ROS-based Architecture for Autonomous Intelligent Campus Automobile (iCab)

Abstract

This paper presents a smart research platform to foster intelligent transportation systems in urban environments, called iCab (Intelligent Campus Automobile) autonomous vehicle. The aim of the paper is to describe the initial steps to achieve a functional autonomous vehicle. The platform is a golf cart vehicle, E-Z-GO model, which is modified to operate in autonomous mode. The software core is based on Robot Operating System (ROS) architecture, which allows the fusion of multiple sensors data and time stamp of different devices in one embedded computer on the board of the platform. The proposed system shows the advantages of ROS-based architecture data management, such as but they are not limited to, huge data handling from the surrounding environment, computer vision system perception and laser scanner data interpretation. The sensors data are integrated with the ROS-based architecture to develop cutting-edge applications, which cope with the autonomous navigation requirements and real-time data processing. The experimental study shows that the ROS-based architecture outperforms former works in autonomous vehicles, for its portability and feasibility to create a network of autonomous vehicles, that is, the autonomous interaction of more than one vehicle in closeness environments fostering the urban mobility.

Full text

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
257
ROS-based Architecture for Autonomous
Intelligent Campus Automobile (iCab)
David Martín Gómez
Doctor en Informática, Investigador Posdoctoral
Pablo Marín Plaza
Máster Universitario en Robótica y Automatización, Doctorando
Ahmed Hussein
Máster en Ingeniería Mecatrónica, Doctorando
Arturo de la Escalera Hueso
Doctor Ingeniero Industrial, Profesor Titular
José María Armingol Moreno
Doctor Ingeniero Industrial, Catedrático de Universidad
Laboratorio de Sistemas Inteligentes
Universidad Carlos III de Madrid
Resumen
Este artículo presenta una plataforma de investigación para fomentar los
Sistemas Inteligentes de Transporte en entornos urbanos. El sistema se
denomina iCab y es el vehículo inteligente y autónomo de la Universidad
Carlos III de Madrid. El objetivo del presente artículo es la descripción de la
línea de investigación base para lograr un vehículo autónomo completamente
funcional y seguro que permita la movilidad de las personas en un entorno
urbano. La plataforma autónoma que se está desarrollando en la Universidad
Carlos III de Madrid, se basa en un carro de golf modelo EZ-GO, que se ha
automatizado para operar en modo autónomo. La arquitectura de percepción
del entorno y control de los actuadores se basa en el sistema operativo (ROS),
que permite importantes funcionalidades para la navegación autónoma, como
son: la fusión de datos de múltiples sensores para la percepción óptima del
entorno o la marca de tiempo de los diferentes dispositivos en tiempo real,
entre otras. El estudio experimental que se presenta en este artículo muestra
las ventajas de una arquitectura basada en ROS, favoreciendo la utilización
de los vehículos autónomos, por su portabilidad y viabilidad para crear redes
de vehículos autónomos, es decir, la interacción y cooperación entre vehículos
autónomos que faciliten la movilidad urbana.

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
258
Abstract
This paper presents a smart research platform to foster intelligent
transportation systems in urban environments, called iCab (Intelligent
Campus Automobile) autonomous vehicle. The aim of the paper is to describe
the initial steps to achieve a functional autonomous vehicle. The platform is
a golf cart vehicle, E-Z-GO model, which is modified to operate in autonomous
mode. The software core is based on Robot Operating System (ROS)
architecture, which allows the fusion of multiple sensors data and time stamp
of different devices in one embedded computer on the board of the platform.
The proposed system shows the advantages of ROS-based architecture data
management, such as but they are not limited to, huge data handling from
the surrounding environment, computer vision system perception and laser
scanner data interpretation. The sensors data are integrated with the ROS-
based architecture to develop cutting-edge applications, which cope with the
autonomous navigation requirements and real-time data processing. The
experimental study shows that the ROS-based architecture outperforms
former works in autonomous vehicles, for its portability and feasibility to
create a network of autonomous vehicles, that is, the autonomous interaction
of more than one vehicle in closeness environments fostering the urban
mobility.
1. Introduction
The data of the World Health Organization (WHO) show that 1.3 million of
people around the world died in 2013 due to road traffic accidents [1]. The
majority of these accidents were because of human error, which could be
avoided and minimized through using autonomous vehicles instead.
The first two main demonstrations of the capabilities of autonomous vehicles
took place in the United States through the Defense Advanced Research
Projects Agency (DARPA) to develop autonomous cars. First one was in 2004
and 2005, the DARPA Grand Challenge was two races in the desert with no
dynamic obstacles [2]. The second one was in 2007, the DARPA Urban
Challenge was a race in an urban circuit among autonomous cars, simulating
the dynamic traffic as in real urban environments [3]. Furthermore, the first
functional autonomous vehicle is the Google Self-Driving Car, it is designed
under the traffic laws after years of research and it resulted to a fully
autonomous car [4]. Autonomous vehicles continue as an important topic in
intelligent transportation systems, where a recent work shows that driver-
less vehicles could become widely available in the next 5 to 10 years [5].
Nowadays the in-vehicle applications outperform the environment perception
through computer vision and laser scanner. They overcome the most
significant technical limitations, such as robustness to face the changes in the
environmental conditions due to illumination variation, such as shadows, low

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
259
lighting conditions, night vision among others. Accordingly, perception
applications ensure the suitable robustness and safety in case of large variety
of lighting conditions and complex perception tasks [6]. Additionally, the use
of computer vision is well-established in recent researches about autonomous
vehicles; for example the route from Mannheim to Pforzheim by Mercedes
Benz S-Class car. The car navigated 103 km on the route autonomously, it
was equipped with computer vision systems and radar sensors along with
digital maps [7].
Moreover, further problems of the autonomous vehicles are autonomous
navigation and path planning. Many researchers implemented several
approaches towards solving the problem in indoors environments. The results
showed the feasibility to generate an obstacle free path from one point to
another and navigate through the generated path with the localization of the
vehicle at each point on route [8, 9, 10]. On the other-hand, for the outdoors
environments, researchers implemented several algorithms aiming to obtain
autonomous outdoor vehicle. Robot Operating System (ROS) based
architecture was used to generate the mapping and localization for the
autonomous navigation and the results outperformed other algorithms [11,
12]. ROS-based systems provide an operating system-like services to operate
robots with the fusion of multiple sensors data and time stamp of different
devices [13].
This paper presents the first steps in the implementation of a smart research
platform to foster intelligent transportation systems in urban environments
and describes the initial steps to achieve a functional autonomous vehicle.
The project main objective is to implement and improve autonomous
navigation and path planning approaches, based on image processing and
laser scanner data interpretation. The implementation is performed over a
smart ROS-based architecture, for real-time processing and communication
of the software processes in an embedded computer. The computer is placed
on the platform, golf cart vehicle, called iCab autonomous vehicle. This
structure achieves the ease of data handling of the on-board sensors, in terms
of camera and laser scanner, with the proposed ROS-based architecture to
research on navigation applications. Also, the advantage of synchronizing
low-level data by means of ROS-based systems is the use of reliable time
stamp for the data acquisition from on-board iCab devices. Hence, the ROS-
based systems allow the coordination of the drivers and middleware, which
aim to simplify the complex task of global data acquisition and sensor
synchronization.
Hence, the iCab applications can foster sensor fusion processes, which allows
the improvement of the performance of each application in high-level stages.
For this purpose, the proposed ROS-based architecture communicates the
processes with each other in order to refine information and knowledge. It

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
260
also provides higher level information to improve the decision making
process, in other words, to avoid safely the collision with an obstacle or
pedestrian in autonomous navigation. This proposed system enables the inter
process communication in an independent and modular way, it also enables
the on-board computer to run multiple and parallel algorithms in order to
achieve both low-level objectives, such as sensor data acquisition and data
preprocessing, and high-level objectives, such as pedestrian detection,
obstacle avoidance, autonomous path planning and navigation. Last but not
least, this architecture facilitates the scalability and adaptability for the
changes of the on-board technology of the iCab, to accommodate novel
sensors or higher requirements of the applications.
The remainder of this paper is organized into five sections. Section 2
introduces the experimental platform, emphasizing on the use of low-level
on-board devices in the iCab, followed by Section 3, which presents the
proposed ROS-based autonomous vehicle architecture. Section 4 explains the
experimental results for different scenarios that will be used in autonomous
navigation through urban environments. Finally, the conclusions and future
work are summarized in Section 5.
2. Platform Description
The selected research platform is an electric golf cart vehicle, E-Z-GO model.
It is modified to fulfil the project objectives, in terms of autonomous
navigation and path planning. Moreover, in order to achieve multiple
autonomous vehicles system, there are two identical golf carts, the first one
see Figure 1.
Fig. 1. Research platform: iCab 1

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
261
Vehicle modifications are for the mechanical and electrical systems.
Accordingly, the steering wheel is removed in order to install the motor-
encoder system and to control the vehicle direction electronically, see Figure
2. Additionally, the throttle paddle is deactivated to control the traction
electric motor of forward and backward motion through a power amplifier and
governed by a PIC microcontroller. The rotor and stator parts of the motor
are independent from each other, to facilitate the control of the power and
torque in different environments, such as hard slope roads and rough terrains.
The inputs of the microcontroller are the percentage of the maximum capacity
for the stator, rotor and desired angle of the steering wheels.
Fig. 2. iCab steering system: motor-encoder
For environment perception, the vehicle is equipped with a laser rangefinder
(SICK LMS 291). The device has over 180 degrees scanning range with 0.25
degrees angular resolution [14]. It is mounted on the front vehicle bumper at
30cm height above the ground. In order to avoid the detection of the steering
wheels, the scanning range is limited to 100 degrees at 20Hz.
Additionally, the vehicle is equipped with a stereo vision binocular camera
(Bumblebee 2). The camera has a maximum of 1032x776 pixels resolution at
20 frames per seconds [15]. It is mounted on the front windshield of the
vehicle at 160 cm height above the ground and orientation of -45 degrees.
The camera has three purposes, first to build a free road map in order to
navigate in the environment, followed by visual odometry and finally
pedestrian or obstacle detection.
These devices are connected to an on-board embedded computer. The
computer has Intel Core i7 processor and is working under Ubuntu operating

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
262
system. The display unit is a 7-inch TFT LCD touchscreen; it is installed on
the vehicle front dashboard in order to view the system's interface software,
display the current and desired locations in the map.
3. Proposed Autonomous Vehicle Architecture
3.1. Proposed Architecture
In this work, the objective is to implement a complete architecture with
various levels of complexity categorized in three layers; deliberative,
sequencing and reactive skills [16]. Figure 3 shows the architecture structure,
the advantages of this architecture are the ability to add more skills and
modify the algorithms to obtain more efficient results during the development
stage. For the architecture layers, the low-level has the simple reaction skills
in the reactive layer, which controls the actuators and read the sensors data
from the environment. It is followed by the sequencer in the hybrid layer,
which incorporates a high level behaviour through logic sequence to the low-
level layer to achieve the required behaviour. The highest level consists of
the path planner in the deliberative layer, which generates the commands for
the iCab to follow.
Fig.3. Three-tier architecture
- Reactive Skills: the initial control structure of the autonomous vehicle is
implemented in this layer, in order to move the vehicle in the environment
with basic commands such as “Move Forward”, “Move Backward”, “Turn Left”,

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
263
“Turn Right” or “Stop”. The layer inputs are the outputs of the sequencing
layer, which are conveyed one by one to generate the movement commands
outputs via ROS-Services and send them to the controller.
- Sequencing: the layer inputs are the outputs of the deliberative layer, each
input is considered as a specific task. The outputs are conveyed to the
reaction skills layer with the desired actions for the vehicle movement. The
complexity of this layer resides in the accuracy of generating simple skills
after splitting the mid-level tasks. The behaviour is formed based on the
accuracy level of these skills, in other words, low accuracy results in no
movement of the vehicle, to avoid false actions.
- Deliberative: the logic in this layer manages the desired actions for the
vehicle, in terms of localization, path planning, navigation and mapping. The
layer inputs are from the user to define the desired destination on the map,
then the layer generates the output tasks for the sequencing layer to split
them in simple skills.
3.2 ROS Packages Description
According to the former description of the architecture and with the
consideration of the previous related work, the proposed architecture is
implemented in ROS-based system. The Figure 4 shows the packages
involved in the first steps of the project as the data sensor acquisition and
the actuators.
The low-level layer is developed in C++ in a ROS package called
“movement_manager”. This node is a server that receives the iCab status
every 20ms (50Hz); in terms of encoders reading, battery voltage, heartbeat,
PID configuration elements and state errors. These readings are published by
“/movement_manager/status_info” topic, see Figure 5. It contains a custom
message, which enables other nodes to subscribe to it and operate with the
information. Additionally, the server waits other nodes to send the clientCall
to perform a specific task. As input is written an incoming topic called
“cmd_movement” which is other way to govern the actions of the iCab. The
simple reactive skills layer contains the information to activate the actuators
in moving forward, moving backward, turning left, turning right and stopping
the vehicle.
Fig. 5. ROS low-level architecture: movement manager

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
264
Fig. 4. ROS low-level architecture
For image data acquisition, there are three main packages, see Figure 6. First
one is the “camera1394stereo_node” package, which receives interpolated
data from both cameras (left and right), then it splits them into two different
name-spaces; “stereo_camera/right” and “stereo_camera/left”. The second
package is the “bumblebee2”, which receives the left and right images from
the first package without any processing as “image_raw”, then it rectifies both
and publishes them in the same name-space as “image_rect”. The last
package is the take “disparity” package, which has the rectified images as
inputs, then it generates the disparity map for the next step.
The system acquire information about the free space of the environment using
an algorithm implemented by Musleh et al. in [17]. The node “free_map”
receives the disparity map as an input, then it publishes the name-space
known as road profile by “free_road” topic. This road profile is the result of

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
265
Fig. 6. ROS low-level architecture: stereo processing
the analysis of the u-v disparity for the environment and it split the image
into free spaces and obstacles.
Last package is for the laser rangefinder, see Figure 7. The
“sicktoolbox_wapper” package is used for the scanner, which receives the
data of the laser rangefinder and publishes them as “LaserScan” messages
through the “sicklms” node.
Fig. 7. ROS low-level architecture: laser rangefinder

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
266
The graphical user interface is for the communication and control of the iCab
in this architecture, see Figure 8. This interface is implemented in ROS node
called “icab_reconfigure”, which sends the clientCall messages to the iCab
server “movement_manager” node. The interface main layout is developed
by Qt-Designer, which displays all the data acquired by the topic
“/movement_manager/status_info” and allows the user to control the vehicle
movements manually and stop in an emergency situation.
Fig. 8. iCab graphical interface v0
4. Exemplified ROS-based Architecture
The following section presents the exemplification of the proposed
architecture, where the low-level structure, the data management of the
perception devices and the time-stamp of the iCab platform are evaluated
based on manoeuvres in urban environments. The urban scenarios have been
evaluated in several experiments, however this section summarizes a
representative scenario of each set of experiments.
The results have been obtained using the iCab platform where ROS-based
architecture has been implemented in the on-board embedded computer. The
algorithms can use the movement commands that control the forward and
backward motion by adding or subtracting 5 % to the motor traction power
to a maximum limit of 40 %. The steering wheel commands control the
vehicle front wheels by adding or subtracting 5 degrees to the heading

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
267
orientation angle, to a maximum limit of 25 degrees in either direction. These
limits are selected as the safety measures during the initial steps.
Three-representative outdoor scenarios have been selected in order to show
the performance of the ROS-based architecture managing the iCab
movement. During all the scenarios, the iCab uses the laser rangefinder and
the stereo camera in three-basic reactive tasks, with all perception devices
publishing and processing data in real-time. The exemplification of the
architecture in these three cases demonstrates that perception data, the low-
level algorithms, and the movement commands are synchronized and time-
stamp aim is achieved by using proposed ROS-based architecture. The
performance of each instantiated scenario is illustrated by plotting the throttle
values, heading angle and the distance to the object versus time.
The first basic reactive task is to follow a wall on one of the roads sides, see
Figure 9(a). The iCab follows the left wall maintaining a parallel position, the
left wall is selected for its uniformity. The starting point is at 5 meters from
the wall, where the iCab motor traction commands are activated at 20 % of
the rotor maximum power, which corresponds to 17.6 cm/s, to follow the wall
for 85 seconds using laser scanner data in real-time. So, this first graph
illustrates the performance of the ROS-based architecture exemplified in an
iCab basic-task: the red curve is the distance from the vehicle to the left wall,
whilst the blue curve is the steering wheel command to maintain the motion
following the wall. The graph shows that the steering wheel command and
laser scanner data are used both in real-time by ROS-based architecture to
accomplish a basic low-level reactive task. The steering wheel commands are
perfectly synchronized with laser data to maintain the distance to the wall.
The second basic reactive task is a straight forward movement with a stop
reactive command when an obstacle (pedestrian in this case) appears in front
of the iCab, see Figure 9(b). The exemplification in this case is the perception-
action control loop based on laser scanner data and stop command in real-
time, where perception-action synchronization is embedded in the ROS-based
architecture. The iCab is moving and laser scanner detects an obstacle
(pedestrian) trying to cross the street in front of it. This exemplification of the
reactive command activation is crucial as low-level basic task in real-time for
autonomous driving within university campus vicinity with many pedestrians.
The basic reactive task stop the iCab whenever a laser scanner measurement
from all array data is minor or equal to a minimum distance of 3.5 m.
In this experiment, the pedestrian crossed in front of the vehicle twice to force
two stop reactive commands. The graph illustrates this performance. The blue
curve is the throttle percentage power; whilst the red curve is the distance to
pedestrian, where one representative distance has been selected from all
measurements from the laser array to be plotted, this distance value belongs

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
268
Fig. 9. Basic reactive tasks using laser scanner data, motor traction
commands and the steering wheel command in real time: (a) follow a wall,
(b) stop with obstacle (pedestrian)
to the laser centre measurement (that is, 0 degrees of the iCab header angle).
The graph shows that laser scanner data in front of the vehicle, and how the
stop command is activated in real-time before the pedestrian is in front of the
vehicle and the throttle power is set to zero. This reactive behaviour by ROS-
based architecture has been achieved successfully and demonstrate the
perception-action loop in real-time of the iCab vehicle in urban environments.
Following, we use data from stereo camera in order to test again the
performance of the proposed architecture in the low-level perception loop.
The implemented algorithm obtains the disparity map and free map of the
road in real-time by using both stereo images. The free map is generated by
applying a specific threshold to the v-disparity of the disparity map. Figure
10 displays in top-right the disparity map, and in the top-left the free space
(binary image), where the right side of the road appears as free space
because of the unfeasibility to distinguish it from the actual road in v-
disparity. The both stereo images are shown also in bottom-left and bottom
right area. The perception algorithm is processed and published in real-time,
that is, stereo images and processing data are available to other future
processes inside proposed ROS-based architecture.
In order to compare the free_map in two different cases, in the next
exemplification, the iCab is approaching to an outdoor exit door at the campus
university and iCab can only navigate by free space, see Figure 11. The
processed data of the stereo camera can be observed where the difference
between the road and the wall is perfectly classified and integrates into

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
269
Fig. 10. Stereo images, disparity map and free map ROS-based low-level
perception processes in real-time without obstacles to accomplish
autonomous navigation
ROS-based architecture. That is, white area corresponds to the free space to
iCab displacement; whilst black areas are objects higher than the road plane.
5. Conclusion and Future Works
This paper presents the design, development and exemplification of a ROS-
based architecture for the iCab autonomous vehicle in urban environments.
The aim of the architecture is to provide the iCab platform the capabilities to
be used as a functional intelligent transportation vehicle. The perception-
action low-level processes are accomplished by proposed architecture using
laser rangefinder, the stereo camera and motor commands, where the
exemplification of real-time data acquisition, time stamp and perception
processing has been demonstrated. That is, the exemplification of the
architecture shows the high performance of the system to obtain the
necessary data from different scenarios to accomplish basic reactive tasks.
The future aspects of research include the integration of ROS-based high-
level reasoning to accomplish path-planning, navigation and trajectory
planning tasks for autonomous movement. At which the vehicle navigates a

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
270
Fig. 11. ROS-based low-level perception processes in real-time with obstacles
information to perform an approximation manoeuvre for taking people into
iCab
given environment avoiding static obstacles and manoeuvring dynamic ones.
Moreover the iCab platform can be extended to deal with more than one
vehicle and create Multiple Vehicle Communication System (MVCS), at which
the coordination and cooperation between the vehicles is necessary to achieve
a network of autonomous transportation systems in urban environments.
Acknowledgments
This work was supported by the Spanish Government through the CICYT
projects (TRA2013-48314-C3-1-R) and Comunidad de Madrid through
SEGVAUTOTRIES (S2013/MIT-2713). The authors wish to express their
gratitude to Francisco Javier Sánchez from the Electrical Engineering
Department at Universidad Carlos III de Madrid for all his knowledge in
electronics.
References
[1] W. H. Organization et al., WHO global status report on road safety 2013:
supporting a decade of action. World Health Organization, 2013.

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
271
[2] S. Thrun, M. Montemerlo, H. Dahlkamp, D. Stavens, A. Aron, J. Diebel, P.
Fong, J. Gale, M. Halpenny, G. Hoffmann, et al., “Stanley: The robot that won
the darpa grand challenge,” Journal of field Robotics, vol. 23, no. 9, pp. 661–
692, 2006.
[3] C. Urmson, J. Anhalt, D. Bagnell, C. Baker, R. Bittner, M. Clark, J. Dolan,
D. Duggins, T. Galatali, C. Geyer, et al., “Autonomous driving in urban
environments: Boss and the urban challenge,” Journal of Field Robotics, vol.
25, no. 8, pp. 425–466, 2008.
[4] M. Harris. (2015) Google’s self-driving car pals revealed. IEEE Spectrum.
[Online]. Available: http://spectrum.ieee.org/cars-that-
think/transportation/self-driving/googles-selfdriving-car-pals-revealed
[5] M. Mitchell, “Autonomous vehicles: No drivers required,” Nature, vol. 518,
pp. 20–23, 2015.
[6] D. Martín, F. García, B. Musleh, D. Olmeda, G. Peláez, P. Marín, A. Ponz,
C. Rodríguez, A. Al-Kaff, A. De La Escalera, et al., “Ivvi 2.0: An intelligent
vehicle based on computational perception,” Expert Systems with
Applications, vol. 41, no. 17, pp. 7927–7944, 2014.
[7] J. Ziegler, P. Bender, M. Schreiber, H. Lategahn, T. Strauss, C. Stiller, T.
Dang, U. Franke, N. Appenrodt, C. Keller, et al., “Making bertha drive? an
autonomous journey on a historic route,” Intelligent Transportation Systems
Magazine, IEEE, vol. 6, no. 2, pp. 8–20, 2014.
[8] S. Shen, N. Michael, and V. Kumar, “Autonomous multi-floor indoor
navigation with a computationally constrained mav,” International
Conference on Robotics and Automation (ICRA’2011). IEEE., pp. 20–25,
2011.
[9] I. Wieser, A. V. Ruiz, M. Frassl, M. Angermann, J. Mueller, and M.
Lichtenstern, “Autonomous robotic slam-based indoor navigation for high
resolution sampling with complete coverage,” Position, Location and
Navigation Symposium-PLANS’2014. IEEE., pp. 945–951, 2014.
[10] M. A. Hossain and I. Ferdous, “Autonomous robot path planning in
dynamic environment using a new optimization technique inspired by
bacterial foraging technique,” Robotics and Autonomous Systems, Elsevier.,
vol. 64, pp. 137–141, 2015.
[11] A. Llamazares, E. Molinos, M. Ocana, and F. Herranz, “Integrating
absynthe autonomous navigation system into ROS,” International Conference
on Robotics and Automation (ICRA’2014). IEEE., 2014.

UNED Plasencia Revista de Investigación Universitaria, Vol. 12
272
[12] S. Zaman, W. Slany, and G. Steinbauer, “ROS-based mapping,
localization and autonomous navigation using a pioneer 3-dx robot and their
relevant issues,” Saudi International Electronics, Communications and
Photonics Conference (SIECPC’2011). IEEE., pp. 1–5, 2011.
[13] M. Quigley, K. Conley, B. Gerkey, J. Faust, T. Foote, J. Leibs, R. Wheeler,
and A. Y. Ng, “ROS: an open-source robot operating system,” in ICRA
workshop on open source software, vol. 3, no. 3.2, 2009, p. 5.
[14] S. S. Intelligence, “LMS200/211/221/291 laser measurement systems,”
Technical Description, 2006.
[15] P. Grey, “Bumblebee: stereo vision camera systems,” Technical
Description, 2012.
[16] R. P. Bonasso, R. J. Firby, E. Gat, D. Kortenkamp, D. P. Miller, and M. G.
Slack, “Experiences with an architecture for intelligent, reactive agents,”
Journal of Experimental and Theoretical Artificial Intelligence, vol. 9, no. 2-3,
pp. 237–256, 1997.
[17] B. Musleh, D. Martin, A. de la Escalera, and J. M. Armingol, “Visual ego
motion estimation in urban environments based on uv disparity,” in Intelligent
Vehicles Symposium (IV), 2012 IEEE. IEEE, 2012, pp. 444–449.