This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Digital Object Identifier 10.1109/ACCESS.2018.DOI

Localization and Navigation for

Autonomous Mobile Robots using
Petri Nets in Indoor Environments
FRANCISCO A. X. DA MOTA1 , MATHEUS X. ROCHA1 , JOEL J. P. C. RODRIGUES, 3,4,5
SENIOR MEMBER, IEEE, VICTOR HUGO C. DE ALBUQUERQUE2 , MEMBER, IEEE, and
AUZUIR R. DE ALEXANDRIA1
1
Industry Department, Federal Institute of Education, Science and Technology of Ceará, Fortaleza/CE, Brazil
2
Graduate Program in Applied Informatics, University of Fortaleza, Fortaleza-CE, Brazil
3
National Institute of Telecommunications (Inatel), Santa Rita do Sapucaí-MG, Brazil
4
Instituto de Telecomunicações, Portugal
5
ITMO University, Saint Petersburg, Russia
Corresponding author: Joel J. P. C. Rodrigues (e-mail: joeljr@ieee.org)
This work was supported by the National Funding from the FCT - Fundação para a Ciência e a Tecnologia through the
UID/EEA/50008/2013 Project; by the Government of Russian Federation, Grant 074-U01; by FINEP, with resources from Funttel, Grant
No. 01.14.0231.00, under the Centro de Referência em Radiocomunicações - CRR project of the Instituto Nacional de Telecomunicações
(Inatel), Brazil; by Brazilian National Council for Scientific and Technological Development (CNPq) via Grants No. 309335/2017-5,
304315/2017-6, 304790/2015-0, and 455441/2012-0.

ABSTRACT In this paper, mobile robotics and present tools used in localization, mapping, and navigation
of a mobile robot are discussed. The main purpose of this work is, given a map represented by the incidence
matrix of a Petri Net (PN), to evaluate the use Radio-Frequency IDentification (RFID) technology to
recognize the position of a robot in this map, as well as use the PN dynamics as the cognition system
of this robot. Thus, cards with RFID technology were placed at each intersection of structured environment
(labyrinth) ways. A robot equipped with an RFID reader at its bottom moves until it passes over these cards.
When this happens, the vehicle performs actions like turning right or left according to the map defined in
its algorithm. Once the actions above are performed, it goes straight to the next card. To ensure that this
happens, there is a black line connecting each card to its neighbor cards. The robot is equipped with three
infrared sensors so it can detect and follow these lines. The results show that the robot can get out of one
RFID card and reach the next one since they are connected by a black line. Without these lines, due to
limitations in the structure of the robot, it loses its way and can not return. Still, the robot is able to execute
the necessary navigation movements, in the case of stopping, moving forward, turning right and left. These
movements are correctly coordinated by the PN dynamics. The robot knows which card it is on and goes to
the next card according to the previously established map. Each path of the robot is mathematically modeled
by the incidence matrix of a Petri Net. Therefore, it managed to reach the destination in each of the four
proposed paths. The PN that represents path 1 has the same number of places and transitions and has curves
only to the left. On path 2 a right turn was added. In 3, besides the two-way curves and RP has more places
than transitions. Lastly in 4 there is a crossing in the path, that is, the robot goes through the same card
twice, making the right and different decision in each case according to the respective maps. Experimental
results show that this approach has feasibility and effectiveness.

INDEX TERMS Smart campus, Robot Perception, Localization, Navigation, Mobile Robotics, Radio-
frequency Identification (RFID), Petri Net modeling.

I. INTRODUCTION in the industry. In the latter case, the Auto-Guided Vehicles

Robots are mechatronic devices used to assist humans in (AGVs) stand out. AGVs are machines capable of moving
various activities, whether dangerous or not, repetitive and and performing various tasks without the intervention of
unhealthy [1]. They are used in indoor environments such as humans [2].
offices, hospitals, production lines, and cargo transportation This category of robots, through sensors and actuators, in-

VOLUME 4, 2016 1

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

teracts with the environment in order to move and perform its tool to model the path that the robot must follow, as well
functions in an autonomous manner. The movement is com- as the dynamics of the PN which define the actions that the
manded by the robot’s navigation system. Given a partially AGV must execute.
known environment, Siegwart and Nourbakhsh [3] present Thus, the main contributions of this work are to use radio
navigation as the ability of the mobile robot to move through frequency identification to determine the localization of the
the environment until one or a series of goal positions, based robot in relation to the path to be navigated and to use RP in
on its knowledge and information provided by its sensors, as the mapping and cognition modules of the mobile robot. The
reliable as possible. methods most presented in the literature for local localization
The navigation system of this kind of robot is typically of mobile robots are Odometry and Computational View.
constituted by modules of perception, localization, cognition The Odometry presents accumulations of error in the
and movement control. These modules are essential to the estimated position of the robot, because the wheels slide
execution of aerial, terrestrial, and aquatic autonomous robot on the surface. On the other hand, the localization through
projects. This is proven by the different areas of activity the reading of cards with RFID technology presents itself
in which this type of machine is inserted. For instance, as a solution that does not accumulate errors and indicates
companies use mobile robots to identify heating points in the localization of the robot in relation to the main map
substations or cracks in pipes. Military personnel use robots efficiently. Computational Vision is also efficient, however,
to deactivate bombs and clean hostile buildings [4], [5]. it requires considerable computing power, in order to per-
In addition, mobile robots perform tasks such as educating, form algorithms for pre-processing, segmentation, attribute
helping or entertaining [6], cleaning buildings and subways, extraction and pattern recognition in captured images. While
or transporting food and medicine [3]. the proposed localization method with cards and RFID reader
A typical mobile robot is composed by the modules: per- demands low computational processing to perform the tasks
ception, localization, navigation or movement control, and of reading the cards; check if the card is part of the predefined
motion [3]. The cycle of actions of a mobile robot begins with route, that is, if it is registered and identify which place of the
the extraction of data from the real environment (Perception RP has a card, that is, the position of the state matrix of the
Module) through the sensors coupled to the robot such as RP has a value of 1.
infra-red, ultrasound, cameras, RFID readers. This data is Another contribution of this work is to mathematically
processed and used as input parameters of algorithms so that model the paths (mapping) to be covered by the robot using
the robot can locate itself (Localization Module). After the RP. Computational vision is used principally in the literature
localization of the robot is estimated, it is compared with to create the topological map of the environment in which the
a global map and the Cognition Module determines which robot is inserted. The processes of treatment of the images
next action to take. The robot then executes movements and captured by the computer vision are greatly influenced by
navigates through the environment until it reaches the goal the illumination, whether natural or artificial. In addition,
position (Movement Control Module). movements of the robot’s trepidation instructed to the dis-
Among these four modules, localization is the problem placement process can blur and / or erase the photograph,
most frequently approached in the literature. Lim et al. [7] rendering it unusable. Mapping using RP, on the other hand,
state that a key prerequisite for success in navigating the is robust and does not suffer interference from environmental
robot is to identify exact current localization. In addition, it is conditions, because it is fully digital.
important to highlight that, besides determining its absolute Finally, another contribution is the purpose for which
position in space, localization also means constructing a map the Petri Net is used in this work. PN has been widely
and determining the position of the robot in relation to it [3]. approached in the literature as a supervisory tool to identify
Localization can be divided in local and global localiza- bottlenecks in the process. In this paper PN is used as an
tion. Overall, using the Global Positioning System (GPS) active tool in the construction of the paths to be covered by
solves the problem of global localization efficiently [8]. The the robot and not for process monitoring as is commonly used
main discussions are instead related to local localization in in the literature.
indoor environments.In this case, methods such as Odometry Thus, the general objective of this paper is, given a map
[9], Computer Vision [10], QR code [11], RFID [12], represented by the incidence matrix of a Petri Net, to use
among others, will be discussed later in this paper. In relation RFID technology to determine the position of the robot in this
to other methods, RFID technology is presented as simple, map, as well as to use the PN dynamics as the cognition sys-
cheap, efficient, and reliable. tem of this robot. It is distributed cards with RFID technology
Once the localization is carried out, the robot needs to at each intersection of labyrinth ways.A robot equipped with
know which movements to make until it reaches the goal an RFID reader at its bottom moves until it passes over one
position. In this case, methods such Guidewire [13], Line of these cards. When this happens, the AGV performs actions
Following [9], Wall Following [14] among others, will like turning right or left according to the map defined in its
be discussed later in this paper. In this context, although algorithm. Once the actions above are performed, it goes
commonly PN is related to the idea of supervising and/or straight to the next card. To ensure that this happens, there
selecting as shown in [15] and [16], this paper presents this is a black line connecting each card to its neighbors cards.
2 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

The robot is equipped with three infrared sensors so it can Binu et al. [8] build the floor map of an indoor environment
properly follow these lines. from arrays composed by the value of the Earth magnetic
As the novelty and the focus of this work is to present a field at different points along a route. To make the map
novel method of localization and mapping for mobile robots, more accurate, Google Maps Route Boxer algorithm is used.
the practical experiments of this work are carried out in a Then, an Android application is used for tracking localization
structured and controlled environment, free of obstacles. in that environment, based on the said map (stored in a
This paper is organized as follows: in section II, mobile database).
robotics are approached and discussed, then techniques used Awad et al. [24] uses a Wi-Fi access point to set the lo-
in robot navigation are presented and compared, more specif- calization. When a device connects to a Wi-Fi network, there
ically related to mapping, localization and movement control; is an Received Signal Strength Indication (RSSI). This, in
In the next section, Petri Nets, Identification by Radio Fre- turn, has a mathematical relationship with the distance to the
quency, Arduino and the other materials and methods used in access point. Thus, from a uniform collection of RSSI data in
this research are described; In sections IV and V it is shown a known environment, the article proposes an optimization of
the results and conclusion obtained regarding the application the WiMAP algorithm (Where is My Access Point) to define
of localization and navigation systems using RFID and PNs the position of the autonomous mobile robot. Tests were
in a mobile robot controlled by an Arduino electronic board. performed with the robot in several scenarios and the results
Finally, it has been the bibliographical references. show that the optimized version of the WiMAP algorithm
surpasses the previous one in both time and localization
accuracy.
II. RELATED WORKS
Xu and Chou [29] join Wi-Fi fingerprint and AMCL
Mobile robotics is expanding and its contribution is present
to discover the exact localization of the robot. The Wi-Fi
in many areas, such as exploration of environments, security,
fingerprint algorithm makes a rough estimate of localization
education, entertainment, performing of dangerous tasks,
based on the strength of the Wi-Fi signal. Thus, for a more
among many other applications.
accurate localization, it uses the Monte Carlo Localization
Thus, for each type of situation there is a suitable mobile Algorithm (AMCL). The idea is to integrate the information
robot. Evaluating the indoor case, in an industrial environ- of the environment map, odometry and Inertial Measurement
ment, robots with wheels are most commonly used. In case Unit (IMU) through AMCL. The experimental results show
of offices, houses, and restaurants, humanoid robots can be that the Wi-Fi-AMCL method can effectively shorten the
used. When speaking of environments with uneven terrain iterative time and improve the localization accuracy. Error in
with ramps, slopes and protrusions the ideal are robots with this case is less than 50 cm.
treads or hexapod robots. As a matter of fact, independent of Dewi et al. [25] apply Neural Networks in a control mesh
the environment, a characteristic common to AGVs is the ex- of a mobile robot, considering the kinematic and dynamical
istence of a navigation system. More specifically, attention is modeling of the robot as input data. The output is a Jacobian
focused on the robot’s mapping, localization, and movement matrix (velocities) and hence, the localization of the robot is
control modules. The localization and navigation of mobile known. It makes a comparison with other studies in which
robots have been widely studied according to an extensive sensor data are considered as input and disregard the kine-
published literature, as shown in Table 1. matic and dynamic models of the robot. The results show
The idea of localization, whether local or global, is closely better performance for controller whose data of the kinematic
associated with the existence of a map. This map can be modeling and dynamics of the robot are considered.
metric or topological. Metric maps capture the geometric Xu et al. [30] use DM (Data Matrix) tags equally spaced
properties of the environment, whereas topological maps at 50 cm along the ground. A robot equipped with a camera
describe the connectivity of different places [31]. passes over these tags and identifies these landmarks, there-
Cordeiro [9] estimates the absolute position of the robot fore knowing its relative position in that indoor environment.
based on odometry. Besides odometry, Santos [10] uses So the idea is to use these tags as landmarks to review the data
computational vision for the same purpose. Regarding the obtained by odometry using Unscented Kalman Filtering
use of computational vision, Bessa et al. [1] use pattern (UKF). This technique is compared to the conventional use
recognition techniques in omnidirectional images to estimate of Odometry and the results for robot localization are better,
the localization of the robot. showing that the method is feasible and effective.
Yuan et al. [22] make simultaneous mapping and local- Zhang et al. [11] use a camera at the top of the robot to
ization based on data from an RGB-D camera and nonholo- read more quickly the QR codes that exist in the ceiling of the
nomic constraints path planning algorithm. It is proposed an environment. Thus, the localization of the robot is estimated
Extended Kalman Filter (EKF) powered by data encoders based on its distance to these landmarks. Also, to avoid
and landmarks to estimate the position and pose of the robot. collisions, it uses a Laser Ranger Finder (LRF). Dijkstra
Laser is used to identify static and moving obstacles, so that algorithm is used to plan global path and Dynamic Window
if the object is too close, the robot takes action to avoid Approach (DWA) to plan local path, which is based on
collision. the package Navigation provided by ROS (Robot Operating
VOLUME 4, 2016 3

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

TABLE 1. Related works.

Author Mapping Localization Cognition

Robot uses the reading of 3
RFID tags located at fixed points.
Abdelgawad (2014) [12] - Localization is estimated from the -
time the signal spends to exit the
tags and reach the robot
Cordeiro (2014) [9] - Odometry Line Following
Cognitive Mapping Algorithm
Ismail et al (2014) [14] - - as an alternative way for the
navigation system of the robot
Pattern recognition techniques in
Bessa et al (2015) [1] - -
omnidirectional images
Movement control system of a
Karakurt et al (2015) [17] - -
spider robot based on Petri Nets
Lim et al (2015) [7] - Particle filter algorithm -
Kinect, in addition to an encoder, to
Mardiyanto et al (2015) [18] - -
compose the system for creating a 2D map
Path planning is made by
Dijkstra algorithm and
Zhang et al (2015) [11] 2D map is built from a laser range finder QR code
Dynamic Window
Approach (DWA)
Fuzzy Petri nets (FPNs) with embedded
Baldoni et al (2016) [19] - -
memory module
The robot moves by sliding
Bao et al (2016) [13] - -
through a guidewire
Floor map built from arrays composed
by the value of the Earth magnetic
Binu et al (2016) [8] - -
field at different points along a route
and Google Maps Route Boxer algorithm
Robot navigates through
an environment from
Boniardi et al (2016) [20] - - the sketch of a trajectory
drawn manually on an
interface created by them
RFID and RGB-D for
Nguyen et al (2016) [21] - -
object localization
Simultaneous mapping and location based on
Yuan et al (2016) [22] data from an RGB-D camera and nonho- -
lonomic constraints path planning algorithm
The robot moves to the
coordinate (rooms) determined
Avutu et al (2017) [23] - -
by the touch point of the user
on the touch screen device
Awad et al (2017) [24] - Wi-Fi access point -
Apply Neural Networks in
a control mesh of a mobile
Dewi et al (2017) [25] - - robot, considering the
kinematic and dynamical
modeling of the robot
Fuzzy Logic to determine
actions to be performed by
Furlan et al (2017) [26] - - a humanoid robot and Petri
Nets are used to monitor
the navigation of the robot
Mohanty et al (2017) [2] - - Deep-Q-Learning
Fixed RFID tag, in which a robot
moves along a known path and
performs successive readings of the
Motroni et al (2017) [27] - same tag. The distances between the -
robot and the tag are calculated for
each reading. Thus, it is possible to
know the relative localization.
Odometry, laser 2D and RGB-D camera to
Wang et al (2017) [28] simulate and simultaneously generate a 3D -
map of the path to be traveled by a robot
Join Wi-Fi fingerprint and AMCL to
Xu and Chou (2017) [29] - -
discover the exact localization
Odometry and Unscented
Xu et al (2017) [30] - -
Kalman Filtering
This work (2018) Incidence matrix of a Petri Net Cards and RFID reader Dynamic of the Petri Net

4 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

System). strategies between sites to construct the model of the environ-

Abdelgawad [12] presents a robot which uses the reading ment with a Fuzzy Petri Net. According to the same author,
of 3 RFID tags located at fixed points. Thus, from the time because the application of inference tools is very useful for
the signal spends to exit the tags and reach the robot (Time route planning. Karakurt et al. [17] propose the movement
Difference Of Arrival - TDOA), its localization is estimated. control system of a spider robot based on Petri Nets.
Nguyen et al. [21] use RFID and RGB-D for localization. Batista [35] also uses Petri Nets to model the navigation
Uses only RFID for rough object localization. Thus, it obtains system of a mobile robot. However, he uses Colored Petri
a more limited region where the object is found, which in Nets (CPN), which is justified due to the fact that no applica-
turn is projected on a 2D map from an RGB-D of a Kinect. tion of CPN to the proposed problem had been found.
Finally, it implements the HOG-SVM technique to detection Alexopoulos et al. [36] use associative memory for the
of objects in the RoS (Region of Search) of the RGB image. robot to remember the paths it has traveled previously within
As the author sates, this technique decreases computational a known environment. In addition, it learns new (unknown
processing, avoids false positives and false negatives. paths on–line). The more the robot moves, the more reliable
Motroni et al. [27] present a method with a fixed RFID tag, the results. Navigation monitoring is done through Modified
in which a robot moves along a known path and performs Petri nets.
successive readings of the same tag. The distances between
Bao et al. [13] presents an interesting navigation model,
the robot and the tag are calculated for each reading. Thus,
in which the robot moves by sliding through a guide-wire.
it is possible to know the relative localization of the robot.
Meanwhile Cordeiro [9] uses the Line Following method,
The author also comments that other tags could be queued
which can be a line painted on the ground or a magnetic
to increase the coverage area. The accuracy is in the order of
tape as the robot’s navigation system. Since the robot follows
centimeters when at least 10 readings are taken and the robot
a predefined route with efficiency, this is a valid and sim-
is about 1 m away from the tag.
ple alternative. He also uses Kinect as a low-cost solution,
The following are some observations regarding the use
when compared to lasers, for identification of obstacles.
of the proposed localization methods. Odometry alone is
Mardiyanto et al. [18] also use Kinect, in addition to an
not reliable because the robot skips [3] and accumulates
encoder, to compose the system for creating a 2-D map for
error [32]. It is usually used with other methods to be
robot navigation.
reset at certain times along the trajectory. Meanwhile, the
Baldoni et al. [19] propose the use of a Fuzzy Petri Nets
use of the gyroscope sensor is good for short periods but,
application with built-in memory module to optimize the
just like Odometry, in long-time work condition accumulates
obstacle bypass process. This way, the time to perform the
measurement error [32].
mapping is shorter, it is possible to map and avoid obstacles
Computational vision is very efficient, however it is very
simultaneously, and the robot navigates with greater fluidity,
much influenced by ambient lighting. Also, GPS is widely
avoiding unnecessary actions.
used for global localization, however it has low accuracy
in indoor environments. Meanwhile, the Kalman Filter (KF) Wang et al. [28] used odometry, laser2D and RGB-D
presents excellent localization results, provided the system is camera to simulate and simultaneously generate a 3D map
linear and Gaussian. However, these two conditions greatly of the path to be traveled by a robot with wheels in an
limit the use of KF in practice [33]. unstructured indoor environment. It is worth mentioning that
The Monte Carlo algorithm overcomes the KF deficien- the use of camera and laser increases the costs of a project,
cies, since it can be applied in non-linear and non-Gaussian as well as requires greater computational power. These two
systems. It is one of the most widely used and efficient meth- questions together can make a project unfeasible. The results
ods of localization, but in large environments there is the risk of simulations in a real environment proved the efficiency and
of particle degradation, where the weight of many particles effectiveness of the system.
becomes very small over a large number of iterations [29]. Furlan et al. [26] use Fuzzy Logic to determine actions to
QR code is cheap and quite efficient as a landmark and be performed by a humanoid robot, such as turning left or
even to determine the absolute position of the robot, but if right. Petri Nets are used to monitor the navigation of the
the QR code is obstructed, even partially, by a sheet of paper, robot. Meanwhile Boniardi et al. [20] present an interesting
for example, the camera can not detect it correctly. proposal in which the robot navigates through an environ-
In this paper, it is presented RFID to determine the local ment from the sketch of a trajectory drawn manually on an
position of the robot. RFID stands out positively because it interface created by them.
does not accumulate errors, is not influenced by the luminos- Avutu et al. [23] divide a rectangular environment into 4
ity of the environment, requires low computational power, rooms. This environment is proportionally represented by
has low cost, maintains the efficiency even after a long time. a Touch Screen. Thus, the robot moves to the coordinate
Also, it functions correctly even when totally obstructed by (rooms) determined by the touch point of the user on the
thin objects such as papers, tree leaves, or plastics. touch screen device. The robot is controlled by an Arduino
About the navigation system and strategies of the robot board and equipped with an ultrasonic sensor to avoid obsta-
mapping. Lope et al. [34] use site references and navigation cles.
VOLUME 4, 2016 5

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

Ismail et al. [14] present the Cognitive Mapping Algorithm the marking Ei+1 , for i = 0,1,2,3 ..., of a PN is defined by
as an alternative way for the robot navigation system. The Equation (1).
validation of their technique is made comparing its perfor-
mance to the one of a navigation system algorithm of a wall Ei+1 = Ei + Ti ∗ G (1)
following robot.
In this work, it is proposed that the movements of the robot Where:
must be defined by the dynamics of a Black Petri Net and • E = marking matrix
Line Following, which are a simpler solutions and present • G = incidence matrix
feasibility and effectiveness when compared to previously • T = transition matrix
discussed methods. This way, regardless of the path to be followed by the
robot, the value of the first term of the initial marking matrix
III. MATERIALS AND METHODS is always 1 and the other values of the matrix are 0. The
In this work, PNs are used for modeling the trajectory that an transition matrix is composed of the indication of which
autonomous mobile robot controlled by an Arduino MEGA transitions are enabled and which transitions are not for each
board must execute. For such a robot to be able to locate itself marking Ti . Thus, for the PN presented in Figure 1, T0 is [1
during its movements, it uses Radio Frequency Identification. 0].
In the following subtopics are described in more detail each Each path has an identity, which is represented by the
of the aforementioned materials and methods. incidence matrix A. The number of rows and columns in this
matrix is determined by the number of transitions and places,
A. PETRI NET – PN respectively. Figure 2 shows a Generic Incidence Matrix G.
Discrete Event System (DES) is defined as a system that
requires the occurrence of events to change its state. The
open/close process of a door, for instance, is a DES, since it
will remain open or closed, i.e., in the same state until some
action is performed on it.
PN is a technique used for modeling DESs, whether com-
plex or not, which is performed by means of graphic or matrix FIGURE 2. Generic Incidence Matrix.
representation. In the graphical representation, there are four
basic components: where:
1) Places: Graphically represented by circles. Places store
−
tokens. anm = a+
nm − anm (2)
2) Tokens: Represented by dots, are shifted from one
From Equation (2), a+ nm is equal to the weight of the arc
place to another by the triggering of transitions.
leaving transition n and arriving at place m. In an analogous
3) Transitions: Are enabled if the immediately preceding
way, a− nm is equal to the weight of the arc leaving place m
place(s) have a number of tokens equal to or greater
and arriving at the transition n.
than the weight of the bow that departs from the
The Petri Net used in this work is of the black-token type
place(s). They are represented by rectangles.
and was developed in the TIme Petri Net Analyzer (TINA),
4) Arcs: Are represented by arrows and must necessarily
which is a toolbox for editing and analysis of Petri Nets
connect a place to a transition or vice versa. Its weight
developed by the Laboratory for Analysis and Architecture
indicates how many tokens will be taken from one
of Systems (LAAS / CNRS). Although TINA has several
place and be sent to another.
configurations for editing and simulation of Petri nets, Time
Figure 1 shows the graphical model of a generic black- Petri Nets and Automata, in this work only Petri nets are
token PN. used, whose parameters are number of places, transitions,
weight of the arcs and quantity of tokens in each place. The
path 1 has the same number of places and transitions and has
curves only to the left. On route 2 a right turn was added. In
3, besides the two-way curves and RP have more places than
transitions. And in 4 there is a crossing in the path, that is,
FIGURE 1. Model of a generic Petri Net. the robot goes through the same card twice, making the right
decision in both cases. The token of PN is represented by the
Thus, p0, p1 and p2 are places; t0 and t1 are transitions; mobile robot, so all the arcs have unit weight.
the arrows are the arcs and the black point indicates that the
place p0 has a token, which means that the initial marking B. RADIO FREQUENCY IDENTIFICATION – RFID
(the matrix representation of the number of tokens that each This form of communication became popular in World War
place has at the instant analyzed) E0 is [1 0 0]. Therefore, II, when it was used by radars to identify allied and enemy
6 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

aircrafts. Powerful antennas were used for communication Among the kinds of maps available, it was decided to use
over long distances. However, after the evolution of this topological maps, since the accuracy of the position of the
technology, today there are more compact versions with less robot is not essential for navigation. Lope et al. [34] explain
reach, used, for instance, as access keys for electronic locks. that metric models are suitable for situations where high
Regarding this latest version, to make Radio Frequency precision is required in robot navigation, whereas topological
Identification possible, two components are required: the models are best suited for situations where the precision is
RFID reader and the RFID card. The reader is responsible for not critical but sufficient for navigation.
emitting and receiving electromagnetic signals at a specific The robot is equipped with an RFID reader and must fol-
frequency such as 13.56 MHz. low the map based on the reading of RFID card. Throughout
The card can be passive (which is energized when ap- the environment there are RFID cards with Unique IDenti-
proaching the RFID reader and then sends a message) or fication (UIDs) distributed at each intersection of structured
active (has its own power supply). For the application of environment (labyrinth) ways. Thus, according to the path to
this work, it is used passive cards with a maximum range of be covered by the robot, only part of that cards are recorded
approximately 3.5 cm. As the reader approaches the label, in the algorithm, so that the robot can pass over unregistered
the latter receives an electromagnetic signal and responds cards, ignore them and move on. On the other hand, when
to it by sending a message (another electromagnetic signal). arriving at a registered card, the matrix of Future Marking is
Considering that they are operating at the same frequency, calculated according to Equation 1. From this information, it
the reader receives the electromagnetic signal and converts it is known to which place of the RP the token should go. It is
into an electrical signal that can be read by a microcontroller, considered that a place of the PN is represented by an RFID
such as the one on the Arduino board. card and the token is represented by the robot.

C. ARDUINO
F. ALGORITHM
Arduino is an electronic board composed of a microcon-
Figure 4 shows the logical sequence that controls the actions
troller and peripheral devices. Some models, such as UNO
of the mobile robot. Initially, it makes configurations such as
and MEGA, provide ease of use even for beginners. Other
the inclusion of libraries, declaration of variables, declaration
electronic components, such as sensors and actuators, can
of the incidence matrix of the PN that models the path to
be connected to it in order to automate a process. It can
be followed, start the RFID reader and the Serial Peripheral
be programmed through a open–source Arduino Software
Interface (SPI). Then the PN waits until the robot is manually
(IDE), by using a programming language similar to C ++.
placed in the initial position of the route.
D. ROBOTS After that occurs, the robot reads RFID cards. If a card
Robots are mechatronic devices capable of performing au- is unregistered, then MoveOn. If the card is registered, then
tonomous and/or controlled actions. Among the variety of verify if the robot is not on the goal point card, if not, then
robots that exist, in this work a mobile autonomous robot is a value is assigned to k, updates Ei+1 and Ti+1 , and updates
used to navigate a structured environment, based on a map. data logging. If the robot is on the goal point card tough, then
The robot used in the tests, shown in Figure 3, consists of Stop, Horn, Reset all variables and goto Start.
an acrylic chassis, two motors (12 V DC) with gearboxes, two The marking matrices and current transitions are calcu-
wheels, an H bridge, an Arduino MEGA 2560 board, jumper lated so that, when applying Equation 1, the future marking
cables, an embedded power source (7.2 V batteries), 13.56 matrix is found, which determines the next place where the
MHz RFID reader, and a 170 point breadboard. token has to go to, and, consequently, what action the robot
must perform. This way, the robot performs the movements
of turning left, turning right or moving on. When performing
a movement, if the robot does not find a new card, it moves
forward until it reaches one. When this happens, the cycle
restarts.

G. METHODOLOGY
The path to be followed is represented by the incidence
matrix of a Petri Net. The algorithm that contains this matrix
FIGURE 3. Robot used in the tests.
is loaded on the Arduino board beforehand. Then, the mobile
robot is positioned at the starting point of the aforementioned
path. The movement is performed according to the position
E. MAPPING AND LOCALIZATION OF ROBOTS of the robot in relation to the trajectory. The robot’s position
The robot must follow the map of a known environment, i.e., is set from reading the RFID cards. Thus, whenever the robot
be able to locate itself in that space and move from one point arrives at a card, a transition of Petri Net is enabled and the
to another, according to previous programming. place that received the token determines the action that must
VOLUME 4, 2016 7

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

FIGURE 5. Representation of the paths (A) 1, (B) 2, (C) 3 e (D) 4 traveled by

the robot.

FIGURE 4. Logical sequence that controls the actions of the mobile robot.

be performed by the robot. The paths traveled by the robot

during the tests are shown in Figure 5.
The maps to be followed can be represented in two ways.
In the first, mathematically, the map is represented by the
incidence matrix of a Petri Net. Thus, for each path there is a
specific matrix that needs to be loaded on the Arduino board
before the robot initiates the movement through that path.
Adapted incidence matrices are shown in Figure 6 and
are, respectively, the PNs shown in Figure 7. It is important
to note that due to reasons of algorithm implementation,
matrices A are the transpose of the incidence matrices.
The second way of representing the maps, graphically, is
shown in Figure 7. The places are represented by the circles
and connected by arcs (arrows) and transitions (squares),
indicating the path to be traveled by the robot. The token
(black dot) indicates the initial position of the robot. The
circles labeled as P1 are the starting points, whereas the
circles with the labels P4, P5 and P6 are the points of arrival.
Besides, the paths taken by the robot are represented by
dashed lines. FIGURE 6. Incidence matrices of the paths (A) 1, (B) 2, (C) 3 e (D) 4.
The robot is placed at the starting point and its objective is,
obeying the dynamics of the PN, to reach the point of arrival.
At each crossing (where two or more paths connect) there a unique ID. With this, the robot identifies its position and
is an RFID card for localization of the robot. Each card has performs one of the following movements: move forward,
8 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

Therefore, when arriving at a crossing, the robot stops,

reads the RFID card, locates itself (place Pi ), interacts with
the map, defines the action to be performed (from the trigger-
ing of the transition Ti ), and finally, moves to the place Pi+1 .
This cycle of actions is compiled and presented in Figure 8.

FIGURE 8. Cycle of actions executed by the proposed robot.

H. TESTS
The tests were performed in a labyrinth. Thus, at each in-
tersection labyrinth ways, there is an RFID card. When the
robot passes over the card, the future marking matrix of the
RP is updated, that is, the token is moved to the next place in
the RP. Thus, the algorithm determines which card the robot
should go to.
The robot was previously programmed and placed it at the
starting point P1. When the robot arrived at its destination, a
FIGURE 7. Petri Net that represents the paths (A) 1, (B) 2, (C) 3 e (D) 4. buzzer sounded. It is carried out the tests on the four trajec-
tories mentioned above. The four paths chosen have specific
characteristics to make the robot more robust, overcoming
turn left, turn right or stop. different situations.
It is important to emphasize that the PN determines what Path 1 is the most basic. All its curves are to the left, as
movement should be performed when the robot encounters shown in Figure 5 (a). Path 2 presents curves to the left and to
the card. However, to ensure that the robot can reach the cards the right, as shown in Figure 5 (b), whereas path 3, according
it is necessary that it has a line following module. to Figure 5 (c), also includes curves to the left and to the
Without this module, due to several factors such as mis- right. Nevertheless, its incidence matrix is not square, which
alignment of the wheels, the robot, which should ideally means that the number of transitions of the PN that defines
move forward, might for instance deviate from the trajectory this map is different from the number of places. At last, path
and not reach the next card. Therefore, it is used a module 4, as shown in Figure 5 (d), represents a peculiarity. The
composed of three optical sensors model TCRT5000. The robot passes through the place P4 twice and follows different
logic of the line following module is shown in Table 2. ways according to the path defined by the PN. The results and
considerations for the tests performed are presented bellow.
TABLE 2. Logic of the line following module.
IV. RESULTS
Left Central Right
Action
The results are compiled in Table 3. The movement control
Sensor Sensor Sensor of the robot is essential for the execution of the task of follow-
Turn
1 1 0 ing the map. The rows 1 and 2 of Table 3 present satisfactory
Right
1 0 0
Turn results for this action, since the robot can identify the black
Right line and perform the movements necessary to remain on the
Reading
Move
1 0 1 path without deviations.
Forward
0 1 1
Turn Along the route, more precisely at the crossing of tracks,
Left there are RFID cards to locate the robot regarding the map
Turn
1 1 1 of the environment. The use of RFID cards to determine the
Left
Note: A 0 represents the sensor over the black position of the robot generated partially satisfactory results,
tape and a 1 represents the sensor outside it as shown in the row 3 of Table 3. This method is efficient for
VOLUME 4, 2016 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

flat floors, given the proximity that the RFID reader, which is as shown in the row 5 of Table 3. RFID cards can be used
coupled to the robot’s bottom, must be to the card. to identify the position of a robot in relation to a trajectory,
which in turn, is modeled by Petri Nets. The tests were
TABLE 3. Compilation of results. performed in a structured environment composed of four
different paths. Thus, the robot traveled through the paths
Actions Question Answer correctly whenever its initial position was the place P1. The
The robot can
Drive and Is the robot able
stop, turn left, experimental results show that this approach is efficient and
synchronization to perform the feasible. As future work, Machine Learning and Computer
turn right and
of the motors intended movements?
move forward Vision Techniques will be applied to substitute RFID tags.
Reading and The sensors can
Does the robot
interpretation identify regions
recognize when it
of the line
is on the black line?
with black line REFERENCES
follower module and without it [1] J. A. Bessa, D. A. Barroso, A. R. da Rocha Neto, and A. R. de Alexandria,
The robot always “Global location of mobile robots using artificial neural networks in om-
performs the nidirectional images,” IEEE Latin America Transactions, vol. 13, no. 10,
Reading and Does the robot
movement required pp. 3405–3414, Oct. 2015.
identification identify and
by the card when [2] P. K. Mohanty, A. K. Sah, V. Kumar, and S. Kundu, “Application of deep
of RFID cards interpret the cards?
using the line q-learning for wheel mobile robot navigation,” in 2017 3rd International
following system Conference on Computational Intelligence and Networks (CINE). IEEE,
The robot follows Oct. 2017.
Is the robot
Mapping of the the map if it is [3] R. Siegwart and I. R. Nourbakhsh, Introduction to autonomous mobile
able to interpret
environment placed in the robots. Cambridge: The MIT Press, 2004.
the map represented
by using PN initial position [4] G. E. Marchant, B. Allenby, R. Arkin, E. T. Barrett, J. Borenstein, L. M.
by a PN?
of the path Gaudet, O. Kittrie, P. Lin, G. R. Lucas, R. O’Meara, and J. Silberman,
The robot is able “International governance of autonomous military robots,” The Columbia
to locate itself Science and Technology Law Review, vol. 12, pp. 272–276, 2011.
Has the overall
Integration using RFID and [5] P. W. Singer, “Military robots and the laws of war,” The New Atlantis,
project objective
of the actions efficiently follow no. 23, pp. 25–45, 2009.
been achieved?
the map modeled [6] G. Bekey and J. Yuh, “The status of robotics,” IEEE Robotics & Automa-
by PN tion Magazine, vol. 15, no. 1, pp. 80–86, Mar. 2008.
[7] J. Lim, S. Lee, G. Tewolde, and J. Kwon, “Indoor localization and navi-
It is important to highlight that the robot reads unregistered gation for a mobile robot equipped with rotating ultrasonic sensors using
a smartphone as the robot's brain,” in 2015 IEEE International Conference
cards that may appear in its trajectory, but it ignores them. on Electro/Information Technology (EIT). IEEE, May 2015.
Also, when there is no black line, the robot does not always [8] P. K. Binu, R. A. Krishnan, and A. P. Kumar, “An efficient indoor location
identify the card. This happens because in order to the tracking and navigation system using simple magnetic map matching,” in
2016 IEEE International Conference on Computational Intelligence and
reading be possible, the maximum distance to the reader has Computing Research (ICCIC). IEEE, Dec. 2016.
to be 3.5 cm, and they must be conveniently aligned. [9] T. F. Cordeiro, “Sistema de deteção e contorno de obstáculos para robótica
Following lines is a technique widely used for indoor móvel baseado em sensor kinect,” Engenharia Industrial, Escola Superior
de Tecnologia e Gestão do Instituto Politécnico de Bragança, Bragança,
testing. As Cordeiro [9] emphasizes, the navigation method 2014, 126 p.
known as Lines System is widely used in small robots in- [10] G. L. Santos, “Localização de robôs móveis autônomos utilizando fusão
tended for indoor localization. sensorial de odometria e visão monocular,” Engenharia de Computação,
Universidade Federal do Rio Grande do Norte, Natal, 2010, 66 p.
The trajectory to be traveled by the robot is modeled by [11] H. Zhang, C. Zhang, W. Yang, and C.-Y. Chen, “Localization and naviga-
Petri Nets. Thus, making an analogy between the PN and a tion using QR code for mobile robot in indoor environment,” in 2015 IEEE
physical system, it is assumed that the token is the robot, the International Conference on Robotics and Biomimetics (ROBIO). IEEE,
Dec. 2015.
place is where the registered card is located and the transition [12] A. Abdelgawad, “Localization system for indoor robot using RFID,” in
is the RFID card. 2014 IEEE Symposium on Industrial Electronics & Applications (ISIEA).
Therefore, the use of PN for the situation proposed, IEEE, Sep. 2014.
[13] X. Bao, S. Guo, N. Xiao, Y. Wang, M. Qin, Y. Zhao, C. Xu, and W. Peng,
presents partially satisfactory results, according to the row “Design and evaluation of a novel guidewire navigation robot,” in 2016
4 of Table 3, because for the robot to interpret and follow IEEE International Conference on Mechatronics and Automation. IEEE,
the map, it is necessary that its initial position is the place P1. Aug. 2016.
[14] A. R. Ismail, R. Desia, M. F. R. Zuhri, and R. M. Daniel, “Implementation
Respecting this condition, the robot was able to perform the of cognitive mapping algorithm for mobile robot navigation system,” in
planned route in an autonomous way and navigate through 2014 4th World Congress on Information and Communication Technolo-
the map in a satisfactory manner, since it arrived at the place gies (WICT 2014). IEEE, Dec. 2014.
[15] C. bae Moon and W. Chung, “Design of navigation behaviors and the
of destination. selection framework with generalized stochastic petri nets toward depend-
able navigation of a mobile robot,” in 2010 IEEE International Conference
V. CONCLUSION on Robotics and Automation. IEEE, May 2010.
[16] I. J. L. Batista, G. C. Barroso, O. de M. Almeida, A. T. Varela, and
Considering the general objective of this work, which is, J. R. B. Sousa, “Navegação de robôs móveis com ênfase em planejamento
given a map represented by the incidence matrix of a Petri e supervisão de trajetórias,” XVIII Congresso Brasileiro de Automática,
Net, to use RFID technology to know the position of the 2010.
[17] T. Karakurt, A. Durdu, and E. H. Dursun, “Petri-net based control of
robot in this map, as well as to use the PN dynamics as six legged spider robot,” in 2015 International Conference on Advanced
the cognition system of this robot, results are satisfactory, Robotics (ICAR). IEEE, July 2015.

10 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

[18] R. Mardiyanto, J. Anggoro, and F. Budiman, “2d map creator for robot Francisco A. X. da Mota Master in Telecommu-
navigation by utilizing kinect and rotary encoder,” in 2015 International nications Engineering in progress and Graduated
Seminar on Intelligent Technology and Its Applications (ISITIA). IEEE, in Mechatronics Engineering from the Federal In-
May 2015. stitute of Education, Science and Technology of
[19] P. D. Baldoni, Y. Yang, and S.-Y. Kim, “Development of efficient obstacle Ceará - Campus Fortaleza. Acted as technical-
avoidance for a mobile robot using fuzzy petri nets,” in 2016 IEEE 17th pedagogical coordinator at the Blue Genius Tech-
International Conference on Information Reuse and Integration (IRI). nology School. Participates in actions related to
IEEE, July 2016.
Educational Robotics promoted by the Mechanical
[20] F. Boniardi, A. Valada, W. Burgard, and G. D. Tipaldi, “Autonomous
Testing Laboratory (LEM) of the IFCE Campus
indoor robot navigation using a sketch interface for drawing maps and
routes,” in 2016 IEEE International Conference on Robotics and Automa- Fortaleza. Research Area: Mobile and Educational
tion (ICRA). IEEE, May 2016. Robotics.
[21] T.-S. Nguyen, T.-H. Tran, and H. Vu, “Accurate object localization using
RFID and microsoft kinect sensor,” in 2016 Eighth International Confer-
ence on Knowledge and Systems Engineering (KSE). IEEE, Oct. 2016.
[22] W. Yuan, Z. Li, and C.-Y. Su, “RGB-d sensor-based visual SLAM for
localization and navigation of indoor mobile robot,” in 2016 International
Conference on Advanced Robotics and Mechatronics (ICARM). IEEE,
Aug. 2016.
[23] S. R. Avutu, D. Bhatia, and B. V. Reddy, “Design of touch screen based
robot with obstacle detection module for autonomous path navigation,” in
TENCON 2017 - 2017 IEEE Region 10 Conference. IEEE, Nov. 2017. Matheus Xavier Rocha Undergraduate student
[24] F. Awad, A. Omar, M. Naserllah, A. Abu-Hantash, and A. Al-Taj, “Access at Federal Institute of Ceará - IFCE, Fortaleza
point localization using autonomous mobile robot,” in 2017 IEEE Jordan campus, course of Machatronics Engineering. Has
Conference on Applied Electrical Engineering and Computing Technolo- a technical degree in Automotive Maintenance
gies (AEECT). IEEE, Oct. 2017. (IFCE, Campus Fortaleza). Scientific research
[25] T. Dewi, P. Risma, Y. Oktarina, and M. T. Roseno, “Neural network scholarship at Laboratory of Mechanical Testing
controller design for a mobile robot navigation: A case study,” in 2017 (LEM) by CNPq, IFCE.
4th International Conference on Electrical Engineering, Computer Science
and Informatics (EECSI). IEEE, Sep. 2017.
[26] F. Furlan, E. Rubio, H. Sossa, and V. Ponce, “Humanoid robot hierarchical
navigation using petri nets and fuzzy logic,” in 2017 56th Annual Confer-
ence of the Society of Instrument and Control Engineers of Japan (SICE).
IEEE, Sep. 2017.
[27] A. Motroni, A. Buffi, and P. Nepa, “Localization of a mobile device
equipped with an RFID reader,” in 2017 IEEE International Conference
on RFID Technology & Application (RFID-TA). IEEE, Sep. 2017.
[28] C. Wang, L. Meng, S. She, I. M. Mitchell, T. Li, F. Tung, W. Wan, M. Q.-
H. Meng, and C. W. de Silva, “Autonomous mobile robot navigation in
uneven and unstructured indoor environments,” in 2017 IEEE/RSJ Inter-
national Conference on Intelligent Robots and Systems (IROS). IEEE, Joel J. P. C. Rodrigues [S’01, M’06, SM’06]
Sep. 2017. is a Professor and a Senior Researcher with the
[29] S. Xu and W. Chou, “An improved indoor localization method for mobile National Institute of Telecommunications (Inatel),
robot based on WiFi fingerprint and AMCL,” in 2017 10th International Brazil, and a Senior Researcher with the Instituto
Symposium on Computational Intelligence and Design (ISCID). IEEE, de Telecomunicações, Portugal. He has been a
Dec. 2017. Professor with the University of Beira Interior,
[30] B. Xu, X. Zhou, T. Cheng, Z. Su, and J. Wu, “A new proposal for Portugal, and a Visiting Professor with the Uni-
localization of omni-directional mobile robot by DM tag in indoor en- versity of Fortaleza, Brazil. He is also the Leader
vironment,” in 2017 IEEE International Conference on Cybernetics and of the Internet of Things Research Group (CNPq),
Intelligent Systems (CIS) and IEEE Conference on Robotics, Automation the Director for Conference Development – IEEE
and Mechatronics (RAM). IEEE, Nov. 2017. ComSoc Board of Governors, an IEEE Distinguished Lecturer, the President
[31] S. Thrun, “Robotic mapping: A survey,” School of Computer Science, Feb. of the Scientific Council at ParkUrbis – Covilhã Science and Technol-
2002. ogy Park, the past Chair of the IEEE ComSoc Technical Committee on
[32] Y. Xiao, Y. Ou, and W. Feng, “Localization of indoor robot based on
eHealth, the past Chair of the IEEE ComSoc Technical Committee on
particle filter with EKF proposal distribution,” in 2017 IEEE International
Communications Software, a Steering Committee Member of the IEEE
Conference on Cybernetics and Intelligent Systems (CIS) and IEEE Con-
ference on Robotics, Automation and Mechatronics (RAM). IEEE, Nov. Life Sciences Technical Community and the Publications Co-Chair, and
2017. a Member Representative of the IEEE Communications Society on the
[33] Z.-B. Song, Y. H. Zweiri, L. D. Seneviratne, and K. Althoefer, “Non- IEEE Biometrics Council. He is the Editor-in-Chief of three international
linear observer for slip estimation of tracked vehicles,” Proceedings of journals and an Editorial Board Member of several high-reputed journals.
the Institution of Mechanical Engineers, Part D: Journal of Automobile He has been general chair and TPC Chair of many international conferences,
Engineering, vol. 222, no. 4, pp. 515–533, Apr. 2008. including IEEE ICC, IEEE GLOBECOM, and IEEE HEALTHCOM. He is a
[34] J. Lope, D. Maravall, and J. G. Zato, “Topological modeling with fuzzy member of many international TPCs and participated in several international
petri nets for autonomous mobile robots,” Lecture Notes in Artificial conferences organization. He has authored or coauthored over 600 papers in
Inteligence, pp. 290–299, 1998. refereed international journals and conferences, 3 books, and 2 patents. He
[35] I. J. L. Batista, “Modelo de navegação para robôs móveis baseado em redes had been awarded several Outstanding Leadership and Outstanding Service
de petri coloridas,” Engenharia de Teleinformática, Universidade Federal Awards by IEEE Communications Society and several best papers awards.
do Ceará, Fortaleza, 2008, 87 p. Prof. Rodrigues is a licensed professional engineer (as senior member),
[36] A. Alexopoulos, L. Zouaghi, and E. Badreddin, “Associative memory for member of the Internet Society, and a senior member ACM and IEEE.
modified petri-net based monitoring of mobile robot navigation,” in 2012
First International Conference on Innovative Engineering Systems. IEEE,
Dec. 2012.

VOLUME 4, 2016 11

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2846554, IEEE Access

Mota et al.: IEEE Access

Victor Hugo C. de Albuquerque has a PhD in

Mechanical Engineering with emphasis on Materi-
als from the Federal University of Paraíba (UFPB,
2010), an MSc in Teleinformatics Engineering
from the Federal University of Ceará (UFC, 2007),
and he graduated in Mechatronics Technology at
Instituto Federal de Educação, Ciência e Tecnolo-
gia do Ceará - IFCE, Campus Fortaleza (IFCE,
2006). He is currently Assistant VI Professor of
the Graduate Program in Applied Informatics, and
coordinator of the Laboratory of Bioinformatics at the University of Fort-
aleza (UNIFOR). He has experience in Computer Systems, mainly in the
research fields of: Applied Computing, Intelligent Systems, Visualization
and Interaction, with specific interest in Pattern Recognition, Artificial
Intelligence, Image Processing and Analysis, as well as Automation with
respect to biological signal/image processing, image segmentation, biomedi-
cal circuits and human/brain-machine interaction, including Augmented and
Virtual Reality Simulation Modeling for animals and humans. Additionally,
he has research at the microstructural characterization field through the
combination of non-destructive techniques with signal/image processing and
analysis, and pattern recognition. Prof. Victor is the leader of Computational
Methods in Bioinformatics Research Group. He is editorial board member of
the IEEE Access, Computational Intelligence and Neuroscience, Journal of
Nanomedicine and Nanotechnology Research, and Journal of Mechatronics
Engineering, and he has been Lead Guest Editor of several high-reputed
journals, and TPC member of many international conferences. He has
authored or coauthored over 160 papers in refereed international journals,
conferences, 4 book chapters, and 4 patents.

Auzuir Ripardo de Alexandria received his de-

gree in Electrical Engineering (1993) and a Bach-
elor degree in Computer Science (1994) from
the Universidade Federal de Campina Grande –
UFCG, and master (2005) and doctor (2011) de-
grees in Teleinformatics Engineering from the
Universidade Federal do Ceará – UFC. He worked
in local companies between 1993 and 2005, in the
industrial and hardware and software development
areas. He is a professor at the Instituto Federal
de Educação, Ciência e Tecnologia do Ceará - IFCE, Campus Fortaleza,
Department of Industry, since 2003. He participated in the conception of the
Mechatronics Engineering course where he acted as coordinator. Participates
as permanent professor in the Graduate Program in Telecommunications
Engineering and in the Graduate Program in Renewable Energy. His topics
of research are: Computational Vision, Mobile Robotics, Biomedical Engi-
neering, Artificial Neural Networks and Industrial Automation; coordinating
and guiding many projects. He is leader of the Computational Simulation and
Robotics research groups of IFCE.

12 VOLUME 4, 2016

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.