IEEE Computational Intelligence Magazine - May 2022 - 90

booking apps such as Uber, which have
recently become popular, can gather
information from both passengers and
taxis. This information can be used to
make a global dispatch of taxis for passengers.
Situ et al. [86] proposed an
ACO-based algorithm to generate a
taxi-passenger matching scheme for
maximizing profit. To address a large
number of passengers and taxis, a region
is divided into several subregions, and
dispatch in each subregion is conducted
in parallel. Gong et al. [87] considered
both the profit and QoS when formulating
the taxi-passenger matching
scheme. Specifically, a fuzzy logic control
system is proposed to score each
potential taxi-passenger pair based on
the profit and QoS, in which DE is
employed to optimize membership
functions and fuzzy rules. Jung et al. [88]
focused on the shared-taxi scenario that
allowed a taxi to detour to pick up
other passengers when some passengers
were already in the taxi during rush
hours. A simulated annealing-based
algorithm is proposed to minimize the
passenger travel time and to maximize
the profit. In the shared-taxi scenario,
some passengers need to spend more
travel time due to the detour so they
expect lower prices. However, the profit
of drivers should also be considered. To
satisfy both passengers and drivers,
Zhang et al. [89] proposed a GA-based
algorithm to optimize the pricing
scheme. Liang et al. [90] investigated the
dispatch of electric taxis. Since the
charging of electric taxis requires much
more time than the refueling of traditional
petrol taxis, the dispatch of electric
taxis should consider the remaining
battery capacity and the location of
charging facilities. The primary objective
of this problem is to maximize the
number of passenger requests that are
satisfied, while the secondary objective is
to reduce the total waiting time of passengers
whose requests are satisfied. An
ACO-based algorithm with a pre-selection
strategy and local pruning strategy
is proposed in [90] to solve this problem.
Carpooling is also called ride-sharJack
...
Amy
John
Linda
Susan
Charles
Fiona
FIGURE
4 Example of carpooling.
TABLE I The objectives considered in the related studies on carpooling.
REFERENCE
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
DRIVERS'
TRAVEL
DISTANCE
PASSENGERS'
TRAVEL
DISTANCE
NO. OF
MATCHED
PASSENGERS
PASSENGERS'
WAITING
DISTANCE
SEAT
USAGE
RATE
ing, and an example is shown in Fig. 4.
People nearby may have the same or
nearby destination(s) (e.g., people living
in the same block also work in the same
company). Rather than driving to the
destination in separate cars, several people
could share one car. In Fig. 4,
Charles will drive the blue car and take
Amy and Jack together, while Fiona will
drive the gray car and take John, Linda,
and Susan together. There are three benefits
of carpooling. The first benefit is
that the fuel cost of drivers can be
shared by the other passengers, and the
passengers also spend less money compared
to driving by themselves. The second
and third benefits are reducing traffic
congestion and the emission of
greenhouse gas, respectively, because
fewer cars are on the road. However,
drivers must detour from their shortest
route to pick up passengers, which
incurs additional driving distance, fuel
costs, and time costs. Passengers also
have certain demands, including the
pick-up time of cars. Thus, efficient
algorithms for matching drivers and passengers
should be developed to provide
high-quality carpool services.
To manage carpooling, Huang and
Jiau investigated various optimization
objectives in carpooling [91]-[98]. The
details of the objectives considered in
the related studies are compared in
Table I. In [94] and [98], the concept of
a time window (i.e., a specific window
of time when passengers require cars to
pick them up) is considered. Early
arrival or a delay will affect the user
experience. To address multiple optimization
objectives, integrated models that
combine several objectives into a single
objective function are formulated in
[91] and [96], while dual-objective
90 IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | MAY 2022
...
IEEE Computational Intelligence Magazine - May 2022

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - May 2022
