Das, S., & Maurya, A. K. (2018)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

Transport Reviews

ISSN: 0144-1647 (Print) 1464-5327 (Online) Journal homepage: https://www.tandfonline.com/loi/ttrv20

Modelling of motorised two-wheelers: a review of


the literature

Sanhita Das & Akhilesh Kumar Maurya

To cite this article: Sanhita Das & Akhilesh Kumar Maurya (2018) Modelling of motorised
two-wheelers: a review of the literature, Transport Reviews, 38:2, 209-231, DOI:
10.1080/01441647.2017.1317049

To link to this article: https://doi.org/10.1080/01441647.2017.1317049

Published online: 18 Apr 2017.

Submit your article to this journal

Article views: 1213

View related articles

View Crossmark data

Citing articles: 13 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ttrv20
TRANSPORT REVIEWS, 2018
VOL. 38, NO. 2, 209–231
https://doi.org/10.1080/01441647.2017.1317049

Modelling of motorised two-wheelers: a review of the


literature
Sanhita Das and Akhilesh Kumar Maurya
Department of Civil Engineering, Indian Institute of Technology Guwahati, Assam, India

ABSTRACT ARTICLE HISTORY


Motorised two-wheeler (MTW) has emerged as the most widely Received 26 October 2016
used transportation mode for short distance trips in many Asian Accepted 1 April 2017
and European countries because of its affordable price and
KEYWORDS
flexibility in manoeuvring through traffic. Its small size and Motorised two-wheeler;
aggressive riding patterns can considerably affect the overall motorcycle; heterogeneous;
traffic flow characteristics and safety. Modelling the characteristic mixed traffic; modelling
behaviour and complex movement patterns of motorised two-
wheelers is essential due to their pronounced role in microscopic
traffic simulations, traffic management strategies, safety
evaluation, transport planning and intelligent transportation
systems. This paper begins with an overview of the determinant
factors used for MTW modelling, followed by a critical review of
various modelling approaches that have been adopted to depict
the movement patterns of MTWs. The review concentrates on (i)
identifying the factors mostly studied with respect to MTW-
research, (ii) discussing the potential and limitations of each
modelling approach and (iii) reporting some of the aspects that
need further consideration for modelling MTWs in heterogeneous
mixed traffic streams. The findings of the paper suggest that an
integrated approach can be more viable to depict the unique
riding patterns of motorised two-wheelers and some possible
directions for future studies are also proposed.

Introduction
Motorised two-wheeler (MTW), commonly termed as motorcycle, possesses a number of
advantages over other transportation modes in terms of its flexibility, affordability, high
manoeuvrability, convenience, parking space utilisation and reduced emissions. It can at
times prove to be faster than other modes of travel in many dense urban areas. Its
compact size, flexible mobility and lower costs have enabled substantial extension in
MTWs usage, mostly in Asian countries, Southern European countries (Hsu, Sadullah, &
Dao, 2003; Hussain, Farhan, Umar, & Dadang, 2005; Lee, 2008) and also in Latin American
cities as well (Dimitriou & Gakenheimer, 2011). Despite its higher usage and large concen-
tration in many countries, MTWs are seldom considered as a full mode of transportation in
western societies (Wigan, 2002).

CONTACT Sanhita Das sanhi.das25@gmail.com Department of Civil Engineering, Indian Institute of Technology
Guwahati, Assam 781039, India
© 2017 Informa UK Limited, trading as Taylor & Francis Group
210 S. DAS AND A. K. MAURYA

MTWs can be discerned from other vehicle classes both in driving style, and physical
and dynamic attributes. Its small sizes, narrow widths, high power-to-weight ratio, wide
field of view and an intuitive steering method provide MTW riders more flexibility in man-
oeuvring and freedom in choosing high speeds and chaotic trajectories (Vlahogianni,
2014). Lee (2008) summarised some distinctive riding patterns of MTWs as
travelling alongside another vehicle in the same lane, oblique following, moving to the head
of a queue, filtering, swerving or weaving, tailgating, maintaining a shorter headway when
aligning to the lateral edge of the preceding vehicle, travelling according to the virtual
lanes formed dynamically by the vehicles in surroundings, and self-organisation phenomena.

Modelling these unique riding patterns of MTWs through road traffic is indeed challen-
ging because of its complexity in detailing the chaotic and erratic trajectories of MTWs.
Due to the increasing share of MTW traffic in many urban areas, traffic modellers in
many countries, specifically in Asian countries, are showing more interest in modelling
the characteristic behaviour of MTWs as well as its interactions with rest of the traffic
stream. Modelling the dynamic movement patterns in road traffic provides a precise rep-
resentation of the overall traffic flow, which plays a predominant role in microsimulation
models’ development, capacity estimation, safety evaluation and in adopting suitable
traffic management strategies. However, there exists a paucity of research related to the
MTW behavioural modelling in the existing traffic flow studies and this review reflects
the burgeoning academic literature on the modelling of MTWs.
To the best of authors’ knowledge, no comprehensive review of all the modelling
approaches and determinant factors influencing MTW riding patterns is available
(although an overview of MTW-oriented Intelligent Transport Systems research has
been undertaken by Barmpounakis, Vlahogianni, and Golias,2016a). This paper begins
by providing an overview of research conducted on different areas related to MTW
traffic, followed by an identification of various determinants that govern unique man-
oeuvring patterns of MTWs. This paper also aims to offer a systematic review of all the
modelling approaches that have been extensively used to depict the behavioural patterns
of MTWs. The review highlights limitations in the modelling approaches that need further
consideration, followed by a concoction of future research directions and critical gaps in
the prevailing MTW-oriented literature that can be beneficial in further improvement of
the existing models.

Determinant factors for MTW modelling


A large number of studies have characterised the manoeuvring patterns of MTWs with
respect to spacing (Walton & Buchanan, 2012), lane utilisation (Hsu et al., 2003), percola-
tion/filtering/seepage (Agarwal, Zilske, Rao, & Nagel, 2015; Fan & Work, 2015), virtual
lane width determination (Minh, Sano, & Matsumoto, 2012) and overtaking manoeuvres
(Barmpounakis, Vlahogianni, & Golias, 2016b). These phenomena have been investigated
in a number of research work, particularly in congested flow conditions. Most research in
MTW-related modelling studies employs speed, longitudinal/lateral headways and spacing
to the front/surrounding vehicles as significant factors (Lan & Chang, 2003; Lan, Chiou, Lin,
& Hsu, 2010; Meng, Dai, Dong, & Zhang, 2007) in describing different traffic phenomena.
TRANSPORT REVIEWS 211

An overview of studies conducted to elucidate different manoeuvring patterns of MTWs


and the modelled traffic variables are described in this section.

Spacing, positioning and discharge rates


Small sizes and excellent manoeuvring capabilities of MTWs encourage the riders to
exhibit quick lateral manoeuvres, which indeed results in shorter longitudinal and
lateral spacing than other vehicle classes (Spyropoulou & Sermpis, 2009; Walton & Bucha-
nan, 2012; Wong, Lee, & Chen, 2016). The lateral positioning of MTWs depends to a large
extent on congestion levels (Wong et al., 2016). MTWs in Asian countries where left-hand-
driving rule is followed are observed to be concentrated on the left-most lanes (Hsu et al.,
2003; Kanagaraj, Asaithambi, Toledo, & Lee, 2015) while cars tend to utilise the right-most
lanes. In particular, the high manoeuvrability enables MTW riders to attain higher speeds
even at the inner-most lanes (shoulder lanes).
Taking advantage of the compact size, MTWs utilise the available lateral gaps between
vehicles and attempt to proceed in front of a queue near signalised intersections (Fan &
Work, 2015; Nair, Mahmassani, & Miller-Hooks, 2011; Oketch, 2000). This accumulation of
MTWs near the stop line increases the probability of a higher discharge rate (Haque,
Chin, & Huang, 2008) and high number of motorcycles in the queue increases start-up
lost time (Rongviriyapanich, Rongviriyapanish, & Sompakdee, 2010) during the initiation
of green phases. Asaithambi, Kumar, and Sivanandan (2015) suggest that for MTW-domi-
nated traffic (70% MTWs) the discharge rates can be inherently increased (less delays) with
the provision of exclusive stopping space for motorised two-wheelers near the stop line.
But the increased exposure and high accelerating capabilities (Dey, Nandal, & Kalyan, 2013;
Hsu et al., 2003) of MTWs near the stop line makes MTW riders more vulnerable to poten-
tial crashes (Haque et al., 2008), which in particular needs special attention.

Dynamic virtual lane


The terminology “dynamic motorcycle lane” is defined as an unstable imaginary lane
(Sermpis & Spyropoulou, 2007; Sermpis, Spyropoulou, & Golias, 2005) that varies in accord-
ance with the position of subject MTW and rest of the traffic (Minh et al., 2012). The width
of a dynamic lane is related to the operating space of the subject MTW, which is defined as
the minimum lateral clearance between two vehicles moving abreast. Some researchers
have also emphasised on assessing the space occupied by different other vehicle types
on the road (Arasan & Koshy, 2005; Dey, Chandra, & Gangopadhaya, 2006).
Determination of dynamic lane width is considered to be of paramount importance for
the design of an exclusive motorcycle lane. Hussain et al. (2005) computed the physical
width of a static MTW and the average width of the operating space required by an
MTW as 0.8 m and 1.3 m, respectively. They also underlined that MTW paths wider than
1.7 m can enable two riders to conveniently pass each other. The width of the operating
space is however strongly correlated with the speed of MTW. Lee (2008), Nikias, Vlaho-
gianni, Lee, and Golias (2012) and Minh et al. (2012) suggested an increasing linear
trend to describe the relationship between operating space width and speed of the
MTW. A minimum lateral clearance of 0.8 m was used by Minh et al. (2012) to determine
the dynamic lane width of motorcycles, the equation of which was expressed as
212 S. DAS AND A. K. MAURYA

lw = 0.07V + 0.80, where lw is the dynamic lane width of motorcycle (m) and V is the
motorcycle’s average speed (m/s). In another study, Nikias et al. (2012) modelled the over-
taking and filtering manoeuvres of MTWs by relating the dynamic lane width to a number
of traffic determinants (speed, spacing, traffic mix, existence of platoon, etc.). Interestingly,
virtual lane width during filtering was obtained to be significantly higher than during
overtaking.

Filtering
Due to narrow widths and flexible manoeuvrability, MTWs, in general, attempt to utilise
the available lateral spaces described by stationary or almost stationary vehicles near a sig-
nalised intersection and consequently move to the head of a queue using these lateral
gaps, usually termed as lane filtering (passing between stationary vehicles; FEMA, 2009),
lane splitting (passing between moving traffic; FEMA, 2009), lane sharing (Sperley &
Pietz, 2010), creeping (Fan & Work, 2015) and seepage (Agarwal & Lämmel, 2015;
Oketch, 2000; Oketch, 2003) in the literature. The filtering behaviour offers advantages
in an overall trip time reduction, increase in MTW speed and acceleration at traffic
signals. Several studies have attempted to understand the filtering behaviour of MTWs
in heterogeneous traffic streams. Vlahogianni (2014) evaluated the determinants that sig-
nificantly affect MTW riders’ decision to accept critical virtual lane widths during filtering.
Critical lane width is defined as the minimum lateral clearance between any two vehicles
which majority of the MTW riders utilise while filtering. Some of the parameters that influ-
ence riders’ decision are relative speed, spacing, heavy vehicles’ presence and occurrence
of platoon of moving MTWs. On the contrary, filtering behaviour poses safety concerns, as
other vehicles fail to detect the presence of MTWs in the traffic (Hublart & Durand, 2012;
Rogé, Ferretti, & Devreux, 2010) due to its compact size.

Overtaking
In addition to filtering, overtaking has been identified as the most frequent MTW
manoeuvres (Nikias et al., 2012), more insights into which can be useful for understand-
ing traffic flows, associated risks and likelihood of criticalities. Most of the research
works focussing on overtaking manoeuvres consider speed of the subject vehicle,
speed difference with the front vehicle and spacing as the most contributing factors
(Barmpounakis et al., 2016b). The lateral distances and speeds maintained by MTWs
during overtaking manoeuvres have also been demonstrated as useful variables in
some literature. A study by Minh, Sano, and Matsumoto (2005) found that MTWs
attain higher speeds during overtaking in undivided roads when compared to an exclu-
sive motorcycle lane. Also the average lateral distances in exclusive motorcycle lane
and undivided roads are 1.93 m and 1.50 m, respectively. Another study by Law and
Sohadi (2005) estimated the safe lateral distance during overtaking as 1.44 m in an
exclusive MTW lane. Other critical parameters that are addressed in MTW overtaking-
related literature are occupancy (number of persons on a MTW), gender, acceleration,
movement angle, existence of a platoon and oblique gap (Barmpounakis et al.,
2016b; Ghaffari, Alimardani, Khodayari, & Sadati, 2011; Minh et al., 2012; Vlahogianni,
2014; Vlahogianni, Yannis, & Golias, 2013).
TRANSPORT REVIEWS 213

Summary
With the increasing share of motorised two-wheelers in many countries, researchers
are becoming increasingly interested to comprehend the unique manoeuvring pat-
terns of MTWs on road traffic and also to identify the factors that significantly
affect their manoeuvres. The preceding section offers a depiction of various factors
that different researchers have used. Some key findings from the literature are
briefed as follows:

. Longitudinal gap and speed are the most important factors that are used to describe
different riding patterns of MTWs. Among other determinants, lateral clearance and
speed difference between the MTW with another vehicle are the contributing factors
to describe overtaking and filtering behaviours.
. The relationship between heavy vehicle involvements with overtaking manoeuvres
of MTWs is unclear. In particular, the common intuition is that heavy vehicles pres-
ence will affect the overtaking manoeuvres because at higher speeds of trucks,
MTWs may choose to adopt a tailgating behaviour rather than to overtake. But
the overtaking behaviour of MTW riders in presence of heavy vehicles still needs
to be assessed.
. High occupancy reduces frequency of overtaking manoeuvres; male riders are engaged
in more risk taking manoeuvres than female riders. Age of the riders, gender, occu-
pancy, helmet usage, traffic situations and weather conditions may have effects on
different movement patterns of motorised two-wheelers, frequency of lateral move-
ments, overtaking, filtering, etc., which can be further explored.
. Travel time and speed seem more important for MTW riders as they always look for
opportunities to creep through the available gaps in order to minimise their travel
times, which indeed results in a heightened risks for road crashes. Hence, efficient strat-
egies for an increased safety of MTWs require special concern.

On the whole, this overview directs the researchers to conduct research on the areas
which needs further exploration and various additional determinants that can be
employed to describe MTW riders’ behaviours. This would comprise the basis for the
further development of a more realistic traffic flow model.

Modelling approaches
On the modelling of MTW traffic, researchers have adopted different approaches to rep-
resent the characteristic behaviour patterns of MTWs, which indeed ameliorates the repli-
cation of riders’ behaviours and development of microsimulation models as well. The
modelling approaches can be categorised as: cellular automata, agent-based modelling,
social force, safety space, porous flow and discrete choice approach. A review of the
works undertaken on the modelling approaches, potential benefits and certain issues of
the modelling approaches that need further considerations to represent mixed traffic
streams are discussed in this section. Table 1 summarises the modelling approaches pre-
senting the basic assumptions, potentials of each approach and issues that need further
consideration.
Table 1. Summary of all the modelling approaches.

214
Modelling approaches
CA Agent-based modelling Safety space model Social force approach Porous flow approach Discrete choice

S. DAS AND A. K. MAURYA


Related Lan and Chang (2003, 2005), Lee (2008), Agarwal et al. Nguyen, Hanaoka, and Nguyen and Hanaoka (2011); Nair et al. (2011); Ambarwati, Pel, Shiomi, Hanamori, and
literature Lan and Hsu (2005, 2006), (2015, 2016), Agarwal and Kawasaki (2012, 2014), Huynh, Boltze, and Vu Verhaeghe, and Van Arem (2013, Uno (2012); Lee,
Hsu, Lin, Chiou, and Lan Lämmel (2015, 2016) and Nguyen and Hanaoka (2013), Huang, Fellendorf, 2014) Polak, and Bell
(2007), Meng et al. (2007), Lee and Wong (2016) (2012) and Nguyen (2012) and Schoenauer (2012), (2009), Lee and
Mallikarjuna and Rao Huang and Fellendorf Wong (2016) and
(2009) and Lan, Chiou, Lin, (2012), Babu, Vortisch, and Minh et al. (2012)
and Hsu (2009, 2010) Mathew (2014) and Hsu
and Chen (2015)
Salient Road section is divided into Each agent has specific Safety space for a vehicle has Social forces for each MTW Each vehicle proceeds through a series Models the decision-
presumptions cells of constant size. goals, characteristics, the form of a half-ellipse. generally comprise of of pores defined by other vehicles. making process of
Vehicle position is updated memory, rules governing Vehicle within the safety Acceleration force to move Class-specific speed, flow, density riders on the basis of
based on their behaviourActions and space is presumed to have in a desired direction diagrams are defined on the basis of utility maximisation
acceleration/ interactions of different a level of safety Repulsive forces from pore-space distribution theoryEmploys
deceleration rules agents in the traffic flow surrounding vehicles and maximum likelihood
can be simulated static obstacles technique for
parameter
estimation
Applicability of Manoeuvrability model for Microscopic modelling Model to describe swerving Driver behaviour model to Continuum model to explain the Decision-making
the model MTWs in mixed traffic technique to simulate the and oblique following describe integrated lateral creeping phenomena of smaller model to describe
streams using different 2-dimensional manoeuvres of MTWs in and longitudinal vehicle vehicles in heterogeneous queues the position choice,
cell sizes manoeuvres, seepage congested regime movements desire of passing
behaviour, queue manoeuvres, gap-
formation, etc. of MTWs. acceptance choices,
etc. of MTWs
Capabilities Very simple Easy to Programming is simple Simple to use Easy to calibrate Simple to understand Models Reasonable theory Replicates the high Easy to understand and
implement Computationally Simulation is easy. Open- Less parameters as the two-dimensional degree lateral movement of smaller immense flexibility in
efficient source packages are compared to social force movement of vehicles of vehicles in non-lane-based traffic handling different
readily available model non-lane-based traffic factors
Easy to calibrate
Specific cases to Refined cell system however Simulating a large number Cannot capture overtaking, Complexity in calibrating the Vehicle heterogeneity in driver Vehicle heterogeneity
be considered increases the of agents at a lower level filtering, grouping behaviour parameters. Calibration of behaviour and vehicular and interactions
computational burden may prove to be Interaction of MTWs with parameters with real data characteristics within the stationary between different
due to enhanced computationally intensive. other vehicle types need to are not carried out. Number state is not considered Temporal vehicles need to be
resolution of finer cell be explored of parameters increases evolution of pore-space distribution is studied.
systemLateral movement of when different vehicle not considered Representation of
MTWs cannot be captured types are incorporated macroscopic
preciselyInhomogeneity in characteristics needs
vehicle types and their further investigation.
interaction needs further
consideration.
TRANSPORT REVIEWS 215

Cellular automata
A CA model of traffic flow was first proposed by Nagel and Schreckenberg (Na-Sch) in
1992, wherein they depicted the basic features of traffic flows treating only cars on a
single lane traffic stream. Space, speed, acceleration and time were considered as discrete
variables where each discretised cell of 7.5 m length could be either empty or occupied by
a single vehicle. The state of the road at a given time step was attained from one time step
ahead by applying acceleration, braking, randomisation and driving rules for all vehicles.
The single lane Na-Sch model is however inadequate for realistic modelling purposes. A
considerable number of CA models have been modified in a later date (Bham & Benekohal,
2004; Chowdhury, Wolf, & Schreckenberg, 1997; Kerner & Rehborn, 1996; Knospe, Santen,
Schadschneider, & Schreckenberg, 2000; Mallikarjuna & Rao, 2009; Nagel, 1996; Nagel,
Wolf, Wagner, & Simon, 1998). Recently, CA models have gained increasing attention for
its inherent capability in explicating the traffic behaviours of MTWs and other vehicle
types in mixed traffic streams.
Lan and Chang (2003) first developed an inhomogeneous particle-hopping CA model
for describing the manoeuvrability of MTWs and their interactions with cars in urban
mixed traffic streams on 2.5 and 3.75 m lanes using each cell unit as 1.25 × 1.25 m (Lan
& Chang, 2005). They defined motorcycle equivalents (me) for different car percentages
and flow mixes and found that me values range from 2.63 to 5.27 for 10–100% of car
mixes. Lan and Hsu (2005, 2006) introduced a common unit (CU) of squared grid 1.25 ×
1.25 m to represent a fine cell and a fine site for gauging the non-identical vehicle sizes
and non-identical vehicle widths respectively to describe different vehicle types and
their required clearances for safe movements on different lane widths. Based on their
field observations in Taiwan, an MTW was represented by 1 × 2 cells, taking up 2 sites
of road space whereas a car and bus were represented by 2 × 6 cells and 3 × 12 cells,
respectively. The same concept was further extended by Hsu et al. (2007) by reducing
the CU size to a grid of 1.25 × 1.0 m and they introduced the generalised spatiotemporal
definitions of occupancy, flow and speed to precisely capture the collective behaviours of
traffic flow patterns. The simulation results demonstrated that their CA models are able to
capture the effects of both stationary and moving bottlenecks on traffic flows. Another
study by Meng et al. (2007) incorporated cars and MTWs into simulation by categorising
a single lane into three virtual sub-lanes to reflect the parallel movements of two MTWs
within a single lane in which each MTW was assumed to occupy a cell of 3.75 m length
(length of motorbike, 2 m + some safety distance). Lan et al. (2009) revised the conven-
tional CA particle hopping discrete speed variation to be a piecewise linear speed variation
in order to more accurately explicate the abrupt deceleration behaviour of vehicles while
approaching traffic jams or stationary obstacles or signalised intersections. They proposed
a refined CA model (1.25 × 1.0 m) for cars and introduced the concept of limited decelera-
tion capability of vehicles into their CA simulations in order to ascertain safe distance with
the front vehicles.
They further developed a sophisticated CA model to elucidate the erratic MTW beha-
viours in mixed traffic conditions (Lan et al., 2010) which, in addition to the conventional
longitudinal movement and lane-changing rules, the model can also explain the filtering
behaviour of MTWs between two moving cars and the transverse crossing behaviour of
MTWs through the gap between two stationary cars in the same lane. According to
216 S. DAS AND A. K. MAURYA

their simulation results, the CA model can precisely illustrate the unique lateral move-
ments of MTWs in mixed traffic contexts. In addition, the simulations revealed that intro-
duction of more MTWs impair the cars’ movement, thereby resulting in a lower traffic flow.
This phenomena is found to be consistent with the results demonstrated by Meng et al.
(2007).
Owing to the enhanced resolution of finer cell system, a precise representation of the
traffic conditions can be attained but at the same time, it can significantly increase the
computational burden, thereby reducing the simulation efficiency. In addition, deciding
a proper cell size for simulating mixed traffic is more crucial because the dynamic charac-
teristics of vehicles such as speed, acceleration and deceleration are highly affected with
the length of the cell. The cell-based simulation approach is based on the assumption that
each lane-changing manoeuvre is accomplished in the immediate next time step (1 s) but
in real-world traffic scenario, a typical lane change requires the involvement of a multiple
of time steps. Considering the applicability of CA models to describe mixed traffic streams,
it is still a matter of concern whether the simple rules can suitably replicate the dynamic
behaviour of different vehicle types. A trade-off between simulation efficiency and realism
of simulation still needs further evaluation.

Agent-based modelling approach


Agent-based modelling is a powerful microscopic simulation modelling technique in
which the actions and interactions of a cluster of computational agents (Franklin & Graes-
ser, 1997; Langton, 1989) in the traffic flow are simulated repeatedly over time. An agent
has specific characteristics, memory and goals to identify its neighbourhood and meet the
design objectives. This modelling is appropriate when agents are heterogeneous, inter-
actions of agents are complex, nonlinear; and agents exhibit complex behaviour (Bona-
beau, 2002).
Substantial research work exists to model the seepage behaviour, queue formation,
decision-making process and complex interactions of MTWs using agent-based modelling
technique. Lee (2008) adopted the concept of complexity theory (Bar-Yam, 1997) to model
the complex behaviour of MTWs simply by describing the basic rules of their movements.
The author integrated the swerving/unswerving manoeuvre of MTWs moving behind
another vehicle in a lane, the safety distance maintained by a vehicle while following obli-
quely and the path choice model to choose a desired path (left, straight or right) into an
agent-based simulation model system (BikeSim) to capture the characteristic movement
patterns of MTWs. The simulator was able to reproduce the chaotic trajectories and the
dynamic lane-based behaviour patterns of MTWs. Agarwal et al. (2015) used a multi-
agent transport simulation framework (MATSim) to simulate cars and MTWs in mixed
traffic conditions, based on a queue model which employs the earliest-link-exit-time
approach, enabling faster vehicles to overtake slower vehicles in uncongested regime.
They validated the model with the help of fundamental diagrams (FDs) in which the pas-
senger car unit (PCU) for each vehicle type is estimated using the effective area method
(Agarwal, Zilske, Rao, & Nagel, 2013). According to the simulation results of mixed traffic
flows (with an equal modal split in PCU for cars and MTWs), the study demonstrated
that in the free speed regime, MTWs and cars maintained their respective free speeds
whereas at congested regime, speeds of cars were noticeably higher than that of
TRANSPORT REVIEWS 217

MTWs, which is indeed unrealistic. Agarwal and Lämmel (2015) modelled the seepage
behaviour of MTWs in the congested regime in which they introduced the concept of
backward travelling holes (resembling with Newell’s simplified kinematic wave model;
Newell, 1993) that results inherently due to the space (or hole) freed by a leaving
vehicle at the downstream end of the link. The seepage behaviour was validated with
the help of FDs and spatiotemporal plots but the congested regime of FDs for MTWs
still remained unclear. Therefore, this work was further extended by Agarwal and
Lämmel (2016) to validate the seepage behaviour with real-world data of Patna, India.
The simulation results revealed that the seepage behaviour of bikes (slower mode)
resulted in a significant decrease in average trip time for bike mode and an increase in
trip time for cars and MTWs. Agarwal, Lämmel, and Nagel (2016) extended the queue
model with backward travelling holes to simulate mixed traffic behaviour. The study
revealed that the queue model with holes can explicate the realistic behaviour of mixed
traffic at large-scale scenarios. The agent-based technique was further used by Lee and
Wong (2016) to simulate the queue formation process of MTWs in heterogeneous traffic
near a signalised intersection using the BikeSim simulator. The simulator could represent
the lateral position choice and longitudinal movements of MTWs while approaching a
queue. They further demonstrated that the agent-based simulator could integrate the dis-
crete choice models for the microsimulation of heterogeneous traffic. As listed in many
articles, agent-based modelling framework has also been extensively used to simulate
complex traffic phenomena by modelling lane-changing (Hidas, 2002, 2005), car-following
behaviour of different vehicles (Chong, Abbas, & Medina, 2011; Panwai & Dia, 2005), multi-
vehicular interaction (Lee, Polak, & Bell, 2008, 2009), modelling the aggressive level of
drivers (Paruchuri, Pullalarevu, & Karlapalem, 2002), etc.
The widespread use of agent-based approach to simulate complex systems is due to its
simplicity in the programming technique, ability to recognise and distinguish the attri-
butes of other agents and ease in decision-making process. Simulating a large number
of agents at a lower level to describe the aggregate behaviour can however prove to
be extremely computational intensive. Hence a proper balance between computational
power and execution speed need to be attained to account for an improved performance
of the model.

Social force approach


The concept of social forces was pioneered by Helbing and Molnár (1995) to explain the
movement patterns of pedestrians (Helbing & Tilch, 1998; Helbing, Molnár, Farkas, & Bolay,
2001). Later this approach has been employed for modelling MTW movements as well.
These social forces are the measures of instantaneous psychological behaviours experi-
enced by the drivers either to accelerate/decelerate or change direction in order to
avoid collisions with other vehicles or static obstacles. Modelling the behaviour of
MTWs using social force approach was first proposed by Nguyen and Hanaoka (2011).
The social forces for each MTW essentially comprise three kinds of forces (depicted in
Figure 1), namely

(a) A driving force or an acceleration force FaA leading the MTW, α to move in a specified

direction ea oriented towards the destination (straight, left or right) from the actual
218 S. DAS AND A. K. MAURYA

Figure 1. Illustration of social forces for the subject MTW, α.


speed 
va to desired speed va0 ea within a relaxation time ta ,
 
FaA = (va0 ea − 
1
va ).
ta

R
(b) A repulsive force from surrounding vehicles Fab to keep a safe lateral distance from
another vehicle, β in order to avoid collisions: closer the surrounding vehicle, the
semi-minor axis of the ellipse becomes smaller, more is the reduction in the subject
vehicle’s speed, more is the tendency to move away from the surrounding vehicle
and consequently repulsive force gets bigger.
 
R
Fab = Vab (b( rab )),

  
where rab = ra − rb is the distance vector between motorcycle α and motorcycle β.
Repulsive function Vab (b) is monotonically decreasing function of the semi-minor axis
of the ellipse b.

(a) A repulsive force from the infrastructural boundaries (median, strip, guardrail) FaB so as
to avoid hitting the static obstacles: repulsive force increases when vehicles are closer
to the boundaries
 
FaB = UaB ( rab ),

  
where rab = ra − rb indicates the distance from location of motorcycle to border B,
which is nearest to the motorcycle.
TRANSPORT REVIEWS 219

R
(a) A repulsive force for an angle of sight Fab that motorcycle drivers cannot see. A
weight factor of 1 is given if a motorcycle β is in angle of sight and if it is out of
the angle of sight, weight factor becomes c, 0 < c < 1. The repulsive force of other
vehicle is given as

 
R
Fab = wVab (b( rab )).

The equation of motion is then defined as the total of social forces, which are equal to
the change of actual speed va within certain time dt.

d
va    
= FaA + R
Fab + FaB .
dt b b

In addition to acceleration and repulsive forces, Huynh et al. (2013) included attrac-
tive force to explain the grouping behaviour of MTWs near a signalised intersection.
Vehicle trajectories have mostly been simulated using VISWALK software where
majority of the parameters have been modified according to trial-and-error basis
(Huynh et al., 2013) but calibration of the parameters using real field data could sig-
nificantly improve the model performance. Moreover, VISWALK cannot suitably assim-
ilate the attractive force in the model to describe the vehicle grouping behaviour. The
turning behaviour, obstacle passing/overtaking and heterogeneous queue forming
behaviour of MTWs and cars near the approach of a T-intersection using social
force model are also deliberated by Huang et al. (2012) and Huang and Fellendorf
(2012). In another study, Babu et al. (2014) introduced the concept of perception
lines to describe the two-dimensional movements of MTW riders in a mid-block
section. They integrated the theories of intelligent driver model and social force
approach to depict the longitudinal and lateral movements of MTW, respectively in
mixed traffic situations. The developed model was able to simulate the trajectories
involving the interaction of an MTW with different vehicle types (i.e. car, MTW,
motorised three wheelers). Hsu and Chen (2015) introduced a force field interactive
concept to describe MTWs’ riding behaviour using a series of attractive and repulsive
forces. In their work, the parameters of the model are calibrated by maximum likeli-
hood method and validation of model showed mean absolute percentage errors in
speed as 33.8%, which is quite high.
The social force approach has emerged as an extensively used modelling technique
which can suitably replicate the integrated lateral and longitudinal behaviour of MTWs
on a two-dimensional space. Addition of more attractive/repulsive social force com-
ponents can provide detailed understanding of vehicle movements but at the same
time, increasing number of social force components can induce difficulty in the calibration
of a large set of parameters associated with the components of each social force. More-
over, inhomogeneity in vehicle-types and interaction of different types of surrounding
vehicles with MTWs needs to be considered in the future research work without compro-
mising the computational efficiency.
220 S. DAS AND A. K. MAURYA

Safety space model


Nguyen (2012) first proposed the concept of safety space to describe the dynamic behav-
iour of MTWs under congested traffic conditions. Safety space for an MTW is the space sur-
rounding a single subject MTW that has the form of an approximate half-ellipse, the
boundaries of which are equipotential lines where all vehicles on the line are presumed
to have the same level of safety as perceived by the subject vehicle (presented in
Figure 2).
The acceleration and deceleration behaviour of motorcyclists in response to changes in
size of safety space are described using a simple model with few parameters. If the influ-
ential vehicle moves in or out of the safety space, the subject vehicle will either decelerate
or accelerate. The model assumes that a subject vehicle α responds to the most influential
vehicle β by adjusting its acceleration, which equals the gradient ∇ of a level of safety Vb
for the current position of the influential vehicle with respect to a relative speed vector

vab =  va − vb . The level of safety for vehicle β can be described as an exponential func-
tion of the distance between two vehicles and the acceleration of vehicle α can be formu-
lated as

aa (t + Ta ) = −∇  Vb (t),
vab

   
x2 y2
aa (t + Ta ) = cases − ∇  A exp − + /B , if x
vab (ta va )2 (Wa + dy )2
   
y2
≥ 0 − ∇  A exp − /B , if − 2dx ≤ x ≤ 0cases
vab (Wa + dy )2

where Ta is the reaction time, ta is the relaxation time or the time needed to avoid a col-
lision either by braking or swerving, (ta va ) is the length of the semi-major axis, Wa is the
lateral distance on the y-axis between two MTWs, dx and dy are the physical size of the
MTW on each axis, A and B are the parameters where A represents the magnitude of

Figure 2. Safety space for the subject motorcycle, α (Nguyen et al., 2012).
TRANSPORT REVIEWS 221

safety level and B represents the tendency of safety level to increase when both the
vehicles come near each other.
The proposed safety space model could correctly describe more than 61% of left/ right
swerving manoeuvres. This model assumes that the vehicle will be the most influential if
the subject vehicle responds to it with the maximum magnitude of acceleration. The fol-
lowing angle and the width of route to follow a leader were also considered to affect the
selection of the most influential vehicle (Nguyen, Hanaoka, & Kawasaki, 2014). Motorcy-
clists rather prefer to follow a leader at a smaller following angle and the furthest
vehicle ahead provided there are no vehicles in between. The similar concept was used
by Nguyen et al. (2014) to describe the behaviours of increasing or decreasing speeds
by introducing two new parameter sets for acceleration and deceleration. The new
model correctly reproduced more than 63% of left or right swerving manoeuvres and
58% of acceleration behaviours (Nguyen et al., 2012, 2014; Nguyen & Hanaoka, 2012).
This model was extended to comprehend the interactions of cars and MTWs in
which MTWs were observed to follow cars at a smaller headway than that of cars
(Nguyen, 2012).
The safety space model has been found to suitably describe the acceleration, oblique
following and swerving manoeuvres of the MTWs in congested traffic stream. Akin to
social force approach where the subject vehicle maintains a safety sphere from the sur-
rounding vehicles in the form of an ellipse, the safety space model utilises the same
concept and is more suitable because of its simplicity, ease of application and calibration
of a few parameters as compared to social force model. Moreover the safety space varies
across motorcyclist riders with different reaction times but this variability has not been
considered in the studies. Though this model can describe the oblique following and swer-
ving manoeuvres of MTWs, it cannot reproduce other behaviours of MTW traffic such as
filtering and grouping behaviour.

Porous flow approach


A mixed traffic system with weak-lane discipline traffic (where vehicles of different
dimensions do not move within well-demarcated lanes) can be suitably elucidated
on the basis of fluid flow through a porous medium because each vehicle in a
mixed traffic stream often tend to manoeuvre through the available pore space
depending on its vehicle dimensions. To address the creeping phenomena of smaller
vehicles and dynamics of different vehicle classes in heterogeneous queues of weak-
lane-disciplined traffic streams, Nair et al. (2011) developed a continuum model using
an equilibrium speed–density relationship, considering that vehicles traverse through
a series of pores defined by other vehicles on the traffic stream. In this approach,
the complex interactions of vehicular traffic is assessed on the basis of pore-space dis-
tribution for each vehicle class since all vehicle types cannot utilise the same pore
space (larger vehicles cannot use the same pore that smaller vehicles can). The
porous flow model is based on multi-class kinematic wave modelling (Logghe &
Immers, 2008), where each vehicle class in a traffic stream is assumed to have an inde-
pendent fundamental diagram. To account for the lateral dynamics of vehicles, the
overall speed of traffic stream for each vehicle class is considered to consist of free
222 S. DAS AND A. K. MAURYA

vehicles uf (v, x, t) and restrained vehicles ur (v, x, t):



r(v) 
1

u (v, x, t) = ur (v, x, t) fp (rp, x, t)drp + uf (v, x, t) fp (rp, x, t)drp ;


0 r(v)

∀v = 1, 2 . . . n,

where
⎛ ⎞ar

r(v)

ur (v, x, t) = uf (v) ⎝1 − fp (rp, x, t)drp ⎠ ,


0
⎛ ⎞ af

r(v)

uf (v, x, t) = uf (v) ⎝1 − fp (rp, x, t)drp ⎠ ,


0

fp (rp, x, t)drp indicates the fraction of pores with pore size rp , r and f indicates
restricted and free vehicles, ar ≥ af , uf (v) is the free flow speed of vehicle class v.
A full description of the model can be found in Nair et al.’s (2011) research work.
This model was verified by Ambarwati et al. (2013) using empirical data from Indonesia.
In their study, the pore size was found to be dependent on vehicle dimensions and type of
vehicles they interact with. The authors further extended their work in 2014 with a view to
analyse vehicle-type-specific critical pore sizes, develop pore size density distribution and
to produce the vehicle-type-specific speed–density and flow–density diagram for MTW-
MTW, MTW-car and car-car (Ambarwati, Pel, Verhaeghe, & Van Arem, 2014). Results of
the study indicated that the pore size distribution depends on traffic composition and
density; traffic composition also affects the fundamental relationships for mixed traffic
flows.
One drawback of this approach is that it cannot implement more than two vehicle
classes because the analysis becomes cumbersome for the multi-dimensional case. The
pore size distribution for each vehicle class is expected to remain stationary, which may
not always hold true, especially in the case while approaching a signalised intersection.
The evolution of pore-space distribution with time should be explicitly included in the
model. The porous flow model considers equilibrium speed–density relation where tran-
sitions in stationary traffic states are only allowed. The equilibrium assumption however
induces uncertainty in describing the stationary state as there exhibits heterogeneity in
driver behaviour and vehicular characteristics even within the stationary state. Hence
this approach cannot explicitly model the micro-level traffic flow characteristics.

Discrete choice approach


Discrete choice modelling approach has been extensively used to model the decision-
making processes and has been applied to describe the lane-changing manoeuvres,
lateral position choices, desire of passing manoeuvres, gap-acceptance choices, etc. of
MTWs in mixed traffic scenario. The discrete choice approach is easy to understand and
offers immense flexibility in handling different kind of factors as explanatory variables.
TRANSPORT REVIEWS 223

Modelling the riding behaviour of motorcycles using discrete choice approach was first
undertaken by Shiomi et al. (2012). They developed a microscopic model for MTW-domi-
nated traffic by modelling the interactions of cars and MTWs. In the proposed model (orig-
inally developed for pedestrian movements by Robin, Antonini, Bierlaire, & Cruz, 2009),
motorcyclists and car drivers are assumed to make discrete decisions in choosing their
acceleration/deceleration and turning angle simultaneously at each time step from the
defined choice set. The choice set is defined based on the speed and moving direction
of the vehicles. Cross Nested Logit model (CNL) and nested logit model were used to
capture the movement of MTWs and cars, respectively. The proposed model could
capture the interactions of cars and MTWs in a mixed traffic stream; however the lane-
changing behaviour of cars could not be properly captured in the model. This approach
has been employed in many research work to model the decision-making processes of
motorcyclists as well as other drivers. A multinomial logit model (MNL) was used to rep-
resent the lateral position choices of MTWs during queue formation near a signalised inter-
section (Lee & Wong, 2016). Literature reveals that this approach has been extensively
used to describe the willingness of motorcyclists to change lanes (Minh et al., 2012),
desire of vehicles to pass through the available pore between other vehicles (Ambarwati
et al., 2014; Farah & Toledo, 2010; Rossi, Meneguzzer, & Gastaldi, 2013) and to model lane-
changing manoeuvres (Ahmed, 1999; Toledo, 2003). In another study, Choudhury and
Islam (2016) used discrete choice framework for the identification of governing leading
vehicle from multiple leaders, which are more prevalent in mixed traffic conditions.
The discrete choice models mostly employ maximum likelihood technique for estimat-
ing parameters and at the same time the accuracy and validity of the model can be stat-
istically evaluated. It has gained considerable attention in depicting the MTW manoeuvres
but the applicability of this model to describe different vehicle-types along with their inter-
actions needs to be further explored. The representation of the macroscopic character-
istics of mixed traffic using this approach also needs further investigation.

Concluding remarks and research directions


MTW has rapidly emerged as a flexible transport mode in many countries, and transport
modellers are showing increasing interests in understanding and modelling the character-
istic behaviours in order to augment the reliability of the existing models which are used in
safety evaluation, traffic operation studies and in intelligent transport systems. This study
reviewed the predominant factors that are mostly incorporated in MTW modelling and
also the existing modelling approaches that are used extensively for MTW-related
studies. Table 2 presents a list of variables that are extensively used in the models to
capture the behavioural phenomena of MTWs. Many MTW-related research works have
investigated a limited number of factors and a more detailed study on MTW modelling
is needed in order to obtain additional insights into the effects of many determinant
factors with appropriate modelling approaches that can improve the replication and pre-
diction of the manoeuvrability patterns of MTWs. This section briefly highlights some poss-
ible directions for modelling motorised two-wheelers and some of the issues are
presented below:
224 S. DAS AND A. K. MAURYA

Table 2. A summary of variables used for depicting riding patterns of motorcycles using different
modelling approaches.
Modelling Vehicle types Authors/nature of model [if
approach Variables considered Factors studied considered any]
LGDF, LGLF, LGRF; speed SV, RF, LF, MTW, car Lan and Chang (2003)
LB, RB; PSV, PLF, PRF; lateral [particle hopping model];
position (R or L) Meng et al. (2007), Hsu
et al. (2007) and Lan et al.
(2009, 2010)
LGTL(F), LGTL(B); deceleration Longitudinal movement, MTW, car Meng et al. (2007) [single
probability of vehicle lane-changing lane CA model]; Hsu et al.
CA

(2007) and Lan et al.


(2009, 2010)
Time the vehicle stops in jam Car, truck Hsu et al. (2007) [CU-based
CA model]; Lan et al.
(2009, 2010)
Minimum longitudinal Car Lan et al. (2009, 2010)
clearance; safe speed; limited [Piecewise linear
deceleration capability of movement CA model]
vehicles
Gaps in the adjacent sub-lanes Filtering MTW, car Lan et al. (2010) [Refined
CA model]
Speed SV, PV; braking Swerving/unswerving, MTW Lee (2008)
deceleration SV; speculative longitudinal following,
deceleration PV, reaction time; oblique following and
Agent-based modelling

lateral speed; relative speed, path choice behaviours


oblique following angle;
vehicle size; lateral clearance
Free speed travel time for each Filtering in congested MTW, car, Agarwal et al. (2015, 2016)
link, length of the link, regime bike and and Agarwal and Lämmel
maximum speed allowed on truck (2016) [Queue model;
the link, maximum speed of MATSim simulator]
the vehicle, storage capacity,
flow capacity and vehicle
type
Lateral gap of SV with two Queue formation process, MTW Lee and Wong (2016)
adjacent vehicles, oblique creeping phenomena [BikeSim simulator]
front lateral gap, LG to the
stationary vehicle ahead,
vehicle type, speed of the
adjacent vehicles and SV
Acceleration force: Actual and Nguyen and Hanaoka
desired speed SV, relaxation (2011), Hsu and Chen
time (2015), Huynh et al.
Repulsive force of other driver: (2013) and Babu et al.
Distance vector and speed Longitudinal and lateral (2014)
between two vehicles, movement of vehicles
Social force approach

Reaction time, semi-minor


axis of ellipse
Repulsive force from border:
Distance of vehicle’s position MTW
from the border
Attractive force between MTW Vehicle grouping Huynh et al. (2013)
with preceding MTWs: Radius
of SV and PV, distance vector
between centre of SV and PV
Attractive force due to lateral Tailgating, swerving Babu et al. (2014)
gap choice: LH with FV, lateral
distance of SV with FV,
maximum turning angle of
SV

(Continued )
TRANSPORT REVIEWS 225

Table 2. Continued.
Modelling Vehicle types Authors/nature of model [if
approach Variables considered Factors studied considered any]
Speed SV; reaction time; lateral Longitudinal movement, MTW, Car Nguyen (2012), Nguyen
Safety space model

clearance between two oblique following and et al. (2014)


adjacent vehicles; width of swerving
SV; relaxation time; length of
SV; relative speed between
SV and influential vehicle in
longitudinal and lateral
direction; following angle;
following route width
Diagonal opening; critical pore Filtering Overtaking MTW, car Nair et al. (2011) [Multi-
Porous flow model

size; speed of free vehicles class kinematic wave


(leaders) and restricted model] Ambarwati et al.
vehicles (followers) for each (2013, 2014)
vehicle class; areal density;
free flow speed of each
vehicle class; traffic density;
relative speed between
overtaking and overtaken
vehicles
Speed PV, LF, RF; size of PV; Path choice MTW Lee et al. (2009) [MNL]
lateral distance to ready-to-
overtake position; lateral
clearance beside PV; decision
choice of last time step
Turning angle; speed; Change in moving MTW, car Shiomi et al. (2012) [CNL]
Discrete choice

maximum speed; lateral direction


distance from the boundary;
road width; relative speed;
speed of leader; LG;
acceleration/deceleration
LGRL, LGLL, LGCL, occupancy, Lane selection model Gap Minh et al. (2012) [MNL]
gender, existence of HV, acceptance
relative speed of SV with
lead/lag vehicles, distance
headway FV
LG to the stationary vehicle Position choice model in MTW Lee and Wong (2016)
ahead; lateral Gap SV, lateral the queue [MNL]
gap to the FV; history of
lateral movements
Note: LG, Longitudinal Gap; DF, Direct front; LF, Left front; RF, Right Front; SV, Subject Vehicle; LB, Left behind and RB, Right
behind;
P: positions (x−y coordinates) of vehicles; R, Right; L, Left; F, Front; B, Behind; TL(F), Front vehicle in Target lane; PV, Pre-
ceding vehicle and LH, Longitudinal headway.
Oblique following angle→ Angle at which SV follows PV obliquely either to the left or right.
Link→ track (road) at which agents (vehicles) move in the simulator.
Storage capacity→ maximum number of vehicles on the network.
Oblique front lateral gap→ lateral clearance of the FV with LF/RF vehicles.
Relaxation time→ time needed by a vehicle to reach desired speed.
Following route width→ virtual lane width that SV maintains while following preceding vehicles.
Critical pore size→ minimum pore size that vehicle requires to traverse in between two vehicles and between vehicles and
roadside.
Number of each vehicle type in the section × Average dimension of each vehicle type
Areal density = ;
Area of road section
CL, Current Lane; RL, Right Lane and LL, Left Lane.
Occupancy→ No. of persons sitting in MTWs; Existence of HVs: Heavy vehicles (trucks) presence at the front of MTW in the
target lane.
226 S. DAS AND A. K. MAURYA

(i) Manoeuvring patterns: There is a paucity of research in depicting the unique behav-
ioural patterns of MTWs. Albeit, overtaking and filtering manoeuvres, discharge rates
characteristics and exclusive space considerations are the commonly studied
phenomena, little research has focussed on describing tailgating issues, oblique fol-
lowing, vehicle grouping, riding abreast and swerving manoeuvres of MTWs. Table 2
clearly depicts that no modelling approach can explicitly amalgamate filtering, over-
taking, oblique following, swerving and tailgating manoeuvres of MTWs in a single
model. Improved understanding of the unique riding patterns of MTWs may
further help in the development of a better simulation model.
(ii) Modelling determinant factors: A number of determinant factors influencing MTWs
movements are already discussed. Some more contributing factors that have not
been assorted in MTW modelling research but can be further addressed in the
future studies can be categorised into socio-economic factors, including age
group, gender and situational factors such as weather conditions, night driving
and traffic density.
(iii) Modelling approaches: Compared to CA and safety space model, the concept of
porous medium closely resembles vehicle manoeuvres in weak lane discipline
traffic stream where vehicles always tend to utilise the empty pores on the road,
but this model has several shortcomings, which has led to its limited use. Besides, dis-
crete choice model seems flexible to model any decision-making process. Hence an
integration of porous flow concept and discrete choice models using agent-based
modelling can strengthen the prediction of realistic drivers’ behaviours, which can
be explored in the future research.
(iv) Interactions of vehicles: Modelling the interactions of MTWs with other vehicle
classes is another little researched area. Considering the interactions of MTWs with
different vehicle types, such as cars, motorised three-wheelers, buses, trucks and
light commercial vehicles that are more prevalent in mixed traffic streams, will
provide deeper insights into the replication and prediction of drivers’ manoeuvring
patterns, which still needs to be dealt in the future works.

A synthesis of the literature reveals that the high percentage of MTW traffic in many
countries has spurred a growing interest in many researchers to model its manoeuvres,
which has proliferated a number of papers related to MTW modelling in the recent
years. As the trend of MTWs usage continues to grow in an increasing number of countries,
opportunities for future research work will also extend. This paper gives a general review
of modelling motorised two-wheelers and future research may be required to revisit the
applicability of the modelling approaches for two-wheelers to describe other vehicles in
mixed traffic streams.

Disclosure statement
No potential conflict of interest was reported by the authors.
TRANSPORT REVIEWS 227

References
Agarwal, A., Lämmel, G., & Nagel, K. (2016). Traffic and Granular Flow’15 (pp. 419–426). Cham:
Springer International Publishing.
Agarwal, A., & Lämmel, G. (2015). Seepage of smaller vehicles under heterogeneous traffic conditions.
Paper presented at the 4th international workshop on agent-based mobility, traffic and transpor-
tation models, methodologies and applications (ABMTRANS 2015), Procedia Computer Science, 52,
890–895. doi:10.1016/j.procs.2015.05.147
Agarwal, A., & Lämmel, G. (2016). Modeling seepage behvaiour of smaller vehicles in mixed traffic
conditions using an agent based simulation. Transportation in Developing Economies, 2, 1–12.
doi:10.1007/s40890-016-0014-9
Agarwal, A., Zilske, M., Rao, K. R., & Nagel, K. (2015). An elegant and computationally efficient approach
for heterogeneous traffic modelling using agent based simulation. Paper presented at the 4th inter-
national workshop on agent-based mobility, traffic and transportation models, methodologies
and applications (ABMTRANS 2015), Procedia Computer Science, 52, 962–967. doi:10.1016/j.
procs.2015.05.173
Agarwal, A., Zilske, M., Rao, K., & Nagel, K. (2013). Person-based dynamic traffic assignment for mixed
traffic conditions. Conference on agent-based modeling in transportation planning and oper-
ations, Blacksburg, Virginia, 12–11.
Ahmed, K. I. (1999). Modeling drivers’ acceleration and lane changing behaviors. Department of Civil
and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, United
States.
Ambarwati, L., Pel, A. J., Verhaeghe, R., & Van Arem, B. (2013). Empirical analysis of heterogeneous
traffic flow. Proceedings of the Eastern Asia Society for Transportation Studies (EASTS), 9.
Retrieved from www.easts.info/on-line/proceedings/vol9/PDF/P10.pdf .
Ambarwati, L., Pel, A. J., Verhaeghe, R., & Van Arem, B. (2014). Empirical analysis of heterogeneous
traffic flow and calibration of porous flow model. Transportation Research Part C: Emerging
Technologies, 48, 418–436. doi:10.1016/j.trc.2014.09.017
Arasan, V. T., & Koshy, R. Z. (2005). Methodology for modeling highly heterogeneous traffic flow.
Journal of Transportation Engineering, 131, 544–551.
Asaithambi, G., Kumar, R. V. Y., & Sivanandan, R. (2015). Microscopic simulation for modeling exclu-
sive stopping space for motorcycles under non-lane based mixed traffic conditions. Transporti
Europei [European Transport], 57), ISSN 1825-3997.
Babu, F. A. M., Vortisch, P., & Mathew, T. V. (2014). Modelling of motorcycle movements in mixed traffic
conditions. Paper presented at the 94th Annual Meeting Transportation Research Board,
Washington, DC.
Bar-Yam, Y. (1997). Dynamics of complex systems, 213. Reading, MA: Addison-Wesley.
Barmpounakis, E. N., Vlahogianni, E. I., & Golias, J. C. (2016a). Intelligent transportation systems and
powered two wheelers traffic. IEEE Transactions on Intelligent Transportation Systems, 17, 908–916.
Retrieved from http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7328721
Barmpounakis, E. N., Vlahogianni, E. I., & Golias, J. C. (2016b). Vision-based multivariate statistical
modeling for powered two-wheelers maneuverability during overtaking in urban arterials.
Transportation Letters: The International Journal of Transportation Science, 8, 167–176. doi:10.
1179/1942787515Y.0000000020
Bham, G. H., & Benekohal, R. F. (2004). A high fidelity traffic simulation model based on cellular auto-
mata and car-following concepts. Transportation Research Part C: Emerging Technologies, 12(1), 1–
32.
Bonabeau, E. (2002). Agent-based modeling: Methods and techniques for simulating human
systems. Proceedings of the National Academy of Sciences, 99, 7280–7287. doi:10.1073/pnas.
082080899
Chong, L., Abbas, M., & Medina, A. (2011). Simulation of driver behavior with agent-based back-
propagation neural network. Transportation Research Record: Journal of the Transportation
Research Board, 2249, 44–51. Retrieved from http://web.mit.edu/linsenc/www/
SimulationofDriverBehaviorwithAgent-BasedBack-Propagation20NeuralNetwork.pdf
228 S. DAS AND A. K. MAURYA

Choudhury, C. F., & Islam, M. M. (2016). Modelling acceleration decisions in traffic streams with weak
lane discipline: A latent leader approach. Transportation Research Part C: Emerging Technologies, 67,
214–226. doi:10.1016/j.trc.2016.02.010
Chowdhury, D., Wolf, D. E., & Schreckenberg, M. (1997). Particle hopping models for two-lane traffic
with two kinds of vehicles: Effects of lane-changing rules. Physica A: Statistical Mechanics and its
Applications, 235(3), 417–439. doi:10.1016/S0378-4371(96)00314-7
Dey, P. P., Chandra, S., & Gangopadhaya, S. (2006). Lateral distribution of mixed traffic on two-lane
roads. Journal of Transportation Engineering, 132, 597–600. doi:10.1061/(ASCE)0733-947X
(2006)132:7(597)
Dey, P. P., Nandal, S., & Kalyan, R. (2013). Queue discharge characteristics at signalised intersections
under mixed traffic conditions. European Transport/ Transporti Europei, 55(7), 1–12. Retrieved from
http://www.istiee.org/te/papers/N55/ET_2013_55_8_Parthaetal.pdf
Dimitriou, H. T., & Gakenheimer, R. (2011). Urban transport in the developing world: A handbook of
policy and practice. Cheltenham: Edward Elgar Publishing.
Fan, S., & Work, D. B. (2015). A heterogeneous multiclass traffic flow model with creeping. SIAM
Journal on Applied Mathematics, 75, 813–835. doi:10.1137/140977977
Farah, H., & Toledo, T. (2010). Passing behavior on two-lane highways. Transportation Research Part F:
Traffic Psychology and Behaviour, 13, 355–364. doi:10.1016/j.trf.2010.07.003
FEMA. (2009). A European agenda for motorcycle safety: The motorcyclists’ point of view. Technical
report, Federation of European Motorcyclists Associations.
Franklin, S., & Graesser, A. (1997). Is it an agent, or just a program?: A taxonomy for autonomous
agents. In J. P. Müller, M. J. Wooldridge, & N. R. Jennings (Eds.), Intelligent agents III: Agent theories,
architectures, and languages (pp. 21–35). London: Springer.
Ghaffari, A., Alimardani, F., Khodayari, A., & Sadati, H. (2011). ANFIS based modeling for overtaking
maneuver trajectory in motorcycles and autos. 2011 IEEE International Conference on Control
System, Computing and Engineering (ICCSCE), IEEE, 68–73. doi:10.1109/ICCSCE.2011.6190498
Haque, M., Chin, H., & Huang, H. (2008). Examining exposure of motorcycles at signalized intersec-
tions. Transportation Research Record: Journal of the Transportation Research Board, 2048, 60–65.
doi:10.3141/2048-08
Helbing, D., & Molnár, P. (1995). Social force model for pedestrian dynamics. Physical Review E, 51,
4282–4286. doi:10.1103/PhysRevE.51.4282
Helbing, D., Molnár, P., Farkas, I. J., & Bolay, K. (2001). Self-organizing pedestrian movement.
Environment and Planning B: Planning and Design, 28, 361–383. doi:10.1068/b2697
Helbing, D., & Tilch, B. (1998). Generalized force model of traffic dynamics. Physical Review E, 58, 133–
138. doi:10.1103/PhysRevE.58.133
Hidas, P. (2002). Modelling lane changing and merging in microscopic traffic simulation.
Transportation Research Part C: Emerging Technologies, 10, 351–371. doi:10.1016/S0968-090X
(02)00026-8
Hidas, P. (2005). Modelling vehicle interactions in microscopic simulation of merging and weaving.
Transportation Research Part C: Emerging Technologies, 13(1), 37–62. doi:10.1016/j.trc.2004.12.0
Hsu, T., & Chen, Y. (2015). Mobility force model for microscopic simulation of motorcycle traffic behav-
iour. International Conference on Modelling, Simulation and Applied Mathematics (MSAM 2015),
49–52. doi:10.2991/msam-15.2015.11
Hsu, C. C., Lin, Z. S., Chiou, Y. C., & Lan, L. W. (2007). Exploring traffic features with stationary ad
moving bottlenecks using refined cellular automata. Journal of the Eastern Asia Society for
Transportation Studies, 7, 2502–2516. doi:10.11175/easts.7.2502
Hsu, T.-P., Sadullah, A. M. F., & Dao, N. X. (2003). A comparison study on motorcycle traffic development
in some Asian countries: Case of Taiwan, Malaysia and Vietnam. Paper presented at the Eastern
Asia Society for Transportation Studies (EASTS) International Cooperative Research Activity,
Tokyo. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.460.2229&rep=
rep1&type=pd
Huang, W., Fellendorf, M., & Schoenauer, R. (2012). Social force based vehicle model for two-dimen-
sional spaces. Paper presented at the 91st Annual Meeting Transportation Research Board,
Washington, DC.
TRANSPORT REVIEWS 229

Huang, W., & Fellendorf, M. (2012). Social force based vehicle model for traffic simulation. hEART:
European Association For Research In Transportation. Retrieved from: http://transp-or.epfl.ch/
heart/2012/latsis2012_submission_34.pdf
Hublart, A., & Durand, J.-F. (2012). Risks for powered two-wheelers and filtering between lanes on urban
motorways. Paper presented at European Transport Conference 2012, Glasgow, Scotland.
Hussain, H., Farhan, M. A., Umar, R. R., & Dadang, M. M. (2005). Key components of a motorcycle-
traffic system: A study along the motorcycle path in Malaysia. International Association for
Traffic and Safety Sciences (IATSS) Research, 29(1), 50–56. doi:10.1016/S0386-1112(14)60118-7
Huynh, D. N., Boltze, M., & Vu, A. T. (2013). Modelling mixed traffic flow at signalized intersections
using social force model. Journal of the Eastern Asia Society for Transportation Studies, 10, 1734–
1749. doi:10.11175/easts.10.1734
Kanagaraj, V., Asaithambi, G., Toledo, T., & Lee, T.-C. (2015). Trajectory data and flow characteristics of
mixed traffic. Transportation Research Record: Journal of the Transportation Research Board, 2491,
1–11. doi:10.3141/2491-01
Kerner, B. S., & Rehborn, H. (1996). Experimental features and characteristics of traffic jams. Physical
Review E, 53, R1297–R1300. doi:10.1103/PhysRevE.53.R1297
Knospe, W., Santen, L., Schadschneider, A., & Schreckenberg, M. (2000). Towards a realistic micro-
scopic description of highway traffic. Journal of Physics A: Mathematical and General, 33, L477.
Retrieved from: http://stacks.iop.org/0305-4470/33/i=48/a=103
Lan, L. W., & Chang, C. W. (2003). Motorbike’s moving behaviour in mixed traffic: Particle-hopping
model with cellular automata. Journal of the Eastern Asia Society for Transportation Studies,
5, 23–37. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=24DBD975
DD41DBB6B1834B724BA10F91?doi=10.1.1.585.9967&rep=rep1&type=pdf
Lan, L. W., & Chang, C. W. (2005). Inhomogeneous cellular automata modeling for mixed traffic with
cars and motorcycles. Journal of Advanced Transportation, 39, 323–349. doi:10.1002/atr.
5670390307
Lan, L. W., Chiou, Y. C., Lin, Z. S., & Hsu, C. C. (2009). A refined cellular automaton model to rectify
impractical vehicular movement behavior. Physica A: Statistical Mechanics and its Applications,
388, 3917–3930. doi:10.1016/j.physa.2009.05.039
Lan, L. W., Chiou, Y. C., Lin, Z. S., & Hsu, C. C. (2010). Cellular automaton simulations for mixed traffic
with erratic motorcycles’ behaviours. Physica A: Statistical Mechanics and its Applications, 389,
2077–2089. doi:10.1016/j.physa.2010.01.028
Langton, C. G. (1989). Artificial life. Redwood City, CA: Addison-Wesley Publishing Company, 1–48.
Lan, L. W., & Hsu, C. C. (2005). Using cellular automata to explore spatiotemporal traffic patterns.
Proceedings of 10th international conference of Hong Kong Society for Transportation Studies,
Kowloon, Hong Kong, 210–219.
Lan, L. W., & Hsu, C. C. (2006). Formation of spatiotemporal traffic patterns with cellular automaton
simulation. Paper presented at 85th Annual Meeting Transportation Research Board,
Washington DC.
Law, T. H., & Sohadi, R. U. R. (2005). Determination of comfortable safe width in an exclusive motor-
cycle lane. Journal of the Eastern Asia Society for Transportation Studies, 6, 3372–3385. Retrieved
from https://www.miros.gov.my/userfiles/file/10-DeterminationOfComfortableSafeWidthInAnExcl
usiveMotorcycleLane.pdf
Lee, T.-C., Polak, J. W., & Bell, M. G. H. (2008). BikeSim User Manual Version 1.0. (Working Paper). Centre
for Transport Studies, Imperial College London, United Kingdom. Retrieved from www.cts.cv.ic.ac.
uk/documents/publications/iccts01156.pdf
Lee, T.-C., Polak, J., & Bell, M. (2009). New approach to modeling mixed traffic containing motorcycles
in urban areas. Transportation Research Record: Journal of the Transportation Research Board, 2140,
195–205. doi:10.3141/2140-22
Lee, T.-C. (2008). An agent-based model to simulate motorcycle behavior in mixed traffic flow.
Kensington: Imperial College London, United Kingdom. Retrieved from www.cts.cv.ic.ac.uk/
documents/theses/LeePhD.pdf www.cts.cv.ic.ac.uk/documents/theses/LeePhD.pdf
230 S. DAS AND A. K. MAURYA

Lee, T.-C., & Wong, K. I. (2016). An agent-based model for queue formation of powered two-wheelers
in heterogeneous traffic. Physica A: Statistical Mechanics and its Applications, 461, 199–216. doi:10.
1016/j.physa.2016.05.005
Logghe, S., & Immers, L. H. (2008). Multi-class kinematic wave theory of traffic flow. Transportation
Research Part B: Methodological, 42(6), 523–541.
Mallikarjuna, C., & Rao, K. R. (2009). Cellular automata model for heterogeneous traffic. Journal of
Advanced Transportation, 43(3), 321–345. doi:10.1002/atr.5670430305
Meng, J., Dai, S., Dong, L., & Zhang, J. (2007). Cellular automaton model for mixed traffic flow with
motorcycles. Physica A: Statistical Mechanics and its Applications, 380, 470–480. doi:10.1016/j.physa.
2007.02.091
Minh, C. C., Sano, K., & Matsumoto, S. (2005). Characteristics of passing and paired riding maneuvers
of motorcycle. Journal of the Eastern Asia Society for Transportation Studies, 6, 186–197. doi:10.
11175/easts.6.186
Minh, C. C., Sano, K., & Matsumoto, S. (2012). Maneuvers of motorcycles in queues at signalized inter-
sections. Journal of Advanced Transportation, 46(1), 39–53.
Nagel, K. (1996). Particle hopping models and traffic flow theory. Physical Review E, 53, 4655–4672.
doi:10.1103/PhysRevE.53.4655
Nagel, K., & Schreckenberg, M. (1992). A cellular automaton model for freeway traffic. Journal de phy-
sique I, 2, 2221–2229. Retrieved from http://www.pd.infn.it/~agarfa/didattica/met_comp/lab_
20140108/1992_origca.pdf
Nagel, K., Wolf, D. E., Wagner, P., & Simon, P. (1998). Two-lane traffic rules for cellular automata: A
systematic approach. Physical Review E, 58, 1425–1437.
Nair, R., Mahmassani, H. S., & Miller-Hooks, E. (2011). A porous flow approach to modeling hetero-
geneous traffic in disordered systems. Transportation Research Part B: Methodological, 45(9),
1331–1345. doi:10.1016/j.trb.2011.05.009
Newell, G. F. (1993). A simplified theory of kinematic waves in highway traffic, part I: General theory.
Transportation Research Part B: Methodological, 27(4), 281–287.
Nguyen, L. X., Hanaoka, S., & Kawasaki, T. (2012). Describing non-lane-based motorcycle movements
in motorcycle-only traffic flow. Transportation Research Record: Journal of the Transportation
Research Board, 2281, 76–82. doi:10.3141/2281-10
Nguyen, L. X., Hanaoka, S., & Kawasaki, T. (2014). Traffic conflict assessment for non-lane-based
movements of motorcycles under congested conditions. International Association of Traffic and
Safety (IATSS) Research, 37, 137–147. doi:10.1016/j.iatssr.2013.10.002
Nguyen, L. X., & Hanaoka, S. (2011). An application of Social force approach for motorcycle dynamics.
Proceedings of the Eastern Asia Society for Transportation Studies, 8, 1–8. doi:10.11175/eastpro.2011.
0.319.0
Nguyen, L. X., & Hanaoka, S. (2012). A microscopic simulation model for motorcycle traffic safety assess-
ment in Vietnam. Tokyo : Tokyo Institute of Technology, Japan . Retrieved from www.ide.tite-
ch.ac.jp/~hanaoka/JSCE2012haru-3.pdf.
Nguyen, L. X. (2012). A concept of safety space for describing non-lane-based movements of motor-
cycles. Hachioji: Tokyo Metropolitan University, Japan.
Nikias, V. A., Vlahogianni, E. I., Lee, T. C., & Golias, J. C. (2012). Determinants of powered two-wheelers
virtual lane width in urban arterials. Paper presented at the15th International IEEE conference on
intelligent transportation systems, IEEE, 1205–1210. Retrieved from http://ieeexplore.ieee.org/
stamp/stamp.jsp?arnumber=6338619
Oketch, T. (2000). New modeling approach for mixed-traffic streams with nonmotorized vehicles.
Transportation Research Record: Journal of the Transportation Research Board, 1705, 61–69.
doi:10.3141/1705-10
Oketch, T. (2003). Modeled performance characteristics of heterogeneous traffic streams containing
non-motorized vehicles. Paper presented at Transportation Research Board 2003 Annual
Meeting, Washington DC.
Panwai, S., & Dia, H. (2005). Comparative evaluation of microscopic car-following behaviour. IEEE
Transactions on Intelligent Transportation Systems, 6, 314–325. Retrieved from http://ieeexplore.
ieee.org/stamp/stamp.jsp?arnumber=1504791
TRANSPORT REVIEWS 231

Paruchuri, P., Pullalarevu, A. R., & Karlapalem, K. (2002). Multi agent simulation of unorganized traffic.
Proceedings of the first international joint conference on Autonomous agents and multiagent
systems: part 1, 176–183. doi:10.1145/544741.544786
Robin, T., Antonini, G., Bierlaire, M., & Cruz, J. (2009). Specification, estimation and validation of a ped-
estrian walking behavior model. Transportation Research Part B: Methodological, 43(1), 36–56.
doi:10.1016/j.trb.2008.06.010
Rogé, J., Ferretti, J., & Devreux, G. (2010). Sensory conspicuity of powered two-wheelers during filter-
ing manœuvres, according the age of the car driver. Le travail humain, 73, 7–30. doi:10.3917/th.
731.0007
Rongviriyapanich, T., Rongviriyapanish, S., & Sompakdee, P. (2010). Microscopic simulation for mod-
eling effects of motorcycles on traffic operations at signalized intersection. Proceedings of the
Eastern Asia Society for Transportation Studies, 8, 1714–1721. doi:10.11175/easts.8.1714
Rossi, R., Meneguzzer, C., & Gastaldi, M. (2013). Transfer and updating of logit models of gap-accep-
tance and their operational implications. Transportation Research Part C: Emerging Technologies, 28,
142–154. doi:10.1016/j.trc.2011.05.019
Sermpis, D., & Spyropoulou, I. (2007, January). Parameters related to modelling motorcycle movement.
Paper presented at 86th Annual Meeting of Transportation Research Board, Washington, DC, USA.
Sermpis, D., Spyropoulou, I., & Golias, J. (2005). Investigation of the two-wheel vehicle movement at
urban signal-controlled junctions. Paper presented at the 84th Annual Meeting of Transportation
Research Board, Washington DC, USA.
Shiomi, Y., Hanamori, T., & Uno, N. (2012). Modeling traffic flow dominated by motorcycles based on
discrete choice approach. hEART: European Association For Research In Transportation. http://
transp-or.epfl.ch/heart/2012/latsis2012_submission_50.pdf
Sperley, M., & Pietz, A. J. (2010). Motorcycle lane–sharing: Literature review. Technical Report, OR–RD–
10–20, Oregon Department of Transportation Research Section.
Spyropoulou, I., & Sermpis, D. (2009). Performance of junctions with a high motorcycle proportion.
Proceedings of the Institution of Civil Engineers - Transport, 162, 63–69.
Toledo, T. (2003). Integrated driving behavior modeling. Department of Civil and Environmental
Engineering, Massachusetts Institute of Technology, Cambridge, USA.
Vlahogianni, E. I. (2014). Powered-two-wheelers kinematic characteristics and interactions during fil-
tering and overtaking in urban arterials. Transportation Research Part F: Traffic Psychology and
Behaviour, 24, 133–145. doi:10.1016/j.trf.2014.04.004
Vlahogianni, E. I., Yannis, G., & Golias, J. C. (2013). Critical power two wheeler driving patterns at the
emergence of an incident. Accident Analysis & Prevention, 58, 340–345. doi:10.1016/j.aap.2012.12.
026
Walton, D., & Buchanan, J. (2012). Motorcycle and scooter speeds approaching urban intersections.
Accident Analysis & Prevention, 48, 335–340. Retrieved from http://acrs.org.au/files/arsrpe/
MotorcycleandScooterSpeedsatUrbanIntersections.pdf
Wigan, M. (2002). Motorcycles as a full mode of transportation. Transportation Research Record:
Journal of the Transportation Research Board, 1818, 39–46. doi:10.3141/1818-06
Wong, K. I., Lee, T.-C., & Chen, Y.-Y. (2016). Traffic characteristics of mixed traffic flow in urban arter-
ials. Journal of Asian Transportation Studies, 4(2), 379–391.

You might also like