Content uploaded by Leo Ordinez
Author content
All content in this area was uploaded by Leo Ordinez on Nov 14, 2014
Content may be subject to copyright.
Design Patterns for Cyber-Physical Systems: The
Case of a Robotic Greenhouse
Ricardo Garro
EEA Anguil “Ing. Agr. Guillermo Covas”
INTA
Anguil, Argentina
rgarro@anguil.inta.gov.ar
Leo Ordinez, Omar Alimenti
Instituto de Investigaciones en Ingeniería Eléctrica
DIEC – UNS – CONICET
Bahía Blanca, Argentina
lordinez@uns.edu.ar, iealimen@criba.edu.ar
Abstract— Cyber-physical systems (CPS) are a new discipline
of research that involves electrical engineering, electronics,
computer science, control and communications interacting
with physical processes. This leads to a co-managed domain
where both worlds (cyber and physical) must be taken into
account to decide the actions to take. In this sense, this paper
presents the design of a robotic greenhouse, which involves
basically the greenhouse and a mobile robot interacting with
the physical environment (collecting data and acting on
plants). Thus, the main contribution is related to taking
practical experience and generalize it by applying well-proven
techniques to the development of CPS.
Keywords-cyber-physical system; design patterns; robotic
greenhouse.
I. INTRODUCTION
At the present, production systems under cover or in
greenhouses are a clear example of intensive production
where high levels of performance are achieved. In Argentina,
the area devoted to horticultural production in greenhouses is
approximately 5000 ha. It is located mostly in vegetable belts
of large cities and provinces such as Corrientes, Santa Fe,
Buenos Aires, La Plata, Mar del Plata, among others.
The data show that the crops under cover or in
greenhouses, belong to a highly productive system within the
agricultural sector in Argentina. One of the main productive
advantages of this type of installations is that it is possible to
generate the optimal climatic conditions that allow off-
season crop. This creates a better selling price because the
demand is high and the supply limited. However, there are
some problems such as the control of pests and diseases,
which have a higher incidence than in crops grown outdoors.
In the greenhouse crops this control is usually realized
through the application of chemicals. In recent times
methods environmentally friendly and less harmful to people
have been introduced. However, it still remains the challenge
of finding optimized treatment, from a technical and
agricultural point of view [18].
Based on the foregoing, it is a suitable alternative to use a
mobile robot to develop these tedious and unhealthy tasks.
On the other hand, despite the repetitiveness of the tasks to
be performed and the controlled environment in which they
will be developed, building a robot with these characteristics
represents a current technological and scientific challenge.
Moreover, the strong interaction with the physical world and
the problems it involves requires a multidisciplinary
approach. Consequently, control theory, real-time embedded
systems, software engineering, mechanics and agronomic by
themselves are not enough.
Cyber-Physical Systems (CPS) are an appropriate
solution to this need by proposing an integration of physical
processes with computing. This paradigm is very recent [19]
and is based on others such as control theory, real-time
systems and communications networks. However, CPS
introduced a different look at the system to be developed.
According to [10], CPS are systems in which information
processing and physical processes are so closely integrated
that it is not possible to determine whether behavioral
attributes are the result of computations, physical laws, or
both. On the other hand, [13] state that a CPS is a system
composed of physical subsystems together with computing
and networking. These three components must necessarily be
included in the modeling and system design.
In this context, the objective of this paper is to present the
design a Robotic Greenhouse from the point of view of CPS.
This includes taking account of the characteristics of this
kind of systems to model the case study. To do that, firstly
the CPS under construction is described in terms of
environmental variables. Then, based on those variables, a
general design of the main subsystems involved is proposed.
In order to ease the approach, a series of Design Patterns are
identified to be adapted in order to fulfill the requirements of
the Robotic Greenhouse. Additionally, through a division of
the CPS in subsystems, new Design Patterns are found and
roughly described. With all, the article intends to help the
reader in the development of a CPS with some experiences
taken from the practice.
The main entities involved in the system are the
environmental and phytosanitary conditions of the
greenhouse, and a mobile robot capable of collecting
environmental data and perform the application of chemicals
to plants. In addition, as part of the system it is included a
remote server capable of holding information in a database
and to command the robot manually. From the entities
involved it can be seen that the development can be projected
2011 Brazilian Symposium on Computing System Engineering
978-0-7695-4641-4/11 $26.00 © 2011 IEEE
DOI 10.1109/SBESC.2011.10
15
2011 Brazilian Symposium on Computing System Engineering
978-0-7695-4641-4/11 $26.00 © 2011 IEEE
DOI 10.1109/SBESC.2011.10
15
2011 Brazilian Symposium on Computing System Engineering
978-0-7695-4641-4/11 $26.00 © 2011 IEEE
DOI 10.1109/SBESC.2011.10
15
2011 Brazilian Symposium on Computing System Engineering
978-0-7695-4641-4/11 $26.00 © 2011 IEEE
DOI 10.1109/SBESC.2011.10
15
from the viewpoint of a CPS. This approach includes the
problem statement from the joint perspective of the physical
world involved and computational resources required, along
with a description of those necessary subsystems, arising
from the problem statement and its corresponding
requirements analysis. It must be stressed that unlike an
embedded system, which focuses mainly on the electronics
and computer system to develop, the CPS approach
considers the physical world as a co-domain affected by the
physical laws and computational processes indistinguishable.
From the latter point of view, the Robotic Greenhouse is not
a greenhouse with a robot, but a more complex system,
conformed by the mobile platform, the greenhouse
microclimate, plants and their physical distribution, the
remote server, the mobile control systems, etc.
After this introduction, the rest of the paper is organized
as follows: Section II presents and describes the physical
environment of the greenhouse, its characteristics and
challenges. In Section III, the requirements of the proposed
system are analyzed. The system design is presented in
Section IV. Section V presents some design patterns from
the literature that are suitable for the Robotic Greenhouse
(and for CPS, in general). In addition, that section sketches
some Design Patterns needed but not found in the literature.
In Section VI, some related work that supported the
development of this article is presented. Finally, in Section
VII, conclusions are exposed.
II. C
ASE STUDY
A greenhouse is typically a structure whose roof and
sides are transparent or translucent, allowing a sufficient
quality and quantity of sunlight to produce photosynthesis in
plants contained therein. The greenhouses have been
generally designed to protect plants that are harvested out of
season or not, protecting them from excessive heat or cold
[1]. There are also, in this environment, some aspects
unfavorable and harmful to human health, either by the
conditions of the place or due to the tasks to be performed.
Equipment that have traditionally been used to combat
pests and diseases are spears and hand spray guns (low-cost
equipment, easy maintenance and suitable for specific
phytosanitary problems). Although they can be coupled to a
distributed network of gas pipes, several problems arise such
as low efficiency, difficulty in regulation, operator skill, high
risks of environmental pollution, loss of plant protection
products in soil; health risk for people, etc.
While there are others more technically advanced
equipments (guns, spray facilities), an alternative to the
phytosanitary tasks optimization is the application of
robotics. Fundamentally the presence of operators represents
a number of hazardous tasks, repetitive and somewhat
tedious, that can be robotized. Hence, availability of vehicles
with autonomous movement capabilities would be
convenient [17] [4] [8].
Today, in some parts of Argentina, activities such as field
survey, data acquisition and environmental management are
made in the traditional way. As previously mentioned
backpacks can be used to spray pesticides, measure variables
(i.e., humidity, temperature, Normalized Differential
Vegetation Index (NDVI), etc.) and hand picking. All these
tools generate an effort and some considerable risk. In these
tasks, usually routine or potentially dangerous to humans, the
idea of using an autonomous mobile vehicle is a suitable
alternative. This enables the relocation of people away from
potentially dangerous environments.
In this context, we present the design of a mobile
spraying robot for intensive production under cover. The
robot has a movable structure, with a differential mode of
locomotion, which allows selective application of plant
protection product, according to the needs of the crop.
Importantly, the mobile platform does not work in
isolation, but forms a network with different devices located
at strategic points in the greenhouse, constituting a network
of wireless sensors and actuators (WSANs). This idea is
based on the work proposed in [20]. Particularly, this kind of
technology allows the correlation between various WSANs,
which will translate into a better understanding of the
environmental conditions. Therefore, making possible the
generation of specific applications for intensive cultivation in
the field of precision agriculture [16]. One application is that
where local conditions may dictate the control of the
amounts of pesticides and fertilizers.
The possibility that a greenhouse can vary in their
internal configuration in terms of number of brokers, facings,
lines of crops, etc., makes navigation not an easy task. Thus,
in principle, it is addressed the travelling with previously
established maps, including the possibility of a manual
remote control. However, the design leaves the possibility of
including in a future autonomous navigation using a more
complex system. One of the future goals is to use the Zigbee
modules to triangulate the mobile robot's position by
measuring the intensity of the wireless signal [11].
III. R
EQUIREMENTS ANALYSIS
From the context explained in the previous section, it
comes up the need to address the design of the Robotic
Greenhouse from the viewpoint of CPS. As mentioned
earlier, these systems represent a very strong integration of
the physical processes with computing [12] [10]. Such
integration is obvious in this case, because the physical
variables that must be handled are closely related to the
software and hardware components that affect them. To this
point there is such coexistence between physical and
computational processes, that for an external observer it is
difficult to determine if environmental conditions of the
greenhouse and plants are given by natural causes or by the
actions of the computer system. In other words, it is
impossible to distinguish the environmental process from the
computational one, both of which make up the Robotic
Greenhouse. This makes it difficult to determine where one
ends and the other begins.
Requirements analysis is about understanding the
problem [6]. For the particular case of CPS, requirements
analysis is given mainly by the establishing of the physical
processes involved, the identification of their variables and
the setting up of the relationships among those variables, as
well as their interactions with the computational aspects
(hardware and software associated). This section identifies
16161616
the variables to be taken into account in the CPS, both for the
specific monitoring of the greenhouse, and for the mobile
platform.
A. Environmental Variables of the Greenhouse
The main variables to consider are: humidity,
temperature and solar radiation. These three variables are
essential to the specialized staff that manages the
greenhouse, since with them inferences about behavior and
evolution of different crops can be made.
B. Variables of the Spraying Robot
In the case of the other outstanding component of the
CPS (i.e., the robot), the set of variables is constituted by
position, speed, acceleration, distance to objects, electric
power, pressure, fluid flow rate and Zigbee signal strength.
This set of variables is the one that will enable the mobile
robot to travel across the greenhouse in a controlled way and
to perform its tasks (both the ones of spraying plant
protection product and those typical of navigation within the
greenhouse).
C. Requirements Synthesis
Based on the previous variables and the functional and
nonfunctional requirements common in CPS [13], the
following issues have to be considered in the design stage:
motion; fixed sensing and positioning; Human-Machine
remote interface; communication; spraying; vision and
imaging and general coordination.
IV. S
YSTEM DESIGN
In order to represent the architectural design of the
system, Deployment Diagrams of the Unified Modeling
Language (UML) [2] were chosen. Figure 2. shows how the
Robotic Greenhouse CPS is composed. This diagram allows
having a wide view of the system and its subsystems, as well
as the relationships between its components at the hardware
and software level. On the other hand, Figure 1. shows the
platform where the experiments are performed.
Figure 1. Image of the spraying robot.
Figure 2. Deployment diagram of the CPS.
According to the requirements synthesized in the
previous section, next the different subsystems that constitute
the Robotic Greenhouse are described.
A. Motion Subsystem
This subsystem (Figure 3.) is in charge of acquiring
relative and absolute position variables through the reading
of several sensors. Besides, it drives the motors of the robot.
One of the most important tasks to be carried out by this
subsystem is to sample, with the frequency set by the real-
time scheduler, the different sensors. These data are sent to
the central coordinator which performs the calculations of
trajectory. With the returned data, control and trajectory
tracking algorithms are executed. The path tracking is
achieved with a control signal acting on the H-bridge,
indicating the direction and intensity of rotation of each
motor individually.
Figure 3. Deployment diagram of the motion subsystem.
Finally, this subsystem is responsible for making the
triangulation and definition of the relative position according
to the measurement of Zigbee signal strength with respect to
different fixed positioning modules distributed within the
greenhouse.
B. Fixed Sensing and Positioning Subsystem
This module (see Figure 4.) is responsible for measuring
humidity, temperature and solar radiation in certain strategic
17171717
points within the greenhouse. The collected information is
then sent to a remote server by means of a Zigbee module.
Figure 4. Deployment diagram of the fixed subsystem.
As previously mentioned, the mobile platform also has a
Zigbee module, so it can triangulate its relative position with
respect to these fixed modules, located in the greenhouse.
C. Human-Machine Remote Subsystem
This subsystem, composed mainly of a server and a
Zigbee communication module, is responsible for enabling
the robot to be manually operated. In this case, the
perception of the environment and path planning is left to the
user. However, the robot has mechanisms to alert for any
possible inconvenience of operation. In turn, this subsystem
processes the information sent by the robot, which will give
end users a comprehensive and quantified look of the
environmental variables of the greenhouse and the mobile
platform.
D. Robot Communication Subsystem
This subsystem performs the necessary actions to allow
communication between the mobile robot and the remote
server as well as the fixed subsystems. Thanks to this
subsystem data from the robot can be collected and guidance
within the greenhouse can be done remotely. This subsystem
consists mainly in a Zigbee transceiver.
E. Spraying Subsystem
Figure 5. Deployment diagram of the spraying subsystem.
It is responsible for carrying out the spraying process.
Here, it is important to control pressure and fluid flow rate
variables, since abrupt changes in their values may cause
breakage of equipment. Another variable to consider is the
temperature, because this directly influences the
evapotranspiration of the product and should be taken into
account when making any application of it. Finally, this
subsystem is also responsible for activating and deactivating
the valves that actually make the application of the products,
according to the spatial configuration of the greenhouse (i.e.,
number of aisles, facings, etc.). The deployment diagram of
this subsystem is shown in Figure 5.
F. Vision and Imaging Subsystem
This subsystem has the objective of capturing and
processing images of the environment with the aim of
establishing the characteristics of the plants (color, size,
NDVI, etc.).
In the future, this subsystem will be also used to aid the
motion subsystem and the fixed positioning one in path
planning.
G. Coordinator Subsystem
This subsystem (Figure 6.) is the heart of the Robotic
Greenhouse. It is responsible for coordinating all activities
through the real-time scheduler, which gives different
priorities to each task according to their criticality and how it
may impact on the overall system an eventual failure.
Figure 6. Deployment diagram of the coordinator subsystem.
All measurements obtained and processed by other
subsystems are sent to this one. Here, calculations of
position, obstacle avoidance, motion control, spraying,
among others, are performed. Once information is processed,
new information is returned to the subsystem that generated
the response, so that the necessary control actions are taken.
Also, this subsystem is responsible for sending the collected
data (through the communication subsystem of the robot) to
the central server to be processed, displayed and stored in a
database.
V. D
ESIGN PATTERNS
Once the structure of the Robotic Greenhouse is
presented from the viewpoint of CPS, in this section, a series
of design patterns will be identified and applied to that
design. To this aim, different alternatives are explored in
order to fulfill the needs of a CPS. Additionally, in those
cases that, to the authors’ belief, there are no design patterns
developed, they will be determined and roughly sketched.
Buschmann et al. [5] describe an architectural software
pattern as a particular and recurrent design problem, that is
stated in specific contexts, and that is suitable of having a
generic plan for solving it. Therefore, the solution scheme is
specified through the description of the main elements, the
responsibilities, relationships and the way in which the
entities involved cooperate.
18181818
It is noteworthy that there are not only design patterns for
software, but as explained in [3], design patterns have
become an important area of research in many fields. Some
of them are fault tolerance, telecommunications, embedded
systems, security, among others. Each of these areas has its
own patterns, but all follow the same basic principle stated
above.
In the first place, some well-known design patterns will
be revisited.
Konrad and Cheng [9] make a deep review of design
patterns for embedded systems. Following, the ones that best
fits the requirements of the Robotic Greenhouse, in
particular, and of any CPS, in general, are enumerated.
Controller Decompose: It deals with decomposing an
embedded system into different components based on their
responsibilities. This decomposition is essential since there
are many activities involved and responsibilities must be
shared correctly.
Actuator-Sensor Pattern: This pattern is useful for
specifying several kinds of sensors and actuators in an
embedded system. In the Robotic Greenhouse it becomes
necessary because of the amount of sensors and actuators
present.
Watchdog Pattern: The objective of this pattern is to
monitor a device and initiate a corrective action if it does not
behave properly.
Monitor-Actuator Pattern: This pattern has to do with
increase of safety and reliability by monitoring the actuator
behavior for errors.
The use of the following patterns is important because of
the nature of the system under design in which a malfunction
can cause considerable human and material damage.
Fault Handler Pattern: It deals with the integration of
fault handling functionality into an embedded system. Given
the nature of the presented CPS, it is important to have an
error handling mechanism integrated into the system.
Mask Pattern: This pattern is used to reduce the burden
placed on the computing component when many sensors and
actuators are present, and their values need to be sorted or
filtered into single values for the computing component. In
the proposed design it is important to distribute the
processing load among the processors since there are many
sensors and actuators related to various processing units.
Channel Pattern: It is useful to organize communication
between two components. This pattern is present in several
subsystems due to communication is inherent in CPS and
especially in the robotic greenhouse.
From the book of Pont [15] some patterns are also
extracted as follows:
Hardware Patterns: They deal with characteristics such
as microcontroller, oscillator, memory issues and DC and
AC loads.
Port Access Patterns: They are useful for managing
interfaces with different kinds of devices directly from the
microcontrollers’ ports.
User Interface Patterns: These patterns deal with those
devices common in embedded designs that act as human-
machine interfaces. For example, LCDs, keypad, switches.
Serial Peripherals Patterns: They were designed to
attend those situations where there is a need to access
devices through serial Communications channels (e.g., I2C,
SPI, RS-232, RS-485).
Mono and Multiprocessor Patterns: They propose
solutions to situations according to different structural
designs depending on whether a single processor is enough
or a multiprocessor scheme is better.
Monitoring and Control Patterns: These are very
simplistic but represent a good starting point. They provide
different alternatives related to aspects such as Pulse Width
Modulation (PWM), analogue-to-digital converters y
Proportion-Integral-Derivative control.
The key benefit of design patterns usage in CPS is to take
advantage of well-tested solutions. Solutions that, in specific
contexts of design, address recurring problems that may
arise. However, as mentioned above, there are aspects
particular to CPS that are not covered in the literature.
Therefore, here are some possible design patterns needed in
the design of a CPS. They are presented as merely
illustrative and schematic just to identify points to take into
account when building a CPS based on this approach.
Patterns for handling different power requirements:
The objective of these patterns is to solve the power handling
for differential traction motors, using two H-bridges and
controlled by PWM. It is also useful for the spray pump of
the robot.
Patterns of interconnection of microcontrollers: The
purpose of these patterns is to provide solutions to the
problems presented in the application design in terms of
multiple processors required to carry out the development.
Patterns of real-time scheduling: The goal of these
patterns is to provide a solution to the need to schedule tasks
with different timing constraints. This implies single and
multiprocessor schemes.
Patterns of control applied to mobile robotics: These
patterns are intended to present solutions to the problems of
digital control, from the approach of state variables. These
will be applied primarily to the control of trajectories. Its
main application is in the Motion Subsystem.
Patterns for image capture and processing: The
purpose of these patterns is to provide a solution to capture
and image processing inside the greenhouse. Its application
is in the Vision and Imaging Subsystem.
Patterns for data fusion of multiple sensors: These
patterns aim to address how to integrate information from
different sensors (sonar module, Zigbee RF, images,
encoders, gyroscope, GPS, etc.). They are applied in the
Coordinator Subsystem.
Design patterns for real-time communications: The
purpose of these patterns is to respond to the problems
presented in the transmission of data from the robot or fixed
station to the remote server. They are applied in the H-M
Remote Subsystem.
VI. R
ELATED WORK
Progressively they have been incorporated into the
agribusiness sector new information technologies,
communication and control techniques. Mainly those related
19191919
to automation and robotics. Some robotic applications
developed for the agribusiness sector, particularly for
greenhouses, are as follows.
One example is a mobile robot called AURORA [14].
The AURORA project is designed to replace those jobs that
involve health risks inside greenhouses by means of an
autonomous mobile robot. The robot has various devices for
inspection and operation. Another example is that of the
University of Almería [8] [17], where it has been developed
an autonomous mobile platform called Fitorobot. In this
case, the platform can move between the rows of crops to
perform, in an early stage, phytosanitary applications by
using a vertical spray bar. Similarly, research works were
developed at the MIT [7]. There they made the
implementation of a distributed system of autonomous
gardening, where the garden was established as a network
between the robots and plants. Robots "gardeners" are
mobile and have a robotic arm, which holds a camera. They
are able to locate plants and water them as needed.
It should be noted that the works relating to the generic
application of design patterns to CPS have been analyzed in
Section V of this article.
VII. C
ONCLUSIONS
Currently, the integrated knowledge is more than the set
of specialists or disciplines working in teams on specific
issues. Value added in the incorporation of technologies to
improve agricultural production is to optimize resource
management and data collection. It also improves efficiency
and protects human beings from relocating their workplaces
and substitution of tasks that can be risky for their health. A
viable alternative would be to address the problem from the
conceptual idea provided by the CPS. In particular, this
paper attempts to contribute in this regard.
From the viewpoint of the CPS, in a given system there is
no distinction between physical, biological and structural
parts from those electronics, mechanics and computational
ones. The integration is such that the physical environment
becomes a co-domain affected by the physics laws and
computational processes.
This paper specifically presents the design of a Robotic
Greenhouse, from the CPS approach. In this sense, it
considers the whole set ecosystem-greenhouse and a mobile
robot as a single system. Based on this the main system
requirements were identified in terms of the physical
variables involved. Then, from these variables, subsystems
necessary to meet requirements were established. Finally,
these subsystems were characterized and described using
UML Deployment Diagrams.
Additionally, Design Patterns were used to describe the
solutions to various problems present. Thus, classic patterns
were highlighted and new ones were identified, according to
the particularities of each case.
R
EFERENCES
[1] Encyclopedia Britannica. (2011) “Enciclopedia Británica”,
http://www.britannica.com/EBchecked/topic/9620/agricultural-
technology/67808/Greenhouses, June 2011.
[2] UML. (2011) “Unified Modeling Language”, http://www.uml.org,
June 2011.
[3] Ashraf Armoush, Falk Salewski, Stefan Kowalewski. (2008)
“Effective Pattern Representation for Safety Critical Embedded
Systems”. CSSE (4): 91-97
[4] Baturone, A. O. (2001). Robótica: manipuladores y robots móviles,
ISBN: 8426713130. Páginas: 464. Marcombo.
[5] Buschmann, Frank, Regine Meunier, Hans Rohnert, Peter
Sommerlad, Michael Stal, and Michael Stal. (1996). “Pattern-
Oriented Software Architecture Volume 1: A System of Patterns”. 1o
ed. Wiley, Agosto 8.
[6] Cheng, B. H.C, and J. M Atlee. (2007). “Research directions in
requirements engineering.” In 2007 Future of Software Engineering,
285–303.
[7] Correll, N., N. Arechiga, A. Bolger, M. Bollini, B. Charrow, A.
Clayton, F. Dominguez, et al. (2009). “Building a distributed robot
garden.”, In Intelligent Robots and Systems, 2009. IROS 2009.
IEEE/RSJ International Conference on, 1509–1516.
[8] González, R., F. Rodríguez, J. Sánchez-Hermosilla, and J. G.
Donaire. (2007) “Experiencias en sistemas de navegación de robots
móviles para tareas en invernadero”. AgroIngenieria, Septiembre
2007, Albacete, Spain.
[9] Konrad, S., and B. H.C Cheng. (2005) “Requirements patterns for
embedded systems.” In Requirements Engineering, 2002.
Proceedings. IEEE Joint International Conference on, 127–136.
[10] Krogh, B. H., E. Lee, I. Lee, A. Mok, R. Rajkumar, L. R. Sha, A. S.
Vincentelli, et al. (2008). “Cyber-Physical Systems, Executive
Summary.” CPS Steering Group, Washington DC, March.
[11] Kuo, W. H, Y. S Chen, G. T Jen, and T. W Lu. (2010) “An intelligent
positioning approach: RSSI-based indoor and outdoor localization
scheme in Zigbee networks.” In Machine Learning and Cybernetics
(ICMLC), 2010 International Conference on, 6:2754–2759.
[12] Lee, E. A. (2006). “Cyber-physical systems-are computing
foundations adequate.” In Position Paper for NSF Workshop On
Cyber-Physical Systems: Research Motivation, Techniques and
Roadmap, 1:1–9.
[13] Lee, E. A, and S. A Seshia. (2010). “Introduction to Embedded
Systems-A Cyber-Physical Systems Approach.” Lee & Seshia.
[14] Mandow, A., J. L. Martinez, V. F. Munoz, A. Ollero, and A. Garcia-
Cerezo. (2002). “The autonomous mobile robot AURORA for
greenhouse operation”. Robotics & Automation Magazine, IEEE 3
(4): 18–28.
[15] Pont, M. J. (2001). “Patterns for time-triggered embedded systems:
Building reliable applications with the 8051 family of
microcontrollers.” ACM Press/Addison-Wesley Publishing Co.
[16] Rodolfo Bongiovanni, Evandro Chartuni Mantovani, Stanley Best,
and Álvaro Roel. (2006). “Agricultura de precisión: Integrando
conocimientos para una agricultura moderna y sustentable.”
PROCISUR. Montevideo, UY.
[17] Sánchez, R. G, F. R Díaz, J. S.H López, and J. G Donaire. (2006)
“Algoritmo de navegación reactiva de robots móviles para tareas bajo
invernadero”. In XXVII Jornadas de Automática.
[18] Sánches-Hermosilla, J., A. S.G López, and Y. R.M Anzano. (2007).
“Equipos de aplicación de productos fitosanitarios en invernadero”.
Horticultura global: 26–31.
[19] Sha, L., S. Gopalakrishnan, X. Liu, and Q. Wang. (2009). “Cyber-
physical systems: A new frontier.” Machine Learning in Cyber Trust:
3–13.
[20] Stankovic, J. A. (2008). “When sensor and actuator networks cover
the world”. ETRI journal 30 (5): 627–633.
20202020
