Towards Infrastructure-Aided Self-Organized Hybrid Platooning

Full text

Towards Infrastructure-Aided Self-Organized
Hybrid Platooning
Christian Krupitzer
Universit¨
at W¨
urzburg, Germany
christian.krupitzer@uni-wuerzburg.de
Samy El-Tawab
James Madison University, USA
eltawass@jmu.edu
Michele Segata
University of Trento, Italy
msegata@disi.unitn.it
Sven Tomforde
Universit¨
at Kassel, Germany
stomforde@uni-kassel.de
Martin Breitbach
Universit¨
at Mannheim, Germany
martin.breitbach@uni-mannheim.de
Christian Becker
Universit¨
at Mannheim, Germany
christian.becker@uni-mannheim.de
Abstract—Nowadays, the technical feasibility of au-
tonomous driving is out of the question. This technology
opens the door for several research topics (e.g., information
exchange, security of information) in the IoT domain.
One such application is platooning: coordinated driving
of vehicles in convoys. Therefore, such platoons set up
Internet-of-Vehicles. Although platooning has been mainly
researched on highways, the idea of platooning in smart
cities can have several benefits, such as efficient use
of roads, time-saving through route optimization, and
minimizing traffic in peak times. In this position paper, we
present our vision of a traffic management approach for
combining platooning on highways and urban areas. We
explain the idea of platooning, how to manage it efficiently,
and the differences between platooning on the freeway and
in smart cities.
Index Terms—Internet-of-Things, Smart Traffic, Pla-
tooning, Coordination
I. INTRODUCTION
According to the U.S. Department of Transportation,
94% of all traffic accidents are caused by human errors,
accumulating an annual cost of four billion dollars.
Recent technological advances in computing power,
sensor technology, and wireless communication have
resulted ”Autonomous Vehicles” (AVs) that can sense
their surroundings and make intelligent decisions by
communicating with neighboring vehicles and nearby
infrastructure. Today, it is not the question anymore
whether autonomous driving will be technically feasi-
ble. Waymo’s self-driving cars have driven more than
8,000,000 miles1in the last few years, even in crowded
urban areas. Whereas Waymo relies on expensive state-
of-the-art laser technology, car manufacturers, such as
Mercedes-Benz, use (close to) series-technology as base
for their autonomous driving initiatives. So today, the
technical feasibility leads to another issue: How can we
use autonomous driving efficiently?
Drivers of autonomous vehicles enjoy more comfort
and safety as human errors – which are one of the
main reasons for accidents – are eliminated. But the
real strength in using autonomous driving comes into
1Announced in July 2018: https://waymo.com/ontheroad/
play when smart infrastructure is combined with the
vehicles’ autonomous capabilities through vehicle-to-
infrastructure (V2I) communication. This enables ef-
ficient traffic management through dynamic routing,
adaptive traffic light control (A-TLC), as well as pla-
tooning. Project studies have shown, that cooperative
driving, meaning that vehicles share information, such as
brake or acceleration data, via wireless communication
is superior to autonomous driving based on local sensing
only [1].
In this position paper, we present our vision for
infrastructure-aided and self-organized platooning of ve-
hicles. Whereas most approaches focus on platooning on
highways only, the objective of our work is to provide
an integrated approach for efficient urban and inner-city
traffic management. Therefore, we combine platooning
and smart navigation on highways with platooning en-
abled through dynamic navigation and A-TLC within
cities. We focus on using existing infrastructure and
technology combined with state-of-the-art technology in
autonomous driving [2].
The remainder of the paper is structured as follows.
In Section II, we define platooning and present our
approach. Section III presents our research objectives
and questions. In Section IV, we discuss related work
in platooning and route guidance. Section V depicts our
project status and future work.
II. A HYBRID AP PROACH FO R PLATO ONI NG
In this section, we present a scenario that shows effi-
cient highway and inner-city traffic management through
platooning, dynamic routing, and Adaptive Traffic Light
Controller (A-TLC). The scenario is based on existing
standard technologies which can be commonly found
in cities and on highways, such as cameras, induction
loops, or variable-message signs (VMSs) which are
small matrix signs based on LED technology. This
technology is combined with state-of-the-art Vehicle-to-
Everything (V2X) communication technology as well as
autonomous driving [3].

A. The Principle of Platooning
In accordance with literature (e.g. [4]), we define
platooning as ”spontaneous and dynamic forming of
convoys of vehicles, so called platoons”. Each vehi-
cle drives within a short distance to the next vehicle.
Vehicles need to be driven autonomously or at least
support the driver in holding the distance and keeping
the vehicle within the lane boundaries, which involves
autonomous braking. It can be argued that a single
vehicle is a platoon already. However, as we assume
self-driving vehicles which can drive without support
through a Platooning Control System (PCS), the benefits
of platoons are not valid for single-vehicle platoons.
Therefore, in our view, a platoon consists of at least
two vehicles. Drivers do not have to control forming of
platoons as this is done automatically. The responsibility
for controlling a platoon can be within a specific vehicle
in the platoon or infrastructure-aided [1], [4]. In our
approach, a hybrid PCS manages platooning, i.e., the
highway is divided in sections that are cooperatively
managed in a regional planner-like approach by semi-
autonomous sub-systems of the PCS. The platooning
process is divided into the following activities: (i) find-
ing/joining/creating a platoon, (ii) maintaining platoons
(including lane changing), (iii) leaving a platoon, and
(iv) dissolving. The control system follows the MAPE-K
model known from Autonomic Computing [5]. Accord-
ingly, it integrates (i) monitoring the vehicles and the en-
vironment, (ii) analyzing the need for adaptation actions
– such as joining a vehicle to a platoon or dissolving a
platoon–, (iii) planning these action, and (iv) controlling
the execution of these actions. All modules share a
common knowledge repository. This makes the vehicles
in combination with the control system a self-organized
system. Subsequently, we describe these activities in
more detail.
1) Finding/Joining a platoon: Using the V2I com-
munication infrastructure, vehicles send information that
are relevant for platooning – such as desired speed and
their route – to the PCS. The PCS uses this information
and plans a platoon. It sends the information to the
relevant vehicles. The condition that at least two vehicles
should have a similar route for a threshold time T
where Tis an arbitrary value that can be calculated
depending on the route, and time of the day. Usually,
there should be platoons available if a vehicle enters
the highway or city, so vehicles just join a platoon.
If this is not the case, a vehicle enters the highway
and drives autonomously until a platoon is available. As
soon as a certain number of vehicles is available within
near vicinity, the infrastructure coordinates the formation
process for a new platoon. The vehicles use V2I and
Vehicle-to-Vehicle (V2V) communication as well as
sensors (e.g., distance sensors) for joining a platoon.
Additionally, the PCS delivers information constantly
via V2I communication.
2) Maintaining platoons: Vehicles driving in a pla-
toon use V2V communication for keeping the distance
to the vehicle in front of them. Through V2V and
V2I communication, information is spread very fast,
such that the vehicles are able to react with very small
delay in dangerous situations, such as an incident or
items/people on the track without crashing into each
other. Further, vehicles send their positions to the PCS.
By combining the information from the PCS with inter-
platoon V2V communication, overtaking processes are
possible. Within the PCS, a handover of a platoon’s
coordination is important, as one sub system of the PCS
only controls a region. Further, the PCS is responsible
for merging and splitting of platoons (e.g., at highway
junctions).
3) Leaving a platoon: Usually, the PCS delivers the
information when a vehicle leaves the platoon, (e.g., in
case it has to leave the highway or join another platoon
in the city due to its route). Furthermore, a vehicle
can leave the platoon on its own by signaling it to the
PCS, e.g., for resting purpose on highways or the driver
spontaneously changes the route by taking over control
manually. In both cases, the PCS has to confirm the
leaving request and the vehicle or PCS has to signal it
to the other vehicles within the platoon. Gaps will be
closed automatically by succeeding vehicles.
4) Dissolving a platoon: Through the constant update
of a platoon’s position, the PCS can determine when a
platoon achieve a position where it is dissolved, e.g., a
highway crossing. In this case, the PCS sends messages
to the vehicles for forming new platoons or signaling
the end of the platooning process. A vehicle must wait
for a certain time before joining another platoon.
Figure 1 shows the platooning process on a highway.
In the figure, within platoon (1) a vehicle leaves the
platoon. Platoon (2) overtakes another platoon
(3) and further, an additional vehicle joins platoon
(2). In the following two sections, we motivate the ben-
efits of platooning with an integrated scenario covering
platooning on highways and in cities.
Fig. 1. The platooning process on a highway.
B. Our Approach for Platooning
State of the art platooning approaches focus on high-
ways. In our work, we integrate platooning on highways

with inner-city platooning. In this section, we describe
our scenario for a hybrid approach.
1) Highway: An Intelligent Transportation System
acts as PCS and enables the formation of platoons. It
receives information from drivers as their goal or route
characteristics (e.g., if they want to go the shortest route,
fastest route, or most efficient route) through the V2I
interface. This information is used for finding an existing
platoon or forming a new one. The PCS sends the
information to the involved vehicles. Platooning is con-
trolled by the vehicles via V2V communication. Vehicles
need to provide at least automatic longitudinal control.
Overtaking of platoons is coordinated with the PCS or
with other platoons. Platooning on a highway offers
various advantages. The use of slipstreams lowers fuel
consumption and environmental pollution. Furthermore,
the PCS can coordinate the platoons’ actions, such as
overtaking, and reducing the travel time further through
coordination. Safety is increased as the likelihood of
accidents due to human error is reduced. Streets are
used more efficiently, as the security distance can be
reduced resulting in an optimized utilization. Studies
elaborated that platooning could increase the capacity
of highway lanes by 100-200% in comparison to in-
dividually driven cars [4]. Further, spontaneous traffic
jams can be avoided with platooning. Inhomogeneous
traffic flows arise if high differences in vehicles’ velocity
exist. If this happens in combination with exceeding a
threshold in highway density (vehicles/km) braking can
lead to spontaneous traffic jams. Platooning eliminates
these jams as it homogenizes and optimizes the traffic
flow.
2) City: Whereas most platooning approaches focus
on platooning on highways, our approach combines
platooning on highways with an approach for inner-
city traffic management. Within a city, we focus on
vehicles that come from the highway and aim for
a certain location in the city (e.g., a specific event
location) or want to drive through the city (e.g., as
result of a redirection due to a closed highway). As it
is more complicated to establish a V2I communication
infrastructure in cities, our inner-city approach does not
aim at the formation of closely-driving platoons (which
require strong communication) as on highways. Hence,
we support formation of convoys through prioritiza-
tion of specific vehicles (prioritization can be based
on inner-city vs. going-through traffic or the number
of people in a vehicle) by dynamic routing and A-
TLC [6]. The A-TLC enables Decentralized Progressive
Signal Systems (DPSS, also called ”green waves”) for
convoys of going-through traffic with a common goal
outside of the inner city, as these vehicles should pass
the inner, crowded part of the city faster for lowering
the traffic load there. The use of bus lanes (only for
controllable units, such as autonomous vehicles) can be
allowed dynamically if there is no distraction of public
transportation. The A-TLC is complemented by dynamic
navigation information sent to navigation systems (of
non-autonomous vehicles), to VMSs, or to the control
unit of an autonomous vehicle via V2I. This information
can also be used for specific routing, e.g., for distributing
trucks on different routes in the city, or for separating
vehicles that want to drive to downtown and going-
through traffic. Moreover, the combination and splitting
of platoons through routing and A-TLC is an important
aspect here. Whereas platooning in cities does not gen-
erate slipstream effects, it still optimizes the utilization
of the streets and drivers’ routes.
III. RESEARCH QUESTIONS
The main objective of our work is to enable the
efficient use of autonomous driving through platooning.
This objective can be split to two main research ques-
tions: (i) How to manage platoons efficiently? and (ii)
How efficient is our approach for platooning? Whereas
the first question covers the efficiency of platoon coordi-
nation activities as forming of platoons, the second ques-
tion compares the efficiency of platooning compared to
situations without platoons. Further, differences in pla-
tooning on highways and cities are in our research scope.
In this section, we highlight our research questions.
A. How to Manage Platoons Efficiently?
Different factors influence the efficiency of coordi-
nating such platoons. As vehicles in platoons are highly
fluctuating and their characteristics can change, merging
of platoons, rerouting of platoons, vehicles changing
platoons, as well as overtaking of platoons are relevant
issues. Different quality of service metrics can be used
for optimizing the platooning process, e.g., put the
vehicles that have to leave next at the back for avoiding
acceleration to close a gap. For joining platoons, we
survey factors as whether it is more beneficial to speed
up for joining a platoon versus slow down to wait
for platoons. For platooning on highways, we have
some specific research issues. We assume, that various
platoons with different objectives use the highway at the
same time. Therefore, overtaking processes of platoons
are important here. One important issue is the necessary
degree of interaction between infrastructure and platoons
(as well communication within a platoon). Therefore, we
will compare the amount of messages between a platoon
and the infrastructure against intra-platoon (between
the vehicles of a platoon) and inter-platoon (between
vehicles of different platoons) messages. Based on this,
we provide efficient algorithms for platoon coordination.
B. How does Platooning on Highways Differ from Pla-
tooning within Cities?
In urban environments, prioritization of platoons can
increase the traffic conditions in situations where many
drivers have to leave the city very soon, e.g., after a
sports event. In our work, we model such situations and
derive algorithms and rules for efficient coordination

of platoons and traffic including prioritizing platoons.
Within cities, we have another approach for platooning.
In cities, platooning is achieved by a coordinated A-
TLC, use of bus lanes, and adaptation of traffic signs
through VMSs. As other vehicles should not be affected
negatively, we need efficient forecast algorithms for the
prioritization, e.g., dynamic use of bus lanes for pla-
tooning should not affect adversely the bus traffic. Our
algorithms have to take into account the administrative
effort compared against the usefulness of platoons in
cities considering the length of possible platooning in
cities (which is usually shorter as on highways). We
will pay special attention to the handover of platoons
from highway to urban environments. Vehicle control
will change between different systems. Delays cannot be
tolerated. Efficient algorithms for the handover process
are required.
C. How Efficient is Platooning?
The question regarding the efficiency of platooning
is two-folded. On the one hand, there is the question
about how efficient platooning itself is. On the other
hand, there is the question how efficiently platooning
uses the infrastructure.
1) Which factors enable efficient platooning? Possible
factors for measuring the efficiency of platooning such
as the travel time, the achieved throughput, energy
consumption, or utilization of the street influence the
characteristics of platoons, e.g., platoon size, speed, or
vehicle type. Further, studies can reveal factors that
influence user acceptance of platooning and when it
is beneficial to use platooning. Additionally, efficient
forecast techniques are relevant for efficient platooning
based on dynamic and robust techniques that can cope
with uncertainty, e.g., arising from non-controllable ve-
hicles. Forecasts are relevant for forming platoons and
determining which vehicles should join which platoon.
Further, the quality of forecast techniques determines
the efficiency of prioritization of platoons in urban
environments. Scalability of the platooning approach
is another factor that influences the efficiency of pla-
tooning. We assume that decentralized approaches will
perform better than centralized ones as they offer higher
responsiveness, faster adaptation, and can easier cope
with a high amount of data. However, as we want to
avoid platooning with leading vehicles as coordinator,
the infrastructure has central points that enable the coor-
dination of platoons. Another issue concerns the amount
of autonomous vehicles vs. normal vehicles: Where is
the threshold for the share of non-automatic vehicles
such that platooning works efficiently? As platooning
is dedicated to vehicles that can be controlled by the
control system, we assume that the higher the amount of
such vehicles is, the better the performance of platooning
will be. The ratio of autonomous (controllable) vehicles
versus vehicles, that are not controllable by our system,
influences the efficiency of platooning. This includes
the comparison of different ratios and definition of
thresholds.
2) How to use the existing infrastructure more effi-
ciently with platooning? Further, another objective is to
enable a more efficient use of the existing infrastructure.
Our approach for platooning avoids the use of special
leading vehicles. Instead, the infrastructure enables pla-
tooning. However, our approach should be non-intrusive
to the existing infrastructure. The degree of reuse of the
infrastructure is an interesting research topic within our
studies. Further, the type of communication is within
our research scope. Can we use V2I communication
infrastructure in cities and on highways? Do the re-
quirements for V2I communication differ between these
scenarios? The communication analysis results are the
foundation for the definition of efficient communication
protocols (defining how to exchange which data) and
communication mechanisms. We assume that infrastruc-
tural providers such as telecommunication companies
could play a leading role in offering a platooning in-
frastructure. Another important infrastructural resource
is the street itself: How can platooning use streets
efficiently? This question is obviously related to the
design of efficient platooning algorithms. We assume
that platooning improves the utilization of streets by
increasing the throughput while minimizing traffic con-
gestion. We specially focus on the efficiency of inner-
city through A-TLC.
IV. REL ATE D WORK
In the following, due to space limitations, we shortly
summarize major platooning projects. Our technical
report from [7] presents a comprehensive overview on
platooning. The PATH program [4] is among the first
projects investigating the potential of platooning. There,
Platoons drive on dedicated lanes. Longitudinal control
is achieved by magnetic nail following. All vehicles are
fully automatically driven. Similar to PATH, all vehicles
have equal roles in our approach. However, we aim at
minimal intrusion. A dedicated lane for platooning is not
in our scope. A focus of the SARTRE project [1] was
to enable platooning on existing public roads without
changing the roadside infrastructure. A truck or a bus as
leading vehicle, driven by a trained driver is followed by
autonomous driving buses, trucks, or cars. An additional
remote system guides drivers which are not already
part of a platoon to the nearest convoy with a suitable
destination. Our approach has the same objective than
SARTRE in using the existing infrastructure. However,
we want to avoid the need of specific leading vehicles
as in our opinion this decreases scalability, availability
of platooning, as well as safety. Energy ITS started in
2008 [8]. As Energy ITS platoons only require on-board
equipment, Energy ITS does not need infrastructural
changes. Vehicles have radar and a 2-dimensional lidar
for obstacle detection and gap measurement as well as
DSRC for V2V communication. However, Energy ITS

covers platooning with trucks only. Contrary to Energy
ITS, our approach includes all types of vehicles. The EU
project COMPANION [9] aims at dynamic forming of
platoons and is supported by Volkswagen and Scania.
However, the scope is limited to trucks as in Energy
ITS. All of these approaches are limited to platooning
on highways. None of them offers an approach for cities.
Gershenson [6] studied self-organizing traffic lights in a
multi-agent simulation based on a toroidal traffic grid.
He showed that with simple rules and without direct
communication, he could reduce the average waiting
times at red lights and the number of stopped cars.
The Sotl-request control holds a counter for the number
of waiting cars. With a sufficient number of cars, the
red lights will turn green, creating platoons of cars.
However, this solution is limited to toroidal traffic grids
and limited to inner-city traffic.
In contrast to fixed-time controlled traffic lights,
traffic-actuated controls adapt their signalization to the
monitored traffic situation. This can result in a green
time adaptation, switching to on-demand phases, a phase
sequence adaptation, or a cycle time adaptation. Claes et
al. [10] demonstrated in a real-world traffic scenario how
their delegate multi-agent system reduces congestion
and the average travel times for traffic participants. Their
decentralized approach mimics ant behavior, using real
time traffic information from probe vehicles. Research
by Pan et al. [11] showed that re-routing strategies
based on real-time traffic data from vehicles lowers the
average travel time with less re-routing frequency. They
present three novel algorithms and evaluate them in a
simulation-based study.
V. PRO JEC T STATUS
So far, this paper presented the vision of an integrated
approach to coordination of platooning. The project
contains two elements: (i) the PCS for coordination of
platoons on highways [12] and (ii) the Organic Traffic
Control (OTC) system [13], [14]. For both elements,
prototype systems exist. Currently, we are working on
the integration of these system. This section presents
these systems as well as the challenges for the connec-
tion of both systems into an integrated testbed.
A. Adaptive Traffic Control
Earlier work applied the Observer/Controller archi-
tecture known from Organic Computing to traffic signal
control resulting in the Organic Traffic Control (OTC)
system (see Figure 2). The basic OTC system is respon-
sible for the adaptation of the green times at traffic lights
at intersections according to the present traffic condi-
tions. The self-learning, self-optimizing system follows a
safety-oriented concept that allows OTC to adapt within
defined boundaries. Each individual instance of OTC is
fully decentralized and controls one intersection only.
Traffic light controllers (TLCs) located on nearby inter-
sections which are directly connected via streets may
communicate with each other to establish a distributed
coordination of TLCs. The self-organized route guidance
mechanism is able to calculate the fastest routes through
the network to prominent places based on current traffic
flows. Techniques from the Internet domain, such as the
Distance Vector Routing (DVR) and the Link State Rout-
ing (LSR) protocols, are applied to road traffic guidance.
In every time step, the monitoring component receives
the current raw traffic data from sensors located at Layer
0 and processes them before the data is forwarded to
other modules such as the prediction component or
the situation analyzer. The latter generates a situation
description of the current traffic flow for an intersection
(vehicles/hour for every intersection), whereas the pre-
diction component forecasts traffic flows for upcoming
time steps. The prediction component itself consists of
several prediction methods that each make forecasts.
These methods range from simple smoothing algorithms
that calculate a forecast based on the recent data to more
sophisticated algorithms such as Kalman Filter, Artificial
Neural Networks, or ARIMA [13].
Fig. 2. Architecture of the Organic Traffic Control system.
B. Platooning Coordination System
In the iCOD project2, we implement a system for
self-organized infrastructure-aided cooperative driving
through coordination of platoons on highways. In [12],
we describe the implementation of a demonstrator3of
the PCS with LEGO Mindstorms robots. LEGO robots
drive autonomously (following a colored lane) and drive
in platoons (using the distance sensor). Forming a pla-
toon is supported by the PCS. The PCS is designed
as a self-adaptive system and implemented using the
FESAS framework [15]. It assigns vehicles to platoons
based on individual preferences of the driver, vehicle
factors, context factors as well as the integration of
a compensation system. V2I communication is WiFi-
based. Based on the demonstrator, we recently built a
testbed which integrates the PCS with the platooning
2iCOD project website: http://icod.bwl.uni-mannheim.de/
3A video showing the platooning demonstrator can be found at:
https://www.youtube.com/watch?v=Nnrbq-4Dn24

simulator PLEXE [16] for evaluating various coordi-
nation strategies. The usage of a large-scale simulator
allows a detailed quantitative analysis of the approach.
Additionally, it is possible to compare different coordi-
nation algorithms and to evaluate their performance with
realistically simulated vehicles. Therefore, algorithms
for coordination can easily be plugged into the system
and the system chooses at runtime the algorithm that
corresponds to the driver’s preferences. Figure 3 depicts
the architecture of the simulation testbed. The left hand
side shows the real world system; right hand side shows
the testbed with PLEXE that integrates SUMO for
traffic simulation and the OMNeT++ and VEINS for V2I
simulation.
PCS /
FESAS
VEINS /
OMNet
SUMO
PLEXE
Fig. 3. The simulation testbed for the Platooning Coordination
System.
C. Open Challenges
For future work, we have to link the OTC with the
PCS for designing an integrated test bed. This raises
several challenges, which can be categorized into (i) im-
plementation challenges and (ii) research challenges. In
the following, we present these challenges.
The existing OTC approach supports dynamic routing
through A-TLC. Currently, we extend the system by
an approach for dynamic, temporary assignment of bus
lanes to normal vehicles. Further, we survey methods
how to establish inner-city platooning through traffic
light control for DPSS creation and intelligent naviga-
tion of traffic. Additionally, we increase the usability of
the existing PCS testbed for avoiding the need to handle
all the necessary tools and to offer an out-of-the-box
approach for testing coordination algorithms. Last, the
integration of both testbeds might raise implementation
challenges, e.g., the integration of the AIMSUN simula-
tion used in the OTC system with the SUMO simulation
from the PCS testbed.
For profound analysis of the coordination perfor-
mance, we need to define suitable metrics and objectives.
However, as platooning is a multi-dimensional optimiza-
tion problem, this is challenging. Possible objectives,
that might conflict each other, are the optimization of
the traveling time, fuel consumption / environmental
pollution, or traffic throughput. Objectives of individual
drivers might conflict with global optimal objectives,
e.g., travel as fast as possible vs. reducing environ-
mental pollution. In contrast to existing solutions, a
platoon might be composed integrating different driver
preferences. Hence, objectives need to be balanced and
suitable metrics to measure them needs to be defined.
Especially a reward / compensation system must be
defined for two reasons. First, the position in the platoon
influences the slipstream effects and the benefits in terms
of fuel saving. Second, for inner-city platooning, the
system might redirect individuals to longer routes for
the sake of global optimization.
Our approach also takes inter-platoon interactions into
account. This is not integrated in current platooning
approaches so far. Additionally, we assume that not all
vehicles are able to platoon, hence, interactions with
normal traffic are necessary. However, this makes the
solution ready for use when the very first platooning
vehicles are on the streets.
REFERENCES
[1] C. Bergenhem, H. Petterson, E. Coelingh, C. Englund,
S. Shladover, and S. Tsugawa, “Overview of Platooning Sys-
tems,” in Proc. ITSWC, 2012, pp. 1393–1407.
[2] S. El-Tawab and S. Olariu, “Communication protocols in
FRIEND: A cyber-physical system for traffic Flow Related In-
formation Aggregation and Dissemination,” in Proc. PerComW,
2013, pp. 447–452.
[3] P. Ghazizadeh, S. Olariu, A. G. Zadeh, and S. El-Tawab, “To-
wards fault-tolerant job assignment in vehicular cloud,” in 2015
IEEE International Conference on Services Computing, June
2015, pp. 17–24.
[4] S. Shladover, “PATH at 20–History and Major Milestones,” IEEE
Trans. ITS, vol. 8, no. 4, pp. 584–592, Dec. 2007.
[5] J. O. Kephart and D. M. Chess, “The Vision of Autonomic
Computing,” IEEE Computer, vol. 36, no. 1, pp. 41–50, 2003.
[6] C. Gershenson, “Self-Organizing Traffic Lights,” Complex Sys-
tems, vol. 16, no. 1, pp. 29–53, 2005.
[7] T. Kalbitz, “A comparison of approaches for platooning man-
agement,” Universit¨
at Mannheim, Tech. Rep., 2017. [Online].
Available: http://ub-madoc.bib.uni-mannheim.de/42305/
[8] S. Tsugawa, S. Kato, and K. Aoki, “An Automated Truck Platoon
for Energy Saving,” in Proc. IROS, 2011, pp. 4109–4114.
[9] M. Pillado, D. Gallegos, M. Tobar, K. H. Johannson, J. Martens-
son, X. Ma, S. Eilers, T. Friedrichs, S. S. Borojeni, M. Adolfson,
and H. Pettersson, “COMPANION - Towards Co-Operative
Platoon Management of Heavy-Duty Vehicles,” in Proc. ITS,
2015, pp. 1267–1273.
[10] R. Claes, T. Holvoet, and D. Weyns, “A Decentralized Approach
for Anticipatory Vehicle Routing Using Delegate Multiagent
Systems,” IEEE Trans. ITS, vol. 12, pp. 364–373, 2011.
[11] S. J. Pan, I. S. Popa, K. Zeitouni, and C. Borcea, “Proactive
vehicular traffic rerouting for lower travel time.” IEEE Trans.
Vehicular Technology, vol. 62, no. 8, pp. 3551–3568, 2013.
[12] C. Krupitzer, M. Breitbach, J. Saal, C. Becker, M. Segata,
and R. Lo Cigno, “RoCoSys: A framework for coordination of
mobile IoT devices,” in Proc. PerComW, 2017, pp. 485–490.
[13] M. Sommer, S. Tomforde, and J. H¨
ahner, “An organic computing
approach to resilient traffic management,” in Autonomic Road
Transport Support Systems, 2016, pp. 113–130.
[14] S. Tomforde, H. Prothmann, F. Rochner, J. Branke, J. H¨
ahner,
C. M¨
uller-Schloer, and H. Schmeck, “Decentralised Progressive
Signal Systems for Organic Traffic Control,” in Proc. SASO,
2008, pp. 413–422.
[15] C. Krupitzer, F. M. Roth, C. Becker, M. Weckesser, M. Lochau,
and A. Sch¨
urr, “FESAS IDE: An Integrated Development Envi-
ronment for Autonomic Computing,” in Proc. ICAC, 2016, pp.
15–24.
[16] M. Segata, S. Joerer, B. Bloessl, C. Sommer, F. Dressler, and
R. Lo Cigno, “PLEXE: A Platooning Extension for Veins,” in
Proc. VNC, 2014, pp. 53–60.