Conference PaperPDF Available

V2XSim: A V2X Simulator for Connected and Automated Vehicle Environment Simulation


Abstract and Figures

In this work we propose a vehicle-to-everything (V2X) simulator, called V2XSim, for connected vehicle (CV) environment simulation. As this field is rather new, many researchers focusing on CV-based research do not have access to real-world test-beds to validate their methodologies. As such, there is a timely need for simulation platforms that can incorporate precise vehicle mechanics, accurate physics of vehicle movement, and communications in the transportation network. Accordingly, this paper provides an integrated V2X simulation platform, built with the Gazebo robot simulation engine. The simulated world is constructed with two types of models, namely dynamic and static models. we provide a detailed communication architecture to simulate the vehicle communication process. In this architecture, a closed-loop control module is built to pass control commands and control vehicle models without human intervention. We also provide multiple APIs for users to simplify the vehicle control process and make the simulator easy to use.
Content may be subject to copyright.
V2XSim: A V2X Simulator for Connected and Automated Vehicle
Environment Simulation
Ethan Zhang1and Neda Masoud1
Abstract In this work we propose a vehicle-to-everything
(V2X) simulator, called V2XSim, for connected vehicle (CV)
environment simulation. As this field is rather new, many
researchers focusing on CV-based research do not have access
to real-world test-beds to validate their methodologies. As
such, there is a timely need for simulation platforms that
can incorporate precise vehicle mechanics, accurate physics of
vehicle movement, and communications in the transportation
network. Accordingly, this paper provides an integrated V2X
simulation platform, built with the Gazebo robot simulation
engine. The simulated world is constructed with two types
of models, namely dynamic and static models. we provide a
detailed communication architecture to simulate the vehicle
communication process. In this architecture, a closed-loop
control module is built to pass control commands and control
vehicle models without human intervention. We also provide
multiple APIs for users to simplify the vehicle control process
and make the simulator easy to use.
Keywords: Connected vehicles, V2X, Simulation, Gazebo
The development of the connected and automated vehicle
(CAV) technology is envisioned to bring a plethora of
opportunities, including higher sustainability, mobility, and
safety levels in transportation systems [1]–[4]. However,
extensive research is needed to quantify such benefits before
widespread deployment of CAVs can be justified. CAV
research is still in its early stage, partially due to the scarcity
of controlled testing facilities. The gap between the need for
research and the scarcity of resources motives the simulation
platform presented in this paper.
In the past few years a number of real-world testbeds
have been built for the purpose of collecting CAV data
and testing CAV-related algorithms [5], [6]. Although real-
world testbeds are useful, and even a necessity for certain
applications, their initial deployment as well as their re-
configuration can be costly and time-consuming. Addition-
ally, collecting data from real-world testbeds can create ad-
ditional social concerns related to cyber security and privacy
[7]–[9]. Consequently, at the early stages of a project, it
might be more practical and economical to adopt a simulation
platform. A simulation environment not only can largely
reduce experimental costs, but also can allow for reducing
the set of configurations under which an algorithm needs to
be implemented and tested in a real-world testbed. Safety
is another main concern when it comes to testing new
technology–although real-world large-scale experiments are
currently in place to test mobility benefits of the CAV
1E. Zhang and N. Masoud are with Department of Civil and Environ-
mental Engineering, University of Michigan, Ann Arbor, MI, USA, 48105
technology [5], safety-critical applications of this technology
are harder to assess in real-world settings, especially when
it comes to new applications at the proof-of-concept stage.
CAV simulation environments can be used to replace real-
world facilities. However, their level of effectiveness depends
on the degree to which they are capable of capturing the
elements that are critical in the application at hand. In
this work, we provide an agent-level vehicle-to-everything
(V2X) simulator, with the goal of enabling researchers to
design and simulate customized CAV environments for a
wide range of applications. Unlike most existing simulators
that use game engines or traffic-focused simulators [10],
[11], the proposed simulator is built on top of a robotics
engine, Gazebo, enabling it to simulate CAVs at an agent
level with great physics properties. The simulation envi-
ronment provides an integrated platform that enables infor-
mation sharing, allowing for testing of safety-critical CAV
applications, e.g., vehicle control algorithms in a connected
vehicle environment. One of the critical aspects of a CV
environment is information sharing. In transportation field,
many researchers are interested in investigating the behavior
of vehicles in connected environment [12], [13]. In that
case, the information sharing between vehicle plays the most
important role. In the proposed simulator, each vehicle is
regarded as an individual mobile robot formed by multiple
joints and sensors. The main concentrations of the simulator
is to help researchers achieving information sharing between
vehicles, and to provided them a integrated platform by
which users can easily test their vehicle behavior control
algorithms in connected vehicle environment.
The simulator uses a decentralized simulation architecture:
Each vehicle’s motion is controlled by its own control logic,
which can be defined by the user. The state of all vehicles is
collected by their own state listeners. A virtual RSU-based
communication architecture is designed to collect messages
from all vehicles, and transmit them to other vehicles or other
desired destinations. With these characteristics, the simula-
tion environment provides an integrated platform for users to
simulate physics-based 3-D connected vehicle environments,
and test detailed vehicle behavior.
V2X is becoming an emerging technology in modern
transportation systems. Unlike traditional vehicles, through
the V2X technology connected vehicles share information
with other road users as well as various infrastructure. The
information sharing between vehicles and their surrounding
objects is the most essential part of V2X. Researchers think
that by utilizing connected vehicles and V2X technology,
the society can come up with some better way to improve
transportation systems’ efficiency. For example, reducing
traffic congestion and increasing driving safety [1]. In the
rest of this section, we discuss existing simulators including
their capabilities in modeling V2X communications, as well
as different components that build the proposed simulator.
A. Existing simulators
Existing intelligent vehicle simulation platforms can be
categorized into two types, namely, hardware-in-the-loop
(HIL) simulation and full simulation. HIL simulation con-
tains real world hardware [14], which are typically real
vehicles in a connected environment. It captures real-world
experimental vehicle data and integrates them with data from
the simulated environment. A full simulation simulates all
necessary components and functions through a simulation
1) Hardware-in-the-loop simulation: M. R. Zofka et al.
propose a simulation framework to evaluate and validate
automated driving. Their simulation platform provides the
capability to combine real-world autonomous vehicles with
a simulated traffic environment [15]. Its simulation frame-
work focuses on a single vehicle’s functions (the target
vehicle) and other surrounding vehicles are simulated with
the traffic simulator SUMO, whose connectivity functions,
like communications between vehicles are not considered.
W.M. Griggs et al. propose a SUMO-based HIL emulation
platform to provide human drivers with the experience
of driving in a connected vehicle environment, so as to
help researchers examine the change in drivers’ behavior
[16]. Kevin S. Swanson et al. create an HIL simulator
to combine a simulated world environment with hardware
through robot operating system (ROS) and Gazebo. In the
simulator, Gazebo is used to create a virtual environment
[17]. Similar to Zofka’s work, their simulation platform
focuses on a single vehicle, which indicates that vehicle
connectivity is not implemented. M. Tideman et al. propose
an HIL simulator tool suited to provide simulation support
for Advanced Driver Assistance Systems (ADAS) [18]. It
focuses on a single vehicle’s detailed ADAS performance in a
simulated connected vehicle environment, with the dynamics
of surrounding simulated vehicles pre-defined through its ITS
modeller, i.e., the interactions between the target vehicle and
other vehicles are not dynamic. Among HIL simulators, the
existing work either only considers a single target vehicle,
or otherwise does not consider the connectivity between
2) Full simulation: Jos ´
e L. F. Pereira et al. propose an
integrated architecture for autonomous vehicle simulation
[19]. The simulator combines a robotics game engine (US-
ARSim) and the traffic simulator SUMO. It simulates vehicle
agents in a traffic stream, with the purpose of controlling
autonomous vehicles. However, it can only apply the control
mechanism on a single vehicle. In addition, the simulator
uses a game engine to simulate vehicles–since the physics
of the vehicle engine is mainly designed for game usage,
there is no real-world control performance guarantee. A.
Hussein et al. provide a framework for intelligent vehicle
control and simulation. Their framework combines ROS and
the Unity engine to create a simulated world, focusing on
a single intelligent vehicle [10]. L. Zhao et al. provide
a simulation framework for platooning under a connected
vehicle environment through VISSIM [11]. This framework
is designed based on macroscopic traffic flow and facilitates
simulation of Cooperative Adaptive Cruise Control (CCAC).
The limitation of this simulator is that it only provides
a 2-dimensional simulation environment, where real-world
physics and the surrounding environment (buildings and etc.)
are also not considered. The VISSIM-based testbed adopted
by J. Lee et al. has similar properties and suffers a similar
set of shortcomings [13].
B. Gazebo engine
Gazebo is a simulation engine that provides high-precision
physics for robotics-related simulation [20]. Gazebo is an
open-source and programmable platform–it provides func-
tions to users to build and customize 3D simulated envi-
ronments. It is known in the robotics field by its capability
and accuracy on simulating real-world robot performance.
Gazebo decomposes agent-level control to component-level
control, allowing for controlling agents by controlling their
various components. To obtain agent-level control, users can
combine different components as a joint, and then control
a group of components with a single command through a
joint. Unlike macroscopic traffic simulation platforms that
treat vehicles as moving boxes, using its physics engine
Gazebo allows for capturing microscopic vehicle movement
that prove important when the intent is to capture detailed
vehicle motion in real-world scenarios.
We conclude the literature review section by noting that
connectivity is generally not an integrative component of
existing transportation-focused simulators, specially when
modeling CAVs with both connectivity and autonomy fea-
tures is desired. To facilitate the development of new appli-
cations in Intelligent Transportation Systems (ITS), in this
paper we present an open-source V2X simulator that treats
vehicles as mobile robots and emphasizes the connectivity
among them. The proposed simulator enables connectivity
among vehicles through generating and transmitting basic
safety messages (BSMs) [21], and allows for using ap-
plication programming interfaces (APIs) to simulate and
customize/control intelligent vehicles based on BSMs. A
BSM is a standard format/protocol that contains information
including velocity, position, acceleration, heading angle, etc.
Unlike previous simulators with centralized control, in the
proposed simulator each vehicle has its individual commu-
nication and control modules, i.e., vehicle agents can behave
in a distributed way, and the interactions between vehicles
are controlled based on vehicle-specific control algorithms
and rules. The distributed control design enables users to
test vehicle-specific control tasks.
Fig. 1: Overview of the simulator design
The architecture of the proposed simulation framework
consists of four main components, namely, the vehicle’s
surrounding world, communication, closed-loop control, and
APIs. Each of these components is described in detail.
A. The simulated world
The simulated world consists of multiple modules that
represent the real world environment. In this simulated envi-
ronment, vehicles operate by executing control commands,
interacting with each other and with the static objects, e.g.,
buildings and trees. To build a simulation environment in
Gazebo, we implement 3D models from the Gazebo library.
Two types of models are defined, namely, static and dy-
namic models. Dynamic models are vehicle models, and are
controlled by control modules, while the surrounding static
environment is built with static models.
Dynamic models in the simulator are defined as objects
whose states can change over time. In Gazebo, a vehicle can
be formed by combining many small models, e.g., wheels,
frame, etc., where each component has its own detailed
model configuration. Though control can be performed on
these sub-level models, in this work, we define a vehicle as
ajoint of components, including vehicle body, sensor, tires,
etc. The components of a joint provide a detailed view of
an individual vehicle, allowing the user to focus on different
vehicle components when desired.The workflow overview of
the simulator is shown in Fig. 1.
B. Communication
the communication process is simulated in order to dis-
tribute messages that contain vehicle motion and traffic state
information. In the communication component, we focus on
obtaining the message contents so as to enable users to ana-
lyze them and test connectivity-related algorithms based on
those messages. In each simulation update period, we record
the states of vehicles. Gazebo performs its simulation tasks
by rapidly executing related control modules and updating
models’ states.
To successfully complete the communication process, first,
each vehicle gathers and reports its own state in real-time.
To collect the relevant information, we build two extended
modules on top of each vehicle model, which are the
state listener and state publisher. These extended modules
allow for monitoring each vehicle’s model independently. A
state listener is a monitoring unit that collects the vehicle’s
motion information at the backend when the simulation is
running, while a state publisher is a module that transmits
this information to designated destinations. Unpacking the
information carried by state publishers provides real-time
model states at every simulation update. The advantage of
including state listener and state publisher is providing a
standard format, or protocol, similar to a BSM, for the
transmitted information. Having vehicles autonomously and
indiscriminately transmit their motion information results in
a large number of messages when a simulation runs. It is
likely to congest the communication tunnel in congested
traffic states, causing communication delay and slow down
the simulation process. To avoid this issue, the State listener
module on each vehicle serves as a collector that gathers
information from different sensors of interest during each
simulation update and integrates them in a standard for-
mat/protocol. It also keeps messages organized for future
analysis. By customizing the format of state listener, it is
possible to collect different information of interest. Each
state listener is accompanied with a state publisher. At
each time epoch when a simulation update ends, the state
publisher sends out the integrated information package to its
target destinations.
To facilitate communication in a connected environment
in an effective manner, we create a module to monitor
the communication in the network. After each vehicle ob-
tains the relevant information using its state listener, its
state publisher does not directly transit the BSMs to other
vehicles; rather, all BSMs are sent to a virtual RSU–an
intermediate module layer that serves as a platform to “filter”
messages before forwarding them to suitable destinations.
This communication architecture design is inspired by the
Ann Arbor Connected Vehicle Test Environment (AACVTE),
which is the largest real-world connected vehicles experiment
environment in the world [5]. In AACVTE, when available,
vehicle messages will be collected and stored in centralized
servers through RSUs.
Using a virtual RSU reduces the likelihood of flooding the
communication channel. In a simulated environment with
Nconnected vehicles, without using an RSU to control
the communications, a total of N×N1messages need
to be transmitted at high frequency. Replacing the vehicle-
to-vehicle messages by vehicle-to-RSU and RSU-to-vehicle
messages will reduce the number of communications to
2N.Virtual RSU also provides a platform for users to
monitor the entire system. A receiver and transmitter pair
is built in the virtual RSU, which are designed to receive
messages from vehicles and send out messages to target
destinations. The existence of a Virtual RSU is necessary–
it allows the simulator to provide users with an overview
of all connected devices, informing network-level operations
and control. Instead of fetching these messages one by one,
the virtual RSU provides centralized storage, which is more
suitable for fetching messages. In addition, this architecture
allows the virtual RSU to be built at a different location
than the simulator host, for example, a remote server, which
makes the choice of message storage location flexible. The
virtual RSU can then send out collected information to any
desired addresses. This allows for the V2X communica-
tion process to be completed using a single protocol, i.e.,
we do not need to specify different communication rules
between different objects/models, since all communications
follow the universal virtual RSU architecture. To finish the
communication loop, a receiver module is built on each
destination model, e.g., vehicle model, to receive messages
from the virtual RSU. The receiver is continuously listening
to the messages from the virtual RSU’s transmitter unit. In
each simulation update, the transmitter sends a customized
information package, which can contain other vehicles’ infor-
mation, to each desired destination. All vehicles can obtain
messages from other vehicles using their receiver modules
through the proposed communication loop. Following this
setting, a vehicle in the simulation environment can send
and receive messages from other road users.
C. Closed-loop control
Control is an essential part of the simulation platform.
It provides commands that instruct vehicles on how to
operate. In the architecture, we design a closed-loop control
procedure to help users investigate the impact of vehicle
control algorithms on vehicle motion. We assume the control
of a vehicle depends on other vehicles’ information. When
a vehicle is given a control command, it executes it, which
leads to a state change. With the communication module in
place, the updated state information is sent to the virtual
RSU, and then through it to other vehicles. The control
module of a vehicle obtains messages containing information
from other vehicles from the virtual RSU, decodes these
messages to obtain other vehicles’ state information, and uses
the updated states to inform its control algorithm. Through
this process, closed-loop control can be achieved for each
simulation update.
The control module is built to directly control vehicles’
detailed behavior at the lower-level. Developing control
scripts to control vehicle models requires comprehensive
knowledge of the system itself as well as Gazebo, which
could make the process complex. To simplify the control
process and allow users to develop customized control rules,
we provide a number of upper-level control APIs. Through
these APIs, even without full knowledge of Gazebo , users
can still customize their control commands using a limited
number of pre-defined and common variables, which include
velocity, acceleration and six-degree-of-freedom vehicle pose
(6DoF, which are surge, sway, heave, roll, pitch and yaw),
in a simple manner. The Gazebo plugin system allows
in-depth customization on Gazebo elements. The Gazebo
engine provides different protocols to developers On different
module levels (e.g., sensor, model, world). Based on these
protocol rules, to interact with lower-level settings more
conveniently, in the proposed simulator we develop model
plugins on top of each vehicle model, so as to help users
State listener State Publisher
Control command
Update rule
Control algorithm
Message subscriber
Receiver Messages from
other vehicles
Message analysis,
vehicle monitoring,
Gazebo Plugin
A vehicle model in the simulated world
Control plugin
Control API
Virtual RSU
centralized message
collector & storage
(for simulation
with real world
hardware in the
ROS enabled
(e.g., real world
vehicles, robots,
or sensors)
Fig. 2: Architecture for each simulated vehicle model
pass commands to the vehicle model in a easy manner.
In each plugin, we pre-define lower-level configurations to
interact with the Gazebo engine. Through model plugins,
detailed vehicle model information can be visited. Control
APIs communicate with plugin through the Gazebo internal
messaging protocol, gzmessage. Through the control API,
users/algorithms can directly set their desired values on
vehicle state parameters.
The overall architecture of the proposed simulator is
illustrated in Fig. 2. In addition to the previously introduced
components, this architecture also demonstrates the robot op-
erating system (ROS) extension. The Gazebo engine allows
connections to real world hardware through gazebo ros pkgs,
although the current software version does not implemented
this extention. If users need to combine real world vehi-
cles, robots, or sensors with the simulator, the framework
also allows further implementation of hardware-in-the-loop
simulation with ROS, through the gazebo ros pkgs package.
The package can establish a messaging tunnel between the
Gazebo simulator and ROS-enabled hardware. All simulated
vehicle models use the architecture demonstrated in Fig.
2. It enables each vehicle to to utilize its own customized
control algorithm, enabling users to study different aspects
of heterogeneous traffic streams. Another benefit of this
architecture is that, owing to Gazebo providing sensor-level
simulation, it enables users to investigate performance of
any combination of sensors by simply adjusting the protocol
(message format) of the state listener. Finally, the archi-
tecture provides a convenient platform for sharing different
types of information, which reduces the difficulty of cross-
model information sharing, i.e., connectivity.
The proposed architecture is realized with the Gazebo en-
gine. Backend implementation of the simulation framework
is described in this section.
A. Models and the simulated world
The entire simulation world is constructed with 3D models
provided by the Gazebo model libraries. In the world,
we simulate the vehicle surrounding world with trees, gas
stations, apartments, schools, office building and etc. Several
routes are formed between building blocks. The vehicle
(a) Simulated world view 1 (b) Simulated world view 2
(c) Simulated world view 3 (d) Simulated world view 4
Fig. 3: Different views of an example simulated world
model type used in the simulator is Puris 3D model with
sensors [22], which is an enhanced Puris 3D model equipped
with sensors such as cameras, LIDAR, etc. The simulated
world is shown in Fig. 3, in which the blue lines represent the
LIDAR beams generated by the vehicle’s on-board LIDAR.
To operate the vehicle models, low-level control commands
are provided, which are described in the following section.
B. lower-level modules
We use the Gazebo plugin to implement closed-loop
control. Control APIs are built on top of plugins to send
information to plugins. In each plugin, we set a gzmessage
subscriber object. Correspondingly, a gzmessage publisher
object is built to publish each customized information. A
simple demonstration case is that, when the control unit/user
wants to change velocity (whose value is either generated
by closed-loop control algorithms or defined by the user), it
can simply update the API and the API will automatically
update the Gazebo plugin message through the gzmessage
subscriber-publisher pair. The Gazebo simulator executes the
updated velocity value through the Gazebo plugin for the
corresponding vehicle, and then visualize it in the main
simulation console. Since all messages from vehicles are
collected by the virtual RSU, the system can be monitored
by performing analysis on top of virtual RSU.
C. Simulation example
In this section we demonstrate how the simulator can be
used to simulate scenarios in a connected vehicle environ-
ment. We implement the intelligent driver model (IDM) for
vehicle control in a V2X environment. Each vehicle’s IDM
model is informed by the information collected through the
framework. As there are many versions of IDM, we adopt
the version presented in W. Sultani’s work [23], in which:
dt =α1(vn
where αis a predefined factor and δis a parameter that
governs the changes in acceleration, and vnrefers to the
speed of the nth following vehicle. We set the minimum
distance between vehicles as Smin = 5m. The desired speed
is set as 12 m/s, which is equivalent to 27 mph. With 4
simulated vehicles, under the environment depicted above,
the simulation time of vehicles driving through a 30-meter
long street is about 30 seconds. Define the distance between
two adjacent vehicles as S. In this work we adopt δ= 1,
which indicates a linear decrease in acceleration [24], and
set α=S
. The deceleration mechanism we adopt for the
IDM is based on [24]:
dt =(Smin
We assume the desired speed for a following vehicle
is the same as its leading vehicle’s speed. As indicated
in the mechanism above, a vehicle needs to obtain geo-
locations of other vehicles in order to compute its desired
acceleration. Therefore, the connectivity is required, and is
achieved through the proposed framework. The results are
shown in Fig. 4.
(a) Car following simulation speed illustration
(b) Car following simulation acceleration illustration
Fig. 4: Car following simulation results using the proposed
simulation framework
Collected messages by the virtual RSU are shown in
Figure 5, in which vehicle pose (6-DoF) along with basic
vehicle motion status (i.e., velocity and acceleration) are
collected and printed.
Without any human intervention, the IDM algorithm is
successfully implemented and tested through the connectivity
provided by the proposed simulator. The above example
is just a simple demonstration of a possible scenario that
Fig. 5: Messages collected by the virtual RSU
can be implemented with the proposed simulator. There are
certainly many other research areas can be investigated using
the proposed simulator, including trajectory planning [25],
[26], localization [27], and CAV cybersecurity [7]–[9].
This paper proposes a V2X platform for simulation of
CAV environments. The simulator is developed on top of the
Gazebo engine, and supports connection to ROS for exten-
sion to hardware-in-the-loop simulation. With the proposed
platform, users can carry out detailed simulations of CAVs,
including testing connected vehicle control algorithms. The
proposed simulator uses a virtual RSU to establish communi-
cation between vehicles, which enables users to monitor the
system status and enrich possible simulation scenarios. In the
proposed simulation framework, every vehicle is constructed
as an independent robot model with multiple parts. This
enables modeling traffic streams that are heterogeneous in
terms of the penetration rate of CAVs, connectivity tech-
nologies, and control algorithms. The simulator provides a
platform to investigate CAV environments in a more cost-
effective way compared to real world test-beds.
The video demonstration can be found at: The
open-source source code and the instruction manual of
the simulator can be found at:
[1] N. Lu, N. Cheng, N. Zhang, X. Shen, and J. W. Mark, “Connected
vehicles: Solutions and challenges,” IEEE internet of things journal,
vol. 1, no. 4, pp. 289–299, 2014.
[2] N. Masoud and R. Jayakrishnan, “Autonomous or driver-less vehicles:
Implementation strategies and operational concerns,” Transportation
research part E: logistics and transportation review, vol. 108, pp.
179–194, 2017.
[3] M. Abdolmaleki, N. Masoud, and Y. Yin, “Vehicle-to-vehicle wireless
power transfer: Paving the way toward an electrified transportation
system,” Transportation Research Part C: Emerging Technologies, vol.
103, pp. 261–280, 2019.
[4] M. Abdolmaleki, M. Shahabi, Y. Yin, and N. Masoud, “Itinerary plan-
ning for cooperative truck platooning,Available at SSRN 3481598,
[5] K. Woodhouse, “U-m is turning ann arbor into the
world’s largest lab for wireless vehicle communication,”
Apr 2014. [Online]. Available:
arbor/2014/04/transportation.html#incart m-rpt-2
[6] J. Kianfar and P. Edara, “Placement of roadside equipment in con-
nected vehicle environment for travel time estimation,Transportation
Research Record, vol. 2381, no. 1, pp. 20–27, 2013.
[7] Y. Wang, N. Masoud, and A. Khojandi, “Real-time sensor anomaly
detection and recovery in connected automated vehicle sensors,arXiv
preprint arXiv:1911.01531, 2019.
[8] F. van Wyk, Y. Wang, A. Khojandi, and N. Masoud, “Real-time sensor
anomaly detection and identification in automated vehicles,” IEEE
Transactions on Intelligent Transportation Systems, 2019.
[9] Y. Wang, N. Masoud, and A. Khojandi, “Anomaly detection in con-
nected and automated vehicles using an augmented state formulation,”
arXiv preprint arXiv:2004.09496, 2020.
[10] A. Hussein, F. Garc´
ıa, and C. Olaverri-Monreal, “Ros and unity based
framework for intelligent vehicles control and simulation,” in 2018
IEEE International Conference on Vehicular Electronics and Safety
(ICVES). IEEE, 2018, pp. 1–6.
[11] L. Zhao and J. Sun, “Simulation framework for vehicle platooning
and car-following behaviors under connected-vehicle environment,”
Procedia-Social and Behavioral Sciences, vol. 96, pp. 914–924, 2013.
[12] S. I. Guler, M. Menendez, and L. Meier, “Using connected vehicle
technology to improve the efficiency of intersections,Transportation
Research Part C: Emerging Technologies, vol. 46, pp. 121–131, 2014.
[13] J. Lee and B. Park, “Development and evaluation of a cooperative
vehicle intersection control algorithm under the connected vehicles
environment,IEEE Transactions on Intelligent Transportation Sys-
tems, vol. 13, no. 1, pp. 81–90, 2012.
[14] D. Bullock, B. Johnson, R. B. Wells, M. Kyte, and Z. Li, “Hardware-
in-the-loop simulation,” Transportation Research Part C: Emerging
Technologies, vol. 12, no. 1, pp. 73–89, 2004.
[15] M. R. Zofka, S. Klemm, F. Kuhnt, T. Schamm, and J. M. Z¨
“Testing and validating high level components for automated driving:
simulation framework for traffic scenarios,” in 2016 IEEE Intelligent
Vehicles Symposium (IV). IEEE, 2016, pp. 144–150.
[16] W. M. Griggs, R. H. Ord´
nez-Hurtado, E. Crisostomi, F. H¨
K. Massow, and R. N. Shorten, “A large-scale sumo-based emulation
platform,” IEEE Transactions on Intelligent Transportation Systems,
vol. 16, no. 6, pp. 3050–3059, 2015.
[17] K. S. Swanson, A. A. Brown, S. N. Brennan, and C. M. LaJambe,
“Extending driving simulator capabilities toward hardware-in-the-loop
testbeds and remote vehicle interfaces,” in 2013 IEEE Intelligent
Vehicles Symposium Workshops (IV Workshops). IEEE, 2013, pp.
[18] M. Tideman and M. Van Noort, “A simulation tool suite for devel-
oping connected vehicle systems,” in 2013 IEEE Intelligent Vehicles
Symposium (IV). IEEE, 2013, pp. 713–718.
[19] J. L. Pereira and R. J. Rossetti, “An integrated architecture for
autonomous vehicles simulation,” in Proceedings of the 27th annual
ACM symposium on applied computing. ACM, 2012, pp. 286–292.
[20] N. Koenig and A. Howard, “Design and use paradigms for gazebo, an
open-source multi-robot simulator,” in 2004 IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS)(IEEE Cat. No.
04CH37566), vol. 3. IEEE, 2004, pp. 2149–2154.
[21] B. Cronin, “Vehicle based data and availability,” Oct 2012. [Online].
data availability.pdf
[22] T. Foote, “Demo of prius in ros/gazebo,” Apr 2017. [Online].
Available: demo/
[23] W. Sultani and J. Y. Choi, “Abnormal traffic detection using intelligent
driver model,” in 2010 20th International Conference on Pattern
Recognition. IEEE, 2010, pp. 324–327.
[24] A. Kesting, M. Treiber, and D. Helbing, “Enhanced intelligent driver
model to access the impact of driving strategies on traffic capacity,”
Philosophical Transactions of the Royal Society A: Mathematical,
Physical and Engineering Sciences, vol. 368, no. 1928, pp. 4585–
4605, 2010.
[25] X. Liu, G. Zhao, N. Masoud, and Q. Zhu, “Trajectory planning
for connected and automated vehicles: Cruising, lane changing, and
platooning,” arXiv preprint arXiv:2001.08620, 2020.
[26] F. van Wyk, A. Khojandi, and N. Masoud, “Optimal switching policy
between driving entities in semi-autonomous vehicles,Transportation
Research Part C: Emerging Technologies, vol. 114, pp. 517–531, 2020.
[27] E. Zhang and N. Masoud, “Increasing gps localization accuracy with
reinforcement learning,” IEEE Transactions on Intelligent Transporta-
tion Systems, 2020.
... Gazebo is another open-source, scalable, flexible, and multi-robot 3D simulator that relies on three main libraries: physics, rendering, and communication libraries. It can provide high-precision physics for roboticsrelated simulation [131]. Swanson et al. created a hardwarein-the-loop (HiL) simulator and indicated that although ROS already included a graphical interface RViz for visualization, Gazebo was necessary because it could model much more accurate physics. ...
... The listener and publisher constitute a scalable architecture that allows multiple nodes to control agents, which provides more attack surfaces for evaluation. Zhang et al. developed a Gazebo-based vehicle-to-everything (V2X) platform for the simulation of CAV environments [131]. On top of Gazebo, they extended a communication module receiving and sending information between vehicles and roadside units (RSU). ...
Connected and automated vehicles (CAVs) provide various valuable and advanced services to manufacturers, owners, mobility service providers, and transportation authorities. As a result, a large number of CAV applications have been proposed to improve the safety, mobility, and sustainability of the transportation system. With the increasing connectivity and automation, cybersecurity of the connected and automated transportation system (CATS) has raised attention to the transportation community in recent years. Vulnerabilities in CAVs can lead to breakdowns in the transportation system and compromise safety (e.g., causing crashes), performance (e.g., increasing congestion and reducing capacity), and fairness (e.g., vehicles fooling traffic signals). This paper presents our perspective on CATS cybersecurity via surveying recent pertinent studies focusing on the transportation system level, ranging from individual and multiple vehicles to the traffic network (including infrastructure). It also highlights threat analysis and risk assessment (TARA) tools and evaluation platforms, particularly for analyzing the CATS cybersecurity problem. Finally, this paper will provide valuable insights into developing secure CAV applications and investigating remaining open cybersecurity challenges that must be addressed.
... Several research efforts have been reported in literature on creating a digital version of the real environment using physics-based simulation software. Zhang and Masoud [14] used Gazebo to create a virtual environment to test CAVs due to its ability to capture microscopic vehicle movement. The authors selected Gazebo, rather than a game engine, due to concerns about rigorously replicating the full dynamics of an individual vehicle. ...
... We selected Unity over existing simulation packages, such as Gazebo, as it is easy to deploy and performs well on a variety of platforms. Unlike Zhang and Masoud [14], our interest lies in the systemlevel coordination of CAVs, not the particular dynamics of any individual CAV. Unity also relies heavily on the entitycomponent paradigm of software design, which grants us incredible flexibility in the design and control of vehicles in the virtual environment. ...
... Examples include the SAE J2735 for DSRC message format [3] be exchanged to provide a safer and more efficient use of the transportation network, such as road layout, safety alerts, signal phase and timing, and data on other road users and vehicles. Evidently, the communication component of V2X requires in-depth testing, analysis [5], and simulation [6], as bandwidth and latency can play a major role in the overall quality of the system [7]. In this paper, we focus on the task of V2X-enabled CP, i.e., multiple perception-capable units such as On-board Perception (OBP) and Roadside Perception Units (RSPU) augment their perception via V2X communication [8]. ...
Full-text available
In this paper, we introduce the notion of Cooperative Perception Error Models (coPEMs) towards achieving an effective and efficient integration of V2X solutions within a virtual test environment. We focus our analysis on the occlusion problem in the (onboard) perception of Autonomous Vehicles (AV), which can manifest as misdetection errors on the occluded objects. Cooperative perception (CP) solutions based on Vehicle-to-Everything (V2X) communications aim to avoid such issues by cooperatively leveraging additional points of view for the world around the AV. This approach usually requires many sensors, mainly cameras and LiDARs, to be deployed simultaneously in the environment either as part of the road infrastructure or on other traffic vehicles. However, implementing a large number of sensor models in a virtual simulation pipeline is often prohibitively computationally expensive. Therefore, in this paper, we rely on extending Perception Error Models (PEMs) to efficiently implement such cooperative perception solutions along with the errors and uncertainties associated with them. We demonstrate the approach by comparing the safety achievable by an AV challenged with a traffic scenario where occlusion is the primary cause of a potential collision.
... Examples include the SAE J2735 for DSRC message format [3] be exchanged to provide a safer and more efficient use of the transportation network, such as road layout, safety alerts, signal phase and timing, and data on other road users and vehicles. Evidently, the communication component of V2X requires in-depth testing, analysis [5], and simulation [6], as bandwidth and latency can play a major role in the overall quality of the system [7]. In this paper, we focus on the task of V2X-enabled CP, i.e., multiple perception-capable units such as On-board Perception (OBP) and Roadside Perception Units (RSPU) augment their perception via V2X communication [8]. ...
... Many algorithms such as trajectory planning and control were verified using these systems. Zhang E et al. [16][17][18][19] enhanced the evaluation capability of the testbed. However, their OBU tests are all small in number and only guarantee functional tests in environments with good communication quality, but not in congested environments. ...
Full-text available
Vehicle to Everything (V2X) technology is fast evolving, and it will soon transform our driving experience. Vehicles employ On-Board Units (OBUs) to interact with various V2X devices, and these data are used for calculation and detection. Safety, efficiency, and information services are among its core uses, which are currently in the testing stage. Developers gather logs during the real field test to see if the application is fair. Field testing, on the other hand, has low efficiency, coverage, controllability, and stability, as well as the inability to recreate extreme hazardous scenarios. The shortcomings of actual road testing can be compensated for by indoor testing. An HIL-based laboratory simulation test framework for V2X-related testing is built in this study, together with the relevant test cases and a test evaluation system. The framework can test common applications such as Forward Collision Warning (FCW), Intersection Collision Warning (ICW) and others, as well as more advanced features such as Cooperative Adaptive Cruise Control (CACC) testing and Global Navigation Satellite System (GNSS) injection testing. The results of the tests reveal that the framework (CarTest) has reliable output, strong repeatability, the capacity to simulate severe danger scenarios, and is highly scalable, according to this study. Meanwhile, for the benefit of researchers, this publication highlights several relevant HIL challenges and solutions.
Digital twin cities are frequently used in vehicle and traffic simulations to render realistic on-road driving scenarios under various traffic and environmental conditions. These digital twins provide a high-fidelity replica of the physical world (e.g., buildings, roads, infrastructures, traffic) to create three-dimensional (3D) virtual-physical environments to support various emerging vehicle and transportation technologies such as connected and automated vehicles. These virtual environments provide a cost-effective digital proving ground to evaluate, validate, and test emerging technologies that include control algorithms, localization, perception, and sensors. Replicating a real-world traffic scenario in a digital twin using a traditional 3D modeling approach is a time-consuming and labor-intensive effort. This paper presents a semi-automated spatial framework to construct realistic 3D digital twin cities to support autonomous driving research using readily available geographic information system (GIS) data and 3D prefabricated (prefab) models. We start with a comprehensive review of geospatial data sources of essential digital entities required in a 3D digital twin city and present an integrated GIS-3D modeling pipeline using customized QGIS/GDAL and Blender scripting in Python. The pipeline outputs are realistic 3D digital twin cities compatible with common vehicle simulation software, such as CARLA and IPG CarMaker. The paper closes with a showcase to demonstrate the quality and usability of a digital twin city created to replicate the Shallowford Road corridor in Chattanooga in both Unity and Unreal engine-based virtual environment. The generated digital twin city can be applied to a hardware-in-the-loop simulation environment with an actual testing vehicle to facilitate autonomous driving research.
Conference Paper
Full-text available
Freeway ramp merging is a challenging task for an individual vehicle (in particular a truck) and a critical aspect of traffic management that often leads to bottlenecks and accidents. While connected and automated vehicle (CAV) technology has yielded efficient merging strategies, most of them overlook the differentiation of vehicle types and assume uniform CAV presence. To address this gap, our study focuses on enhancing the merging efficiency of heavy-duty trucks in mixed traffic environments. We introduce a novel multi-human-in-the-loop (MHuiL) simulation framework, integrating the SUMO traffic simulator with two game engine-based driving simulators, enabling the investigation of interactions between human drivers in diverse traffic scenarios. Through a comprehensive case study analyzing eight scenarios, we assess the performance of a connectivity-based cooperative ramp merging system for heavy-duty trucks, considering safety, comfort, and fuel consumption. Our results demonstrate that guided trucks exhibit advantageous characteristics, including an enhanced safety margin with larger gaps by 23.2%, a decreased speed deviation by 30.4% facilitating smoother speed patterns, and a reduction in fuel consumption by 3.4%, when compared with non-guided trucks. This research offers valuable insights for the development of innovative approaches to improve truck merging efficiency, enhancing overall traffic flow and safety.
Many vehicle-to-everything (V2X) applications and use cases require their feasibility to be simulated, tested, and validated in realistic traffic scenarios and under various network conditions before real-time testbed implementation. As cellular-V2X (C-V2X) becomes a superior technology for future connected and autonomous vehicles, the need for a simulation framework, which integrates traffic and network simulators with a realistic channel model, becomes more evident. The challenge is to overcome existing simulation platforms’ weaknesses and improve simulation results’ accuracy while preserving flexibility with manageable implementation complexity. This paper proposed a top-down approach for building reference scenarios with macroscopic and microscopic layers, which interweaves traffic, network, and channel simulators. The basis for the proposed simulation framework is realistic scenario data, which provides the input to the road traffic simulator and the radio channel simulator. The core of the whole simulator is the network simulator interacting with the two other simulators via dedicated interfaces. The road traffic simulator generates the vehicles’ positions and is forwarded to the network simulator, which requests the radio channel simulator for the path loss values on the transmitter–receiver (TX-RX) link. As a proof of concept, the simulation results focused on the load and interference analysis in the absence and presence of V2X traffic.
Full-text available
Autonomy and connectivity are considered among the most promising technologies to improve safety and mobility and reduce fuel consumption and travel delay in transportation systems. In this paper, we devise an optimal control-based trajectory planning model that can provide safe and efficient trajectories for the subject vehicle while incorporating platoon formation and lane-changing decisions. We embed this trajectory planning model in a simulation framework to quantify its fuel efficiency and travel time reduction benefits for the subject vehicle in a dynamic traffic environment. Specifically, we compare and analyze the statistical performance of different controller designs in which lane changing or platooning may be enabled, under different values of time (VoTs) for travelers. Results from extensive numerical experiments indicate that our design can not only provide first-hand cost savings for the subject vehicle but also second-hand savings for vehicles upstream of the subject vehicle. Experiments also highlight that lane changing and platooning can both offer benefits, depending on the relative values of fuel cost and the traveler's VoT: with a small VoT, the fuel efficiency benefits of platooning outweigh time savings offered by lane changing. However, a vehicle with a high VoT may find it more beneficial to travel outside of a platoon and complete its trip faster by leveraging lane changes.
Full-text available
In the future, autonomous vehicles are expected to safely move people and cargo around. However, as of now, automated entities do not necessarily outperform human drivers under all circumstances, particularly under certain road and environmental factors such as bright light, heavy rain, poor quality of road and traffic signs, etc. Therefore, in certain conditions it is safer for the driver to take over the control of the vehicle. However, switching control back and forth between the human driver and the automated driving entity may itself pose a short-term, elevated risk, particularly because of the out of the loop (OOTL) issue for humans. In this study, we develop a mathematical framework to determine the optimal driving-entity switching policy between the automated driving entity and the human driver. Specifically, we develop a Markov decision process (MDP) model to prescribe the entity in charge to minimize the expected safety cost of a trip, considering the dynamic changes of the road/environment during the trip. In addition, we develop a partially observable Markov decision process (POMDP) model to accommodate the fact that the risk posed by the immediate road/environment may only be partially observed. We conduct extensive numerical experiments and thorough sensitivity and robustness analyses, where we also compare the expected safety cost of trips under the optimal and single driving entity policies. In addition, we quantify the risks associated with the policies, as well as the impact of miss-estimating road/environment condition risk level by the driving entities, and provide insights. The proposed frameworks can be used as a policy tool to identify factors that can render a region suitable for level four autonomy.
Full-text available
Automated vehicles are envisioned to be an integral part of the next generation of transportation systems. Whether it is striving for full autonomy or incorporating more advanced driver assistance systems, high-accuracy vehicle localization is essential for automated vehicles to navigate the transportation network safely. In this paper, we propose a reinforcement learning framework to increase GPS localization accuracy. The framework does not make rigid assumptions on the GPS device hardware parameters or motion models, nor does it require infrastructure-based reference locations. The proposed reinforcement learning model learns an optimal strategy to make "corrections" on raw GPS observations. The model uses an efficient confidence-based reward mechanism, which is independent of geolocation, thereby enabling the model to be generalized. We incorporate a map matching-based regularization term to reduce the variance of the reward return. The reinforcement learning model is constructed using the asynchronous advantage actor-critic (A3C) algorithm. A3C provides a parallel training protocol to train the proposed model. The asynchronous reinforcement learning strategy facilitates short training sessions and provides more robust performance. The performance of the proposed model is assessed by comparing it with an extended Kalman filter algorithm as a benchmark model. Our experiments indicate that the proposed reinforcement learning model converges fast, has less prediction variance, and can localize vehicles with 50% less error compared to the benchmark Extended Kalman Filter model.
Full-text available
In this paper we propose a novel observer-based method to improve the safety and security of connected and automated vehicle (CAV) transportation. The proposed method combines model-based signal filtering and anomaly detection methods. Specifically, we use adaptive extended Kalman filter (AEKF) to smooth sensor readings of a CAV based on a nonlinear car-following model. Using the car-following model the subject vehicle (i.e., the following vehicle) utilizes the leading vehicle's information to detect sensor anomalies by employing previously-trained One Class Support Vector Machine (OCSVM) models. This approach allows the AEKF to estimate the state of a vehicle not only based on the vehicle's location and speed, but also by taking into account the state of the surrounding traffic. A communication time delay factor is considered in the car-following model to make it more suitable for real-world applications. Our experiments show that compared with the AEKF with a traditional χ²-detector, our proposed method achieves a better anomaly detection performance. We also demonstrate that a larger time delay factor has a negative impact on the overall detection performance.
Full-text available
A cooperative truck platoon is a set of virtually linked trucks driving with a small intra-vehicle gap enabled by connected and automated vehicle technologies. One of the primary benefits of truck platooning is energy savings due to the reduction of aerodynamic drag on the platooned vehicles. The focus of this paper is on scheduling travel itineraries of a given set of trucks to maximize platooning opportunities and the corresponding energy savings. By constructing a time-expanded network, we formulate the problem as a minimum concave-cost network flow problem, and devise a number of solution methodologies to find the optimal or high-quality solutions. The solution methodologies include an outer approximation algorithm for solving a mixed-integer convex minimization reformulation of the problem, a dynamic-programming-based heuristic scalable to large-scale instances, and a fast approximation algorithm with guaranteed performance for a restrictive version of the problem. All the proposed algorithms are examined and benchmarked on medium to large networks under various test scenarios. The numerical results demonstrate the efficiency of the proposed methods and their applicability in real-world settings.
Full-text available
Connected and automated vehicles (CAVs) are expected to revolutionize the transportation industry, mainly through allowing for a real-time and seamless exchange of information between vehicles and roadside infrastructure. Although connectivity and automation are projected to bring about a vast number of benefits, they can give rise to new challenges in terms of safety, security, and privacy. To navigate roadways, CAVs need to heavily rely on their sensor readings and the information received from other vehicles and roadside units. Hence, anomalous sensor readings caused by either malicious cyber attacks or faulty vehicle sensors can result in disruptive consequences and possibly lead to fatal crashes. As a result, before the mass implementation of CAVs, it is important to develop methodologies that can detect anomalies and identify their sources seamlessly and in real time. In this paper, we develop an anomaly detection approach through combining a deep learning method, namely convolutional neural network (CNN), with a well-established anomaly detection method, and Kalman filtering with a χ <sup xmlns:mml="" xmlns:xlink="">2</sup> -detector, to detect and identify anomalous behavior in CAVs. Our numerical experiments demonstrate that the developed approach can detect anomalies and identify their sources with high accuracy, sensitivity, and F1 score. In addition, this developed approach outperforms the anomaly detection and identification capabilities of both CNNs and Kalman filtering with a χ <sup xmlns:mml="" xmlns:xlink="">2</sup> -detector method alone. It is envisioned that this research will contribute to the development of safer and more resilient CAV systems that implement a holistic view toward intelligent transportation system (ITS) concepts.
Full-text available
To date, three major hurdles have hindered widespread adoption of electric vehicles (EVs): the high cost of batteries, insufficient public charging infrastructure, and the limited driving range of EVs. This study overcomes these three hurdles by introducing a new concept of vehicle-to-vehicle wireless power transfer (V2V WPT) between EVs to facilitate frequent, real-time, and on-demand charging. V2V WPT is enabled by the connected and automated vehicle technology and the sharing economy, and facilitates transfer of power between peer EVs. We formulate the problem of routing, scheduling, and matching vehicles in a V2V WPT platform on a energy-time expanded network, and devise a dynamic programming solution methodology to find the solution efficiently. We quantify the impact of adopting this technology in terms of system-wide energy savings, charging infrastructure requirements, and travel times, and investigate the possibility of reducing battery capacity in EVs as a result of this technology.
Conference Paper
Full-text available
Intelligent vehicles simulations are utilized as the initial step of experiments before the deployment on the roads. Nowadays there are several frameworks that can be used to control vehicles, and Robot Operating System (ROS) is the most common one. Moreover, there are several powerful visualization tools that can be used for simulations, and Unity Game Engine is on the top of the list. Accordingly, this paper introduces a methodology to connect both systems, ROS and Unity, thus linking the performance in simulations and real-life for better analogy. Additionally, a comparative study between GAZEBO simulator and Unity simulator, in terms of functionalities and capabilities is shown. Last but not least, two use cases are presented for validation of the proposed methodology.
Full-text available
Autonomous vehicles are expected to shift not only the driving paradigms but also the notion of vehicle ownership. Although autonomous vehicles are believed to introduce many safety, mobility, and environmental benefits, they will be initially priced relatively highly. This paper assesses the potential for circumventing this barrier by promoting a shared ownership program in which households form clusters that share the ownership and ridership of a set of autonomous vehicles. Such a program will increase the utilization rate of vehicles, making ownership of autonomous vehicles more economical. We study parameters that affect the benefits expected from autonomous vehicles, and introduce policy directions that can boost these benefits.