litera-
and demand allocation are used in 12.82%, 10.26%, 7.69%, ture reviews in software engineering. Technical report, Ver.
and 5.13% of the papers, respectively. The results of the sur- 2.3 EBSE technical report. EBSE
vey on VRP variants demonstrated that about 21.78% of the 17. Tu S, Lai, Li Y (2001) Application of the vehicle routing prob-
lem with time windows—an example of lunch box delivery
papers were related to the capacitated VRP. Furthermore, in graduation term paper, department of transportation technol-
about 76% of the papers, heuristic and metaheuristic meth- ogy and logistics management, Chung Hua University, Hsin
ods were used to solve the problem. Finally, at the end of Chu, Taiwan,
this paper, several suggestions for future research directions 18. Chang J, Cho YJ, Hwang YC (2001) A study on time con-
strained vehicle routing problem for lunch box delivery. In:
were proposed in three categories: future research in the Proceedings of the annual meeting of Chinese Institute of
FVRP, future research in the RTVRP, and future research Industrial Engineering. Kaohsiung, Taiwan
in the CVRP. 19. Chang Y (2006) Model formulation and solution algorithms for
the line-haul feeder problem MS thesis, Dept. Transportation
Technol. Logistics Manage., Chung Hua University, Hsin Chu,
Taiwan,
Declarations 20. Chen HK et al (2011a) The linehaul-feeder vehicle routing
problem with virtual depots. IEEE Trans Autom Sci Eng
Conflict of interest No potential conflict of interest was reported by
8(4):694–704
the authors.
21. Chen HK, Chou HW, Hsu CY (2011b) The linehaul-feeder
vehicle routing problem with virtual depots and time windows.
Math Probl Eng. https://​doi.​org/​10.​1155/​2011/​759418
22. Chen HK, Wang H (2012) A two-stage algorithm for the
References extended linehaul-feeder vehicle routing problem with time
windows. Int J Shipp Transp Logist 4(4):339–356
1. Dantzig GB, Ramser JH (1959) The truck dispatching problem. 23. Chen HK (2015) Issues for the linehaul-feeder vehicle routing
Manage Sci 6(1):80–91 problem with virtual depots and time windows. J East Asia Soc
2. Huang Y-H et al (2019) Solving the feeder vehicle routing prob- Transp Stud 11:678–692
lem using ant colony optimization. Comput Ind Eng 127:520–535 24. Brandstätter C, Reimann M (2018) The line-haul feeder vehicle
3. Long Q (2016) A multi-methodological collaborative simulation routing problem: mathematical model formulation and heuris-
for inter-organizational supply chain networks. Knowl Based Syst tic approaches. Eur J Oper Res 270(1):157–170
96:84–95 25. Brandstätter C, Reimann M (2018b) Performance analysis of a
4. Xu X-F et al (2017) Design optimization of resource combination metaheuristic algorithm for the line-haul feeder vehicle routing
for collaborative logistics network under uncertainty. Appl Soft problem. J Veh Routing Algorithms 1(2–4):121–138
Comput 56:684–691 26. Brandstätter C (2019) A metaheuristic algorithm and structured
5. Vidal T, Laporte G, Matl P (2019) A concise guide to existing analysis for the line-haul Feeder Vehicle Routing Problem with
and emerging vehicle routing problem variants. Eur J Oper Res Time Windows. Central Eur J Oper Res 29:247–289
286:401–416 27. Salehi Sarbijan M, Behnamian J (2022) Multi-fleet feeder vehi-
6. Koç Ç, Laporte G, Tükenmez İ (2020) A review on vehicle rout- cle routing problem using hybrid metaheuristic. Comput Oper
ing with simultaneous pickup and delivery. Comput Oper Res Res 141:105696
122:104987

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

28. Bekta ST, Repoussis PP, Tarantilis CD (2014) Dynamic vehicle 50. Okulewicz M, Mańdziuk J (2017) The impact of particular com-
routing problems. Vehicle routing: problems, methods, and appli- ponents of the PSO-based algorithm solving the dynamic vehicle
cations, 2nd edn. SIAM, Philadelphia, pp 299–347 routing Problem. Appl Soft Comput 58:586–604
29. Yang J, Jaillet P, Mahmassani H (2004) Real-time multivehi- 51. Chen S et al (2018) An adaptive large neighborhood search heu-
cle truckload pickup and delivery problems. Transport Sci ristic for dynamic vehicle routing problems. Comput Electr Eng
38(2):135–148 67:596–607
30. Montemanni R et al (2005) Ant colony system for a dynamic 52. Okulewicz M, Mańdziuk J (2019) A metaheuristic approach
vehicle routing problem. J Comb Optim 10(4):327–343 to solve dynamic vehicle routing problem in continuous search
31. Potvin J-Y, Xu Y, Benyahia I (2006) Vehicle routing and space. Swarm Evol Comput 48:44–61
scheduling with dynamic travel times. Comput Oper Res 53. Archetti C, Guerriero F, Macrina G (2021) The online vehicle
33(4):1129–1137 routing problem with occasional drivers. Comput Oper Res
32. Du T, Wang F, Lu P-Y (2007) A real-time vehicle-dispatching 127:105144
system for consolidating milk runs. Transp Res E 43(5):565–577 54. Guiqin X et al (2021) A combined GA-TS algorithm for two-
33. Jemai J, Mellouli K (2008) A neural-tabu search heuristic for echelon dynamic vehicle routing with proactive satellite stations.
the real time vehicle routing problem. J Math Model Algorithms Comput Ind Eng 164:107899
7(2):161–176 55. Ge X, Jin Y (2021) Artificial intelligence algorithms for proactive
34. Hsueh C-F, Chen H-K, Chou H-W (2008) Dynamic vehicle dynamic vehicle routing problem. Applications of artificial intel-
routing for relief logistics in natural disasters, in vehicle routing ligence in process systems engineering. Elsevier, Amsterdam, pp
problem. IntechOpen, London 497–522
35. de Oliveira SM, de Souza SR, Silva MAL (2008) A solution 56. Fu L (2001) An adaptive routing algorithm for in-vehicle route
of dynamic vehicle routing problem with time window via ant guidance systems with real-time information. Transp Res B
colony system metaheuristic. In: 2008 10th Brazilian symposium 35(8):749–765
on neural networks. IEEE 57. Chang M-S, Chen S-r, Hsueh C-F (2003) Real-time vehicle
36. Okhrin I, Richter K (2009) Vehicle routing problem with real- routing problem with time windows and simultaneous delivery/
time travel times. Int J Veh Inf Commun Syst 2(1–2):59–77 pickup demands. J East Asia Soc Transp Stud 5:2273–2286
37. Liao T-Y, Hu T-Y (2011) An object-oriented evaluation frame- 58. Haghani A, Jung S (2005) A dynamic vehicle routing prob-
work for dynamic vehicle routing problems under real-time infor- lem with time-dependent travel times. Comput Oper Res
mation. Expert Syst Appl 38(10):12548–12558 32(11):2959–2986
38. Khouadjia MR et al (2012) A comparative study between 59. Ferrucci F, Bock S, Gendreau M (2013) A pro-active real-
dynamic adapted PSO and VNS for the vehicle routing problem time control approach for dynamic vehicle routing problems
with dynamic requests. Appl Soft Comput 12(4):1426–1439 dealing with the delivery of urgent goods. Eur J Oper Res
39. Wohlgemuth S, Oloruntoba R, Clausen U (2012) Dynamic vehi- 225(1):130–141
cle routing with anticipation in disaster relief. Socio-Econ Plann 60. Van Woensel T et al (2008) Vehicle routing with dynamic travel
Sci 46(4):261–271 times: a queueing approach. Eur J Oper Res 186(3):990–1007
40. Hong L (2012) An improved LNS algorithm for real-time vehi- 61. Zhao X, Goncalves G, Dupas R (2009) Dynamic vehicle routing
cle routing problem with time windows. Comput Oper Res problem with real-time time-dependent travel times. In: 2009
39(2):151–163 IEEE Intelligent Vehicles Symposium. IEEE
41. Barkaoui M, Gendreau M (2013) An adaptive evolutionary 62. Li J-Q, Mirchandani PB, Borenstein D (2009) A lagrangian heu-
approach for real-time vehicle routing and dispatching. Comput ristic for the real-time vehicle rescheduling problem. Transp Res
Oper Res 40(7):1766–1776 E 45(3):419–433
42. Hu X, Sun L, Liu L (2013) A PAM approach to handling disrup- 63. Li J-Q, Mirchandani PB, Borenstein D (2009) Real-time vehi-
tions in real-time vehicle routing problems. Decis Support Syst cle rerouting problems with time windows. Eur J Oper Res
54(3):1380–1393 194(3):711–727
43. Naoum-Sawaya J, Elhedhli S (2013) A stochastic optimization 64. Chen H-K, Hsueh C-F, Chang M-S (2006) The real-time time-
model for real-time ambulance redeployment. Comput Oper Res dependent vehicle routing problem. Transp Res E 42(5):383–408
40(8):1972–1978 65. Grzybowska H, Barceló J (2012) Decision support system for
44. Jagtenberg CJ, Bhulai S, van der Mei RD (2015) An efficient real-time urban freight management. Procedia-Soc Behav Sci
heuristic for real-time ambulance redeployment. Oper Res Health 39:712–725
Care 4:27–35 66. Messaoud E, Alaoui A (2017) Minimizing the total cost in the
45. de Armas J, Melián-Batista B (2015) Variable neighborhood real time vehicle routing problem with the traffic factors. In: 18th
search for a dynamic rich vehicle routing problem with time ROADEF conference of the French operations research & deci-
windows. Comput Ind Eng 85:120–131 sion support society, February 22–24, Metz, France, pp 1–2
46. Euchi J, Yassine A, Chabchoub H (2015) The dynamic vehicle 67. Cruijssen F et al (2007) Joint route planning under varying mar-
routing problem: solution with hybrid metaheuristic approach. ket conditions. Int J Phys Distrib Logist Manage. https://​doi.​org/​
Swarm Evol Comput 21:41–53 10.​1108/​09600​03071​07525​14
47. Mańdziuk J, Żychowski A (2016) A memetic approach to vehi- 68. Liu R et al (2010) Two-phase heuristic algorithms for full truck-
cle routing problem with dynamic requests. Appl Soft Comput loads multi-depot capacitated vehicle routing problem in carrier
48:522–534 collaboration. Comput Oper Res 37(5):950–959
48. Sarasola B et al (2016) Variable neighborhood search for the 69. Nadarajah S, Bookbinder JH (2013) Less-than-truckload car-
stochastic and dynamic vehicle routing problem. Ann Oper Res rier collaboration problem: modeling framework and solution
236(2):425–461 approach. J Heuristics 19(6):917–942
49. AbdAllah AMF, Essam DL, Sarker RA (2017) On solving peri- 70. Stenger A et al (2013) An adaptive variable neighborhood search
odic re-optimization dynamic vehicle routing problems. Appl algorithm for a vehicle routing problem arising in small package
Soft Comput 55:1–12 shipping. Transport Sci 47(1):64–80

13
 M. Salehi Sarbijan, J. Behnamian

71. Adenso-Díaz B et al (2014) Assessing partnership savings in 93. Wang Y et al (2020b) Collaborative multi-depot logistics net-
horizontal cooperation by planning linked deliveries. Transp work design with time window assignment. Expert Syst Appl
Res A 66:268–279 140:112910
72. Muñoz-Villamizar A, Montoya-Torres JR, Vega-Mejía CA 94. Zhang W et al (2020) Composite multi-objective optimization on
(2015) Non-collaborative versus collaborative last-mile deliv- a new collaborative vehicle routing problem with shared carriers
ery in urban systems with stochastic demands. Procedia CIRP and depots. J Clean Prod 274:122593
30:263–268 95. Wang Y et al (2021) Collaborative multi-depot pickup and deliv-
73. Kimms A, Kozeletskyi I (2016) Core-based cost allocation in ery vehicle routing problem with split loads and time windows.
the cooperative traveling salesman problem. Eur J Oper Res Knowl Based Syst 231:107412
248(3):910–916 96. Wang Y et al (2021) Two-echelon collaborative multi-depot
74. Kimms A, Kozeletskyi I (2016) Shapley value-based cost allo- multi-period vehicle routing problem. Expert Syst Appl
cation in the cooperative traveling salesman problem under 167:114201
rolling horizon planning. EURO J Transp Logist 5(4):371–392 97. Ergun O, Kuyzu G, Savelsbergh M (2007) Reducing truckload
75. Yang F et al (2016) Collaborative distribution between transportation costs through collaboration. Transp Sci 41(2):
two logistics service providers. Int Trans Oper Res 206–221
23(6):1025–1050 98. Ergun O, Kuyzu G, Savelsbergh M (2007) Shipper collaboration.
76. Fernández E, Roca-Riu M, Speranza MG (2018) The shared Comput Oper Res 34(6):1551–1560
customer collaboration vehicle routing problem. Eur J Oper Res 99. Özener O, Ergun Ö (2008) Allocating costs in a collabo-
265(3):1078–1093 rative transportation procurement network. Transport Sci
77. Vaziri S, Etebari F, Vahdani B (2019) Development and opti- 42(2):146–165
mization of a horizontal carrier collaboration vehicle routing 100. Yilmaz O, Savasaneril S (2012) Collaboration among small ship-
model with multi-commodity request allocation. J Clean Prod pers in a transportation market. Eur J Oper Res 218(2):408–415
224:492–505 101. Sun L et al (2015) Transportation cost allocation on a fixed route.
78. Dai B, Chen H (2012) Profit allocation mechanisms for carrier Comput Ind Eng 83:61–73
collaboration in pickup and delivery service. Comput Ind Eng 102. Pradenas L, Oportus B, Parada V (2013) Mitigation of green-
62(2):633–643 house gas emissions in vehicle routing problems with backhaul-
79. Zakharov VV, Shchegryaev AN (2015) Stable cooperation in ing. Expert Syst Appl 40(8):2985–2991
dynamic vehicle routing problems. Autom Remote Control 103. Juan A et al (2014) Horizontal cooperation in vehicle routing
76(5):935–943 problems with backhauling and environmental criteria. Procedia-
80. Defryn C, Sörensen K, Cornelissens T (2016) The selective vehi- Soc Behav Sci 111:1133–1141
cle routing problem in a collaborative environment. Eur J Oper 104. Pérez-Bernabeu E et al (2015) Horizontal cooperation in road
Res 250(2):400–411 transportation: a case illustrating savings in distances and green-
81. Krajewska MA et al (2008) Horizontal cooperation among freight house gas emissions. Int Trans Oper Res 22(3):585–606
carriers: request allocation and profit sharing. J Oper Res Soc 105. Muñoz-Villamizar A, Montoya-Torres JR, Faulin J (2017) Impact
59(11):1483–1491 of the use of electric vehicles in collaborative urban transport
82. Dahl S, Derigs U (2011) Cooperative planning in express carrier networks: a case study. Transp Res D 50:40–54
networks—an empirical study on the effectiveness of a real-time 106. Fachini RF, Armentano VA (2020) Logic-based Benders decom-
decision support system. Decis Support Syst 51(3):620–626 position for the heterogeneous fixed fleet vehicle routing problem
83. Li J (2013) Model and simulation for collaborative VRPSPD. J with time windows. Comput Ind Eng 148:106641
Netw 8(2):331 107. Alkaabneh F, Diabat A, Gao HO (2020) Benders decomposition
84. Vornhusen B, Wang X, Kopfer H (2014) Vehicle routing under for the inventory vehicle routing problem with perishable prod-
consideration of transhipment in horizontal coalitions of freight ucts and environmental costs. Comput Oper Res 113:104751
carriers. Procedia CIRP 19:117–122 108. Mühlbauer F, Fontaine P (2020) A parallelised large neighbour-
85. Hernández S, Peeta S, Kalafatas G (2011) A less-than-truckload hood search heuristic for the asymmetric two-echelon vehicle
carrier collaboration planning problem under dynamic capacities. routing problem with swap containers for cargo-bicycles. Eur J
Transp Res E 47(6):933–946 Oper Res 289:742–757
86. Sprenger R, Mönch L (2012) A methodology to solve large-scale 109. Sabar NR et al (2020) An adaptive memetic approach for hetero-
cooperative transportation planning problems. Eur J Oper Res geneous vehicle routing problems with two-dimensional loading
223(3):626–636 constraints. Swarm Evolut Comput 58:100730
87. Dai B, Chen H, Yang G (2014) Price-setting based combinato- 110. Fernstrøm F, Steiner TA (2020) A constant approximation algo-
rial auction approach for carrier collaboration with pickup and rithm for the uniform a priori capacitated vehicle routing prob-
delivery requests. Oper Res Int J 14(3):361–386 lem with unit demands. Inf Process Lett 159:105960
88. Wang Y et al (2017) Profit distribution in collaborative multiple 111. Buckow J-N, Graf B, Knust S (2020) The exponential multi-
centers vehicle routing problem. J Clean Prod 144:203–219 insertion neighborhood for the vehicle routing problem with unit
89. Wang Y et al (2018a) Collaboration and transportation resource demands. Comput Oper Res 120:104949
sharing in multiple centers vehicle routing optimization with 112. Brandão J (2020) A memory-based iterated local search algo-
delivery and pickup. Knowl Based Syst 160:296–310 rithm for the multi-depot open vehicle routing problem. Eur J
90. Wang Y et al (2018b) Economic and environmental evaluations Oper Res 284(2):559–571
in the two-echelon collaborative multiple centers vehicle routing 113. Wu L, Hifi M (2020) Discrete scenario-based optimization for the
optimization. J Clean Prod 197:443–461 robust vehicle routing problem: the case of time windows under
91. Defryn C, Sörensen K (2018) Multi-objective optimisation mod- delay uncertainty. Comput Ind Eng 145:106491
els for the travelling salesman problem with horizontal coopera- 114. Zhang S et al (2020) Fuzzy optimization model for electric vehi-
tion. Eur J Oper Res 267(3):891–903 cle routing problem with time windows and recharging stations.
92. Wang Y et al (2020a) Collaborative two-echelon multicenter Expert Syst Appl 145:113123
vehicle routing optimization based on state–space–time network 115. Lu J et al (2020) The time-dependent electric vehicle routing
representation. J Clean Prod 258:120590 problem: model and solution. Expert Syst Appl 161:113593

13
Emerging Research Fields in Vehicle Routing Problem: A Short Review

116. Kancharla SR, Ramadurai G (2020) Electric vehicle routing 126. Enthoven DL et al (2020) The two-echelon vehicle routing prob-
problem with non-linear charging and load-dependent discharg- lem with covering options: city logistics with cargo bikes and
ing. Expert Syst Appl 160:113714 parcel lockers. Comput Oper Res 118:104919
117. Bahrami S et al (2020) The plugin hybrid electric vehicle routing 127. Zhou X et al (2020) Enhancing artificial bee colony algorithm
problem: a power-management strategy model. Transp Res C with multi-elite guidance. Inf Sci 543:242–258
111:318–333 128. Li J et al (2019) Discrete firefly algorithm with compound neigh-
118. Giallanza A, Puma GL (2020) Fuzzy green vehicle routing prob- borhoods for asymmetric multi-depot vehicle routing problem
lem for designing a three echelons supply chain. J Clean Prod in the maintenance of farm machinery. Appl Soft Comput
259:120774 81:105460
119. Behnke M, Kirschstein T, Bierwirth C (2020) A column genera- 129. Chen R-M, Shen Y-M, Hong W-Z (2019) Neural-like encoding
tion approach for an emission-oriented vehicle routing problem particle swarm optimization for periodic vehicle routing prob-
on a multigraph. Eur J Oper Res 288:794–809 lems. Expert Syst Appl 138:112833
120. Song M-X et al (2020) Metaheuristics for solving the vehicle 130. Santos MJ et al (2020) A robust optimization approach for the
routing problem with the time windows and energy consumption vehicle routing problem with selective backhauls. Transp Res E
in cold chain logistics. Appl Soft Comput 95:106561 136:101888
121. Qi C, Hu L (2020) Optimization of vehicle routing problem for
emergency cold chain logistics based on minimum loss. Phys Publisher’s Note Springer Nature remains neutral with regard to
Commun 40:101085 jurisdictional claims in published maps and institutional affiliations.
122. Liu G et al (2020) Vehicle routing problem in cold Chain logis-
tics: a joint distribution model with carbon trading mechanisms. Springer Nature or its licensor (e.g. a society or other partner) holds
Resour Conserv Recycl 156:104715 exclusive rights to this article under a publishing agreement with the
123. Li H et al (2020) Two-echelon vehicle routing problem with time author(s) or other rightsholder(s); author self-archiving of the accepted
windows and mobile satellites. Transp Res Part B: Methodologi- manuscript version of this article is solely governed by the terms of
cal 138:179–201 such publishing agreement and applicable law.
124. Li H et al (2020) Two-echelon vehicle routing problem with sat-
ellite bi-synchronization. Eur J Oper Res 288:775–793
125. Kitjacharoenchai P, Min B-C, Lee S (2020) Two echelon vehicle
routing problem with drones in last mile delivery. Int J Prod Econ
225:107598

13