=Paper=
{{Paper
|id=Vol-1664/w19
|storemode=property
|title=Agents meet Traffic Simulation, Control and Management: A Review of Selected Recent Contributions
|pdfUrl=https://ceur-ws.org/Vol-1664/w19.pdf
|volume=Vol-1664
|authors=Giuseppe M. L. Sarné,Maria Nadia Postorino
|dblpUrl=https://dblp.org/rec/conf/woa/SarneP16
}}
==Agents meet Traffic Simulation, Control and Management: A Review of Selected Recent Contributions==
https://ceur-ws.org/Vol-1664/w19.pdf
1
Agents meet Traffic Simulation, Control and
Management: A Review of Selected Recent
Contributions
Maria Nadia Postorino and Giuseppe M. L. Sarné
Abstract—In the last decades, transport demand has increased real-time [11], [12]. Among the ITS applications, an increasing
quickly due to several concurrent factors. The negative impacts of attention is given to the advantages deriving by the adoption
increased demand have many effects on both travelers themselves of the intelligent software agent technologies [13]–[15]. They
and communities and some actions need to mitigate them. To can be defined as autonomous software entities able to deal
this purpose, progresses in different scientific fields as computer
with large, uncertain and or dynamic systems in a centralized
science, electronic, communication as well as studies on new and
more sophisticated traffic simulation models contributed to realize
or distributed way. Software agents have learning and adaptive
Intelligent Transport Systems (ITSs), which provide advanced capabilities [16] as well as attitude to mutually cooperate and
transport services for a better and efficient use of transport net- share their knowledge [17]–[20].
works. The adoption of the software agent technology has given Moreover, software agents can simulate different behaviors
a significant contribution to the ITS development, due to their and, to this aim, different agent classifications have been
capability to both simulate traffic scenarios at different levels provided in the literature, as for instance in [21]–[23]. Agent
of detail and provide intelligent decision-making frameworks. most relevant characteristics of interest for the transportation
Intelligent agents make it possible to study human behaviors and science are: (i) Ability, when agent adopts high-levels of
machine-to-machine interactions with the aim to simulate, control abstraction also in a local perspective. (ii) Adaptivity, when
and manage transportation networks. Given their relevance, in
the last years a great body of researches and surveys have been
agents reveal a context-oriented attitude to interact with the
proposed in the literature on this matter. This paper wants surrounding environment (also in a real-time); (iii) Autonomy,
to contribute by providing an overview of the most significant that is the ability of working without the intervention of users,
advancements produced during the period 2013- 2015. also by incorporating Belief-Desire-Intension (BDI) notions;
(iv) Flexibility, in terms of exhibiting reactivity, pro-activeness
Keywords—Software Agents, Traffic Control, Traffic Manage-
and social ability simultaneously; (v) Intelligence, on the basis
ment, Traffic Simulation.
of their degree of reasoning and capability of learning; (vi) Mo-
bility, when agents have the capability of migrating from a
I. I NTRODUCTION system to another. (vii) Pro-activity, when there is a goal-
In the last decades, the combined effect of several factors oriented approach; (viii) Reactive or Deliberative behavior,
such as growing population and number of vehicles, economic when agents can perceive the environment where they live
conditions, technological improvements has generated a quick and react to its changes; (ix) Self-organization, when they have
increase of the demand for mobility which, in turn, produces the ability of coordinating themselves with the other agents to
negative impacts on the life quality in terms of environmental reach their individual goals; (x) Social-oriented attitude, versus
damages, traffic-jams and waste of time among the most collaboration and cooperation with other agents; (xi) Temporal
significant [1]–[5]. Consequently, there is a need for solving continuity, which identifies the persistence over time; Obvi-
or mitigating these problems. A great number of researchers ously, each agent can show one or more characteristics at the
have investigated on traffic issues in order to improve the per- same time.
formances of the transportation systems, mainly by proposing The aforementioned properties of intelligent software agents
methods to optimize the use of existing resources [6]–[10]. (hereafter simply agents) are particularly suitable to cope
Recent progresses in different scientific fields — e.g., com- with almost all the traffic issues by providing an intelligent
puter science, electronic, communication among the others — decision-making framework [24] for both micro and macro
have made available new tools and techniques addressed to approaches [25]. Furthermore, it is possible to simulate a great
realize Intelligent Transport Systems (ITSs). ITSs include a variety of both complex human behaviors and agent-to-agent
large variety of methods, technologies and models addressed interactions, with the consequent associated decision processes
to provide solutions to the transport system representation, con- at different levels of detail and abstraction [26] so that agents
trol and management. Particularly, several ITSs are addressed often embed sophisticated techniques of Artificial Intelligence
to simulate, control and manage large transport networks in (AI). Finally, mutual interactions among agents and agents
interactions with the environment where they live can be
Maria Nadia Postorino is with the Dept. DICEAM, University Mediterranea conveniently represented, usually as agents’ stimulus answers
of Reggio Calabria, Italy, e-mail: npostorino@unirc.it
Giuseppe M. L. Sarné is with the Dept. DICEAM, University Mediterranea messages [27].
of Reggio Calabria, Italy, e-mail: sarne@unirc.it Due to their flexibility and properties, in the later years
112
2
many studies have been produced on the use of agents for Recently, promising opportunities are disclosed by virtual
transportation issues, by covering several aspects and providing reality and the agent technology is perfect for simulating at a
efficient solutions mainly for ITS purposes. microscopic level the huge amount of mutual interactions tak-
In this paper, an overview of the most recent and interesting ing place among the different simulated entities. For instance,
advancements on the use of agents for coping with transport an interesting multi-agent system, VR-ISSV, simulating traffic
issues in the traffic simulation, control and management fields within a virtual reality platform is presented in [33], [34]. This
are presented. Moreover, to complete this overview some system adopts different types of agents with a hierarchical
statistics have been carried out by considering: (i) the agents modular approach characterized by reuse, reconfigurability
purpose for which they have been exploited; (ii) the specific and scalability. The vehicle behavior is simulated by means
intelligence approach embedded into the agents; (iii) the agent of a fuzzy-logic approach able to carry out the interactions
behaviors. The paper is structured as follows. Section II with both the environment (e.g. road and weather) and the
presents some interesting and recent papers dealing with other vehicles included complex traffic maneuvers as lane
different granularity of transportation applications, while some changing and overtaking. Other useful references still dealing
statistics about the current trends in applying agents to trans- with virtual environment are [35]–[37].
portation issues are treated in Section III. Finally, Section IV In a very near future, autonomous vehicles will be a
draws the conclusions. reality since many car companies are developing real drive
prototypes. A particular proposal, based on the hypotheses of
their forthcoming use, is in [38], where an auction approach
II. M ICRO AND M ACRO S IMULATIONS
to decide the crossing order at an intersection in presence
The decision processes underlying planning and manage- of potential conflicts is discussed. More in detail, an auction
ment activities require the knowledge of the current and or mechanism among vehicles – assisted by agents – at traditional
possible future states of the transportation networks. To this intersections is applied by using vertical and horizontal signals,
end, a great variety of models, architecture and approaches traffic lights and autonomous reservation protocols which give
have been developed over time in order to simulate different priority to the vehicle (i.e. agent) with the best bid. The mech-
aspects of a transportation system (such as traffic flows, anism has been tested at a microscopic level on real city maps.
congestion states, travel demand), to control and handle several In the highly automatized future scenarios, software agents are
network elements under different conditions and constraints fundamental for applying protocols and operational paradigms,
with the final aim to obtain the realistic and useful results also derived from other contexts, as in [38]. Other papers of
addressed to improve traffic conditions and environmental interest on agents and autonomous vehicles in microscopic
affects [28]. simulations, which adopt more common approaches, can be
In such context, the benefits deriving from the adoption of found in [39]–[42].
agents are manifold. In fact, agents can be associated with Another interesting, although very complex, feature is the
several entities involved in the simulation – e.g., travelers, task of reproducing human behaviors as accurately as possible
vehicles, signals, among the others – in order to represent based on real data. To this purpose, different strategies can be
autonomy and intelligence features and then reproduce their adopted, although agents dealing with this type of problems
heterogeneous behaviors and interactions over time. It is often need to enclose complex artificial intelligence software
worthwhile to note that the simulation of large and detailed engines (see the next Section). In this case, an example is
transportation systems requires a considerable amount of com- represented by the traffic simulator described in [43] where
putational resources, which often represented a limit to the agents use fuzzy-logic features for predicting the state of the
dimension and the level of granularity of the simulation itself. transportation system and making possible fast productions of
However, new computational paradigms, as parallel and cloud suitable solutions. The behavior of drivers derives by probe-
computing [29], currently permit to simulate enormous trans- car data for estimating the personality of individual drivers
portation networks in a very detailed and realistic manner [30]. in selecting routes and taking into account various metrics
A common and highly distinctive criterion for classifying of routes as number of turns, travel time and distances. A
transportation agent tools is based on the adopted level of massively parallel execution of this system allowed simulating
detail, e.g. microscopic or macroscopic. Agents fit well with traffic flows in a large metropolitan area. The simulation of
both, although one of the main features of the agent-based human driving behavior by means of an agent-based simulation
simulations is the simplicity in adopting extremely accurate is also the focus of [44] for traffic evacuation in emergency
level of details. Therefore, a large body of agent-based in- management tasks. In particular, the proposed system generates
vestigations is carried out at a microscopic level which could agent’s driving behaviors based on multi-level driving decision
be more complex to realize by using other approaches. On the models and, at each level, several widely used behavior models
contrary, when the study is based on a macroscopic level other are combined together. The simulation allowed studying how
approaches are usually more competitive than agents in terms the network evolves in presence of emergency events in order
of both design and use of computational/storage resources. to identify critical situations and improve existing evacuation
Note that in this overview the mesoscopic approaches [31], plans. Other interesting works implementing human driving
[32] are not discussed because they have characteristics de- behavior by using an agent approach are available in [45]–
pending on how much their granularity is closed to micro or [48]. In this field, a number of works also tried to simulate
macro simulations. pedestrian mobility by starting from the agent simulation of
113
3
individual human behaviors. The simulation of driver and performances only by interacting with its own neighbors.
pedestrian behaviors share some fundamental features. How- Therefore, in this case the road traffic flows are not derived by
ever, in the first case the simulation of individual behaviors simulating vehicles as individual entities but by considering
is more complex, in the second case the most difficult aspect each link (i.e. road) of the transportation network in terms
to simulate is represented by the flows of crowds. In fact, for of interaction with the environment (i.e. vehicles and other
particular problems, such as evacuation procedures, pedestrian roads). A new real-time traffic signal control based on fuzzy
behavior is more random than vehicle behavior which, for logic and wireless sensor network to collect traffic information
example, are constrained to move along roads and must follow is presented in [65] where traffic flows are modeled at a
travel directions. Other remarkable papers on these subjects macro level, as in the previous paper. A cooperative and
are [49]–[51]. hybrid agent for traffic signal control optimizes the green time
Another topic widely investigated when using agents within effective utilization rate. The experimental simulation shows
a microscopic approach is the control and managing of the the capability of this approach to increase the capacity of the
transportation network. For instance, a Newells simplified intersection by reducing the vehicle delay and adapting the
kinematic wave and linear car following models have been green time to the real-time traffic flow changes better than
used in [31] to represent traffic flow states. An agent-based other control techniques.
simulator evaluates traffic dynamics and vehicle emission/fuel
consumption impacts of different traffic management strate- III. C URRENT T RENDS IN A PPLYING AGENTS TO
gies. There are also several scientific contributions, as in [52] T RANSPORTATION I SSUES
and [53]. Some extensions of these studies consider new
social phenomena involving alternatives mobility modes as car- In order to sketch the current trends in applying the agent
sharing and car-pooling [54], [55]. The management of traffic technology to the transportation field, in this section some
signals is another interesting topic considered by many authors. statistical results derived by an investigation carried out on 250
A complex proposal is SURTRAC (Scalable Urban Traffic papers written in the 2013-2015 are presented. Even though
Control) [56], a real-time adaptive traffic signal control system three years are a narrow time, however the quick evolution
integrating a multi-agent platform working in a decentralized, of agent technologies in the last years made the scientific
scalable and reliable manner. Its main peculiarities are that production in the examined period on the top for number and
each intersection allocates its green time – independently and variety. In any case, the statistics presented in this section
asynchronously by the other – based on current incoming should be considered just an indication. More in detail, the
vehicle flows, urban (grid-like) road networks are managed 250 papers have been chosen among the most significant and
with multiple (competing) traffic flows and last, but not the most cited papers; they have been considered and analyzed
least, it operates in real-time. Furthermore, the coordination with respect to: (i) the purpose for which the agent technology
among monitored entities (traffic lights) is realized only at a was adopted; (ii) the type of “intelligence” embedded into the
neighboring level and provides them the ability of balancing agents; (iii) the behaviors of the agents.
incoming flows for establishing larger green waves. In such a The first analysis explored the main goals of the researches
way, the system is highly reactive to quick changes in traffic that adopted the agent technology. To this aim, the papers
conditions and can respond by suitably managing the traffic have been grouped in six classes based on the goal of the
signals. Tests conducted on a nine-intersection road network corresponding paper, although the class borders are fuzzy. The
achieved major reductions in travel times and vehicle emissions first three classes focus only on a main topic, while the second
over pre-existing signal control, and in [57] by integrating three are a combination of the first ones. More specifically, the
also a route choice system. Other valuable papers on the same considered classes are:
theme are [58]–[60]. 1) Modeling: it includes all the contributions mainly fo-
In the transportation field, the use of agents for carrying cused on simulations, e.g. about network states, behav-
out a macro simulation has a minor appeal with respect to the iors, vehicle flows, crowd flows.
microsimulation for the reasons previously specified. However, 2) Control & Management: it collects all those papers
a number of interesting researches there exist, although some- involving monitoring, setting, supervising and similar
times they are combined with a micro simulation acting at activities on the transportation networks such as, for
a different level [61]–[63]. Here, the authors propose a new instance, those involving traffic light cycles, transport
method to derive time-varying tolling schemes by using the demand, fleets and so on.
concept of the Network Fundamental Diagram (NFD). NFD is 3) Planning: it is referred to all the works having the aim
based on the marginal cost pricing for studying the congestion of designing network components, evacuation plans, bus
price. A simulation applied to the test case study of Zurich has schedules, among the others;
been realized by means of agents working at a macroscopic 4) Modeling + Control & Management;
level. In [64] a not intrusive grid of agent-based sensors able 5) Modeling + Planning;
to monitor traffic parameters for determining traffic flows on 6) Control & Management + Planning.
the roads is proposed. Such system adopts a grid of agent- The results of this analysis are reported in Table I, in terms
based sensors, each one associated with a road and able to of absolute values and percentages on the whole number of
collect, analyze and aggregate acoustical vehicle data. Each considered papers, and depicted in Figure 1. As it can be seen,
road-agent, based on a distributed trust-system, improves its the most popular classes are respectively the “Modeling” and
114
4
the “Control & Management” that together represent more than class number percentage
the 70% of all the selected papers. Logic approaches 81 32.4
Fuzzy logic 27 10.8
class number percentage Neural Networks 17 6.8
Modeling 109 43.6 Genetic Algorithm 3 1.2
Control & Management 72 28.8 Reinforcement Learning 45 18.0
Planning 20 8.0 Other techniques 31 12.4
Modeling + Control & Management 26 10.4 Not Specified 46 18.4
Modeling + Planning 17 6.8
Control & Management + Planning 6 2.4 TABLE II. A GENT AI APPROACH .
TABLE I. A GENT PURPOSE .
Fig. 2. Agents AI approach.
those previously listed in Section I, with the exceptions of the
Fig. 1. Agent purpose. “intelligence” behavior – because in this study we considered
only intelligent agents – and “flexibility” – which, in turn, is
After, the typology of AI embedded into the agents has already a composed behavior (see Section I). More in detail,
been considered. Given the number of existing techniques only the considered behaviors are:
the most popular among them have been directly considered. 1) Autonomy;
More in detail, the following seven classes have been defined, 2) Reactive or Deliberative;
namely: 3) Social-oriented;
1) Logic approaches (e.g. first or higher-order logics, 4) Adaptivity;
Markov logic) 5) Self-organization;
2) Fuzzy logic; 6) Pro-activity;
3) Neural Networks; 7) Temporal continuity;
4) Genetic Algorithm; 8) Mobile.
5) Reinforcement Learning; These results are presented in Table III, organized similarly
6) Other techniques; to the other tables and shown in Figure 3. The behaviors more
7) Not Specified (i.e. the used approach is not described adopted by agents are evident, while Self-organization and
or not clearly identifiable). mobility are behaviors that have meet more rarely the interest
Also the results of this second investigation are still reported of researches with respect to the considered sample.
in Table II, in terms of absolute values and percentages on the
whole number of papers considered, and depicted in Figure 2.
IV. C ONCLUSIONS
As the results show, the family of AI techniques within
the logic approaches are mostly implemented, although an Transportation researches and application tools can receive
important part of them just adopt a basic logic. Fuzzy logic and many benefits from the use of agent technology. In fact,
reinforcement learning are the other two main AI techniques complex behaviors and interactions involving users and entities
which meet the preferences of scientists, while a very high can be simulated by implementing sophisticated decision pro-
number of papers do not provide an adequate description of the cesses at different levels of detail and abstraction. Therefore,
implemented AI approach or its description is not sufficiently intelligent software agents are really key factors for a suitable
clear to permit its classification. development of ITS systems.
Finally, the last statistic deals with the behaviors of each In this context, this short overview has discussed and
agent. In fact, each agent usually has more than one behavior classified some selected, very recent and innovative papers
at the same time. More in detail, the considered behaviors are grouped for granularity and analyzed with respect to their aim,
115
5
class number percentage [8] M. N. Postorino and M. Versaci, “Modelling user mode choices by
Autonomy 193 18.3 an ellipsoidal fuzzy approach,” International Journal of Modelling and
Reactive or Deliberative 224 21.2 Simulation, vol. 33, no. 4, pp. 235–243, 2013.
Social-oriented 86 8.1 [9] P. Maheshwari, R. Khaddar, P. Kachroo, and A. Paz, “Dynamic mod-
Adaptivity 184 17.4 eling of performance indices for planning of sustainable transportation
systems,” Networks and Spatial Economics, pp. 1–23, 2014.
Self-organization 20 1.9
[10] M. N. Postorino and M. Versaci, “Upgrading urban traffic flow by
Pro-activity 153 14.5
a demand-responsive fuzzy-based traffic lights model,” International
Ability 58 5.5 Journal of Modelling and Simulation, vol. 34, no. 2, pp. 102–109, 2014.
Temporal continuity 132 12.5
[11] K. N. Qureshi and A. H. Abdullah, “A survey on intelligent transporta-
Mobility 7 0.6 tion systems,” Middle-East Journal of Scientific Research, vol. 15, no. 5,
pp. 629–642, 2013.
TABLE III. A GENTS BEHAVIORS . [12] A. Pell, P. Nyamadzawo, and O. Schauer, “Intelligent transportation
system for traffic and road infrastructure-related data,” International
Journal of Advanced Logistics, vol. 5, no. 1, pp. 19–29, 2016.
[13] B. Chen and H. Cheng, “A review of the applications of agent tech-
nology in traffic and transportation systems,” Intelligent Transportation
Systems, IEEE Trans. on, vol. 11, no. 2, pp. 485–497, 2010.
[14] C. Adam and B. Gaudou, “Bdi agents in social simulations: a survey,”
2016.
[15] J. P. Müller and K. Fischer, “Application impact of multi-agent systems
and technologies: a survey,” in Agent-Oriented Software Engineering.
Springer, 2014, pp. 27–53.
[16] L. Busoniu, R. Babuska, and B. De Schutter, “A comprehensive survey
of multiagent reinforcement learning,” Systems, Man, and Cybernetics
(C): Appl. and Reviews, IEEE Trans., vol. 38, no. 2, pp. 156–172, 2008.
[17] V. Tomás and L. Garcia, “A cooperative multiagent system for traffic
management and control,” in Proc. of the 4th Int. Joint Conf. on
Fig. 3. Agents behaviors. Autonomous agents and multiagent systems. ACM, 2005, pp. 52–59.
[18] F. Wang, “Agent-based control for networked traffic management sys-
tems,” Intelligent Systems, IEEE, vol. 20, no. 5, pp. 92–96, 2005.
the agent intelligence and the agent behaviors. Obviously, other [19] B. Chen, H. Cheng, and J. Palen, “Integrating mobile agent tech-
classifications are possible but we think that the one proposed nology with multi-agent systems for distributed traffic detection and
here represents an equilibrium between the transportation and management systems,” Transportation Research Part C: Emerging
Technologies, vol. 17, no. 1, pp. 1–10, 2009.
the computer science point of views.
[20] S. Sen, A. Biswas, and S. Debnath, “Believing others: Pros and cons,”
in Proc. of the 4th Int. Conf. on Multi-Agent Systems, ICMAS’2000.
V. ACKNOWLEDGMENT IEEE, 2000, pp. 279–286.
This work has been supported by the NeCS Laboratory - [21] G. S. Bhamra, A. K. Verma, and R. B. Patel, “Intelligent software
DICEAM, University Mediterranea of Reggio Calabria. agent technology: an overview,” International Journal of Computer
Applications, vol. 89, no. 2, 2014.
R EFERENCES [22] T. Balke and N. Gilbert, “How do agents make decisions? a survey,” J.
of Artificial Societies and Social Simulation, vol. 17, no. 4, p. 13, 2014.
[1] K. Zhang and S. Batterman, “Air pollution and health risks due to
[23] N. Naciri and M. Tkiouat, “Multi-agent systems: theory and applications
vehicle traffic,” Science of the total Environment, vol. 450, pp. 307–
survey,” International Journal of Intelligent Systems Technologies and
316, 2013.
Applications, vol. 14, no. 2, pp. 145–167, 2015.
[2] E. A. Vasconcellos, Urban Transport Environment and Equity: The case
for developing countries. Routledge, 2014. [24] C. Macal and M. North, “Introductory tutorial: Agent-based modeling
and simulation,” in Proceedings of the 2014 Winter Simulation Confer-
[3] M. Treiber and A. Kesting, “Traffic flow dynamics,” Traffic Flow ence. IEEE Press, 2014, pp. 6–20.
Dynamics: Data, Models and Simulation, Springer-Verlag Berlin Hei-
delberg, 2013. [25] A. L. C. Bazzan and F. Klügl, “A review on agent-based technology for
traffic and transportation,” The Knowledge Engineering Review, vol. 29,
[4] P. P. Dubey and P. Borkar, “Review on techniques for traffic jam no. 03, pp. 375–403, 2014.
detection and congestion avoidance,” in Electronics and Communication
Systems (ICECS), 2015 2nd International Conference on. IEEE, 2015, [26] L. Weng and F. Menczer, “Computational analysis of collective behav-
pp. 434–440. iors via agent-based modeling,” in Handbook of Human Computation.
Springer, 2013, pp. 761–767.
[5] M. N. Postorino, “A comparative analysis of different specifications of
modal choice models in an urban area,” European journal of operational [27] L. Braga, R. Manione, and P. Renditore, “A formal description lan-
research, vol. 71, no. 2, pp. 288–302, 1993. guage for the modelling and simulation of timed,” Formal Description
Techniques IX: Theory, application and tools, p. 245, 2016.
[6] M.N. Postorino and M. Versaci, “A neuro-fuzzy approach to simulate
the user mode choice behaviour in a travel decision framework,” [28] J. Barceló et al., Fundamentals of traffic simulation. Springer, 2010,
International Journal of Modelling and Simulation, vol. 28, no. 1, p. 64, vol. 145.
2008. [29] G. Lu and W. H. Zeng, “Cloud computing survey,” in Applied Mechan-
[7] R. Z. Farahani, E. Miandoabchi, W. Y. Szeto, and H. Rashidi, “A review ics and Materials, vol. 530. Trans Tech Publ, 2014, pp. 650–661.
of urban transportation network design problems,” European Journal of [30] R. K. Kamalanathsharma, H. A. Rakha, and B. Badillo, “Simulation
Operational Research, vol. 229, no. 2, pp. 281–302, 2013. testing of connected vehicle applications in a cloud-based traffic sim-
116
6
ulation environment,” in Transportation Research Board 93rd Annual [48] C. Xiong, Z. Zhu, X. He, X. Chen, S. Zhu, S. Mahapatra, G. L. Chang,
Meeting, no. 14-4260, 2014. and L. Zhang, “Developing a 24-hour large-scale microscopic traffic
[31] X. Zhou, S. Tanvir, H. Lei, J. Taylor, B. Liu, N. M. Rouphail, and simulation model for the before-and-after study of a new tolled freeway
H. C. Frey, “Integrating a simplified emission estimation model and in the washington, dc–baltimore region,” Journal of Transportation
mesoscopic dynamic traffic simulator to efficiently evaluate emission Engineering, vol. 141, no. 6, p. 05015001, 2015.
impacts of traffic management strategies,” Transportation Research Part [49] S. Liu, S. Lo, J. Ma, and W. Wang, “An agent-based microscopic
D: Transport and Environment, vol. 37, pp. 123–136, 2015. pedestrian flow simulation model for pedestrian traffic problems,”
[32] T. Jeerangsuwan and A. Kandil, “Agent-based model architecture for Intelligent Transportation Systems, IEEE Transactions on, vol. 15, no. 3,
mesoscopic traffic simulations,” Computing in Civil and Building En- pp. 992–1001, 2014.
gineering, pp. 1246–1253, 2014. [50] W. L. Wang, S. M. Lo, S. B. Liu, and H. Kuang, “Microscopic modeling
[33] Y. Yu, A. El Kamel, and G. Gong, “Multi-agent based architecture for of pedestrian movement behavior: Interacting with visual attractors in
virtual reality intelligent simulation system of vehicles,” in Networking, the environment,” Transportation Research Part C: Emerging Technolo-
Sensing and Control (ICNSC), 2013 10th IEEE International Confer- gies, vol. 44, pp. 21–33, 2014.
ence on. IEEE, 2013, pp. 597–602. [51] L. Crociani, A. Piazzoni, G. Vizzari, and S. Bandini, “When reactive
[34] Y. Yu, A. El Kamel, G. Gong, and F. Li, “Multi-agent based mod- agents are not enough: Tactical level decisions in pedestrian simulation,”
eling and simulation of microscopic traffic in virtual reality system,” Intelligenza Artificiale, vol. 9, no. 2, pp. 163–177, 2015.
Simulation Modelling Practice and Theory, vol. 45, pp. 62–79, 2014. [52] J. Auld, M. Hope, H. Ley, V. Sokolov, B. Xu, and K. Zhang, “Polaris:
[35] D. H. Liu and H. Su, “Microscopic traffic model in virtual environment,” Agent-based modeling framework development and implementation for
J. of Southwest Jiaotong University, vol. 48, no. 1, pp. 165–172, 2013. integrated travel demand and network and operations simulations,”
Transportation Research Part C: Emerging Technologies, 2015.
[36] T. Ridene, L. Leroy, and S. Chendeb, “Innovative virtual reality appli-
cation for road safety education of children in urban areas,” in Advances [53] M. Beutel, S. Addicks, B. Zaunbrecher, S. Himmel, K. Krempels, and
in Visual Computing. Springer, 2015, pp. 797–808. M. Ziefle, “Agent-based transportation demand management,” 2015.
[37] K. Gajananan, G. Kugamoorthy et al., “An experimental space for [54] M. M. Lopes, L. M. Martinez, and G. H. de Almeida Correia,
conducting driving behavior studies based on a multiuser networked 3d “Simulating carsharing operations through agent-based modelling: An
virtual environment and the scenario markup language,” application, application to the city of lisbon, portugal,” Transportation Research
2013. Procedia, vol. 3, pp. 828–837, 2014.
[38] D. Carlino, S. D. Boyles, and P. Stone, “Auction-based autonomous [55] I. Hussain, L. Knapen, A. Yasar, T. Bellemans, D. Janssens, and
intersection management,” in Intelligent Transportation Systems-(ITSC), G. Wets, “Negotiation and coordination in carpooling: An agent-based
2013 16th International IEEE Conf. on. IEEE, 2013, pp. 529–534. simulation model,” 2016.
[39] M. N. Mladenovic and M. Abbas, “Priority-based intersection control [56] S. F. Smith, G. J. Barlow, X. F. Xie, and Z. B. Rubinstein, “Surtrac:
framework for self-driving vehicles: Agent-based model development scalable urban traffic control,” 2013.
and evaluation,” in Connected Vehicles and Expo (ICCVE), 2014 [57] X. F. Xie, Y. Feng, S. Smith, and K. Head, “Unified route choice
International Conference on. IEEE, 2014, pp. 377–384. framework: Specification and application to urban traffic control,”
[40] J. Wei, J. M. Dolan, and B. Litkouhi, “Autonomous vehicle social be- Transportation Research Record: Journal of the Transportation Re-
havior for highway entrance ramp management,” in Intelligent Vehicles search Board, no. 2466, pp. 105–113, 2014.
Symposium (IV), 2013 IEEE. IEEE, 2013, pp. 201–207. [58] S. T. Rakkesh, A. R. Weerasinghe, and R. A. Ranasinghe, “Traffic
[41] C. Wuthishuwong and A. Traechtler, “Coordination of multiple au- light optimization solutions using multimodal, distributed and adaptive
tonomous intersections by using local neighborhood information,” in approaches,” in Advances in ICT for Emerging Regions (ICTer), 2015
Connected Vehicles and Expo (ICCVE), 2013 International Conference Fifteenth International Conference on. IEEE, 2015, pp. 220–225.
on. IEEE, 2013, pp. 48–53. [59] D. McKenney and T. White, “Distributed and adaptive traffic signal
[42] Q. Jin, G. Wu, K. Boriboonsomsin, and M. Barth, “Advanced intersec- control within a realistic traffic simulation,” Engineering Applications
tion management for connected vehicles using a multi-agent systems of Artificial Intelligence, vol. 26, no. 1, pp. 574–583, 2013.
approach,” in Intelligent Vehicles Symposium (IV), 2012 IEEE. IEEE, [60] M. A. Khamis and W. Gomaa, “Adaptive multi-objective reinforcement
2012, pp. 932–937. learning with hybrid exploration for traffic signal control based on coop-
[43] T. Osogami, T. Imamichi, H. Mizuta, T. Suzumura, and T. Ide, “Toward erative multi-agent framework,” Engineering Applications of Artificial
simulating entire cities with behavioral models of traffic,” IBM Journal Intelligence, vol. 29, pp. 134–151, 2014.
of Research and Development, vol. 57, no. 5, pp. 6–1, 2013. [61] X. Wang, “Simulation system for optimizing urban traffic network based
[44] S. Yuan, Y. Liu, H. Zhang, S. Ae Chun, and N. R. Adam, “Agent on multi-scale fusion,” International Journal of Smart Home, vol. 8,
driving behavior modeling for traffic simulation and emergency decision no. 2, pp. 227–236, 2014.
support,” in Proceedings of the 2015 Winter Simulation Conference. [62] N. Wagner and V. Agrawal, “An agent-based simulation system for
IEEE Press, 2015, pp. 312–323. concert venue crowd evacuation modeling in the presence of a fire
[45] M. Rahman, M. Chowdhury, Y. Xie, and Y. He, “Review of microscopic disaster,” Expert Systems with Applications, vol. 41, no. 6, pp. 2807–
lane-changing models and future research opportunities,” Intelligent 2815, 2014.
Transportation Systems, IEEE Transactions on, vol. 14, no. 4, pp. 1942– [63] N. Zheng, R. A. Waraich, K. W. Axhausen, and N. Geroliminis, “A dy-
1956, 2013. namic cordon pricing scheme combining the macroscopic fundamental
[46] K. Darty, J. Saunier, and N. Sabouret, “A method for semi-automatic diagram and an agent-based traffic model,” Transportation Research
explicitation of agent’s behavior: application to the study of an im- Part A: Policy and Practice, vol. 46, no. 8, pp. 1291–1303, 2012.
mersive driving simulator,” in International Conference on Agents and [64] M. N. Postorino and G. M. L. Sarné, “An agent-based sensor grid
Artificial Intelligence (ICAART), 2014, pp. 81–91. to monitor urban traffic,” in Proceedings of the 15th Workshop dagli
[47] M. Madruga and H. Prendinger, “ico2: multi-user eco-driving training Oggetti agli Agenti, WOA 2014, ser. CEUR Workshop Proceedings, vol.
environment based on distributed constraint optimization,” in Proceed- 1260. CEUR-WS.org, 2014.
ings of the 2013 international conference on Autonomous agents and [65] C. Ma, Y. Li, R. He, and X. An, “Traffic signal fuzzy control approach
multi-agent systems. International Foundation for Autonomous Agents based on green time effective utilization rate and wireless sensor
and Multiagent Systems, 2013, pp. 925–932. network,” Sensor Letters, vol. 12, no. 2, pp. 425–430, 2014.
117