AGRIBOT: DEVELOPMENT OF A MOBILE ROBOTIC PLATFORM TO

SUPPORT AGRICULTURAL DATA COLLECTION

R.A. Tabile and R.V. Sousa

Biosystem Engineering Department

School of Animal Science and Food Engineering - University of São Paulo
Pirassununga, São Paulo, Brazil

A.J.V. Porto
Mechanical Engineering Department
Engineering School of São Carlos - University of São Paulo
São Carlos, São Paulo, Brazil

R.Y. Inamasu
Embrapa Instrumentation
Brazilian Agricultural Instrumentation Research Corporation
São Carlos, São Paulo, Brazil

ABSTRACT

Precision Agriculture and agricultural practices that take into account

environment protection, leads to several research challenges. Sampling scale and
the precision required by these new agricultural practices are often greater than
those required by traditional agriculture, raising the costs of production. This
whole process requests an expressive number of studies in developing automation
instruments. Amongst them, the use of remote sensing techniques based on On-
the-Go sensors technology stands out, coupled to a Geographic Information
System (GIS) adapted and developed for agricultural use. Therefore, the
application of Agricultural Mobile Robots is a strong tendency, mainly in the
European Union, USA and Japan. In Brazil, studies are necessary for the
development of robotics platforms, serving as basis for semi-autonomous and
autonomous navigation systems, facilitating data acquisition in the field. The
greatest difference in the agricultural practices between Brazil and other countries
is that, at other countries the skilled labor is not an issue and often the farm owner
or family members perform field operations. Consequently, tasks automation can
provide more comfort and reduce working days. Moreover, as it is considered a
strategic sector, the government provides subsidies to producers in order to ensure
at least part of internal consumption. The access to technology is also a
differential characteristic, due to its high price or even unavailability in Brazil.
Thereby, in Brazil, autonomous systems are supposed to meet the needs of the
scarceness of qualified professionals, in face of the rising demands; in addition to
serve as a laboratory for the development of national technology. The aim of this
study is to describe the project of a mobile robotic platform designed to be used
for the development of control systems, navigation and data acquisition
technologies for agriculture. The main application of the platform is to perform
remote sensing of agronomic parameters in large areas, at the most important
Brazilian crops. The platform does not require actions with high power, as in
traditional agricultural operations, but has to move efficiently in this environment.
The platform should enable the massive data acquisition required to study the
spatial variability, through sensors and equipment that will be embedded in the
platform. The proposal is based on a systematization of scientific work containing
the main methodologies and technologies employed in agricultural vehicles and
robots, which were used as a basis for constructing the presented model.
Furthermore, a preliminary study of working conditions and the desired
characteristics of the project were performed. The design of the mobile robotic
platform has been developed entirely in a virtual environment by 3D CAD
software. This allows checking for interference between components during
operation and, if necessary, changes in the design can be done. Moreover, data
from the computer model are used to create the kinematic and dynamic models. It
was established that the structure would be rectangular in gantry shape, with 1.80
m between the ground and the base of the chassis. The propulsion and the steering
system are 4WD and 4WS, respectively, with each wheel fully independent from
each other. A turbo diesel engine was used as main power source and hydraulic
systems with proportional valves were used for power transmission. The actuators
control is performed by dedicated controllers that receive the control parameters
by network, and perform the control of the actuators in a closed loop control
system. The data transmission between controllers and an embedded computer,
which contains other sub-routines (localization, navigation, data collection), is
carried out by a CAN fieldbus. Finally, a wireless network with Ethernet standard
is responsible for the communication between the mobile robot platform and the
control station.

Keywords: Agricultural automation, remote sensing, agricultural mobile robot,

robotics, ISOBUS, control systems.

INTRODUCTION

Prior to the beginning of the study of robotics in agriculture, it is necessary to

understand how the automation systems and the use of information technology are
inserted in the agricultural sector. Basically, it is possible to say that these topics,
as well as a roll of others, contribute to arrange the management system known as
Precision Agriculture. Amongst the various approaches related to Precision
Agriculture, it is common to find quotes related with massive data collection, geo-
referenced information systems, maps generation and variable-rate application.
Meanwhile, these operations, when conducted isolated or without an appropriate
management, do not provide profits or can provide confusing data.
As seen in the presented scenario, it is assumed that Precision Agriculture
should be seen as a set of techniques for the management of agricultural
production, which aims to reduce uncertainty in making decisions for better
understanding field variables as well as managing these parameters. This method
of management incorporates several areas of science, such as agricultural
sciences, engineering, geostatistics, computer science, amongst others, resulting
in a multidisciplinary system (Srinivasan 2006). Therefore, those involved with
Precision Agriculture development and its use must have vast amounts of data,
which are derived from multiple sources, to perform the tasks on making
decisions (McBratney et al. 2005).
Precision Agriculture allowed changes in management of agricultural
activities. One particular area that was managed as a whole has become
fragmented into sublevels. Due to the increased amount of data to be processed,
only certain operations can be performed using only human intervention. For this
reason, it is important to note Precision Agriculture in terms of spatial and
temporal units for decision making. Due to this fact, different forms of automation
are required, especially in expensive crops.
In order to pursue the goal discussed above, there was an increased use of
automated systems, equipment and procedures for collecting and processing data,
which provided the development of new agricultural practices. These practices are
based on technologies that were already available and were previously used in
other areas, which were adapted to the agricultural environment, such as global
positioning systems, geographic information systems, sensors, communication
networks and interconnection of devices and controllers (Gozdowsk and
Samborski 2007, Lee et al. 2010, Ahamed et al. 2011). Furthermore, it is
important to develop technologies and devices for data acquisition and real time
operations, sensor fusion and communication networks. Such actions aim to
analyze the spatial variability using remote sensing as well as automating farm
equipment, or providing support for new management practices (Auernhammer
and Speckmann 2006, Auernhammer 2004).
Recently, concerns on the development of autonomous vehicles in agriculture
have been raised, in order to support this demand. Most elaborate studies on
driverless vehicles began to be developed in the early 1960s (Fountas et al. 2007).
These initial studies were not successful due to their lack in comprehending the
complexity of the real world. Most of them adopted an industrial-style to
agriculture, where everything was known previously, and the machines could
work entirely with pre-defined paths, like a production line.
The current challenge is to develop apparatus that are smart enough to work in
a non-modified or semi-natural environment. These machines do not need to be as
smart as a human being, but should present a reasonable behavior in recognizing
contexts. Thus, it should incorporate enough artificial intelligence to work
efficiently and safely for long periods of time, autonomously, in semi-natural
environment, while performing an useful task (Pedersen et al. 2005). One way to
understand this complexity is identifying what people would do in certain
situations, and decompose the actions in the machine control (Blackmore et al.
2005).
In scientific literature, studies that aim to adapt the commercial agricultural
machinery to produce autonomous agricultural platforms (autonomous
agricultural vehicles or robots) are found (Reid et al. 2000, Keicher and Seufert
2000). Currently, there is a strong tendency to develop mobile robots and/or
autonomous vehicles for use in specific tasks, causing an efficiency increase of
operations and improved results when compared to the use of large tractors and
traditional accessories (Blackmore et al. 2007), as can be seen in Reid et al. 2000,
Mogensen et al. 2007, Reske-Nielsen et al. 2006, Blackmore et al. 2007.
However, the development of specific platforms presents two main challenges
(Blackmore, Fountas, and Have 2004): developing an adequate physical
infrastructure for the agricultural environment and an electronic architecture for
integration of the various electronic devices.
At this point, it is interesting to compare the objectives of the use of
autonomous systems in Brazil, Europe, Japan and United States. The greatest
difference is that in these countries, unlike Brazil, the skilled labor is not a
problem, and the operations are often performed by the owner or family members.
Thus, task automation can provide comfort and reduction of working hours.
Moreover, as this is a strategic sector, government provides subsidies to
producers, ensuring the production and blocking the entry of imported products.
In Brazil, autonomous systems aim to complement the shortage of professionals
due to the rising demand in this field, in addition to serving as a laboratory for
developing national technology.
Considering this context, the aim of this paper is to describe the development
and implementation of a modular robotic platform for data acquisition and
research of new technologies for remote sensing in agricultural environments. The
robotic platform has a multifunctional characteristic that allows the coupling of
modules for infield data acquisition by means of sensors and portable equipment.
The acquired data will be used in the study of spatial variability. The project aims
to incorporate design characteristics that are able to enhance the remote sensing
activities in agricultural environment, working in perennial, semi-perennial and
annual crops.

DESIGN OF THE MOBILE ROBOT

Concerning the mobile robotic systems, there are technical factors that hinders
the viability of these projects. The main factors are its interdisciplinary character
and the requirements for operation in real time (Yavuz 1999). It is assumed that
mobile robots are designed, built and tested all over the world. However, despite
this popularity, the discussion about complexity of the structural design and the
basic mechanisms of operation is often obscure or concentrated in the lower part
of the work.
Yavuz (2007) performed an analysis in the literature in order to indicate the
areas where there are greater focus of research concerning robotic systems. The
main topics and related articles were: decision-making mechanisms; data
acquisition subsystems; data and signal processing subsystems; adaptive control
and artificial intelligence subsystems; computer hardware subsystems; operating
software and related issues subsystems; control structure software and related
research; sensing systems and related research; actuator systems and their
subsystems.
Each of the topics mentioned above contributes to the overall functionality of
the system, which could increase the list of applications and, in general, its
sophistication. Although a project does not need to cover all these topics, it is
possible to observe the complexity that involves the design, development and
implementation of an autonomous mobile robot, particularly in terms of variety of
interdisciplinary areas. As a result, there are not many robots in service or in
household, as the projections made in the 1980 and 1990 stated.
In the initial phase of the design process, it is necessary to define the
functionality, control architecture, navigation system, size of the robot, power
supply and other requirements. However, selecting a solution amongst many
options available for one of these requirements, is not simple, because the
compatibility and suitability of any choice is dependent on the interaction of the
subsystem of interest to the global system, and its performance to execute its task.
However, nowadays research on agricultural robots focused on the
development of the robot, and not on the needs of agriculture related to robots.
According to (Blackmore et al. 2007), this condition causes the robot projects to
not reach the highest level of quality.
The approach adopted in this study starts with a brainstorm process. In the
scientific literature, it is possible to find some studies that aim to define the design
parameters and customer requirements for an agricultural mobile robot project
(Sørensen et al. 2008, Sørensen et al. 2010, Sørensen et al. 2006, Sørensen et al.
2007, Tabile et al. 2011). In those researches, it was applied the QFD tool in the
process of design of an agricultural mobile robot, following the model presented
by (Chan and Wu 2005). It is possible to evidence that, over the time, new
comparisons and conclusions are presented by the researches aiming to improve
the model capability. Possible “customers needs” have been identified using
several information sources such as: literature reviews, current research in
robotics and selection of existing products.
During this stage, several customers’ requirements were listed, each of them
associated with a relative importance. Customers’ requirements were divided into
three main categories, called generic requirements. The main categories were
defined as: mobility, navigation and autonomy and their respective customer
requirements are listed below.

Mobility: reduces the amount of man-hour; able to transport an external module;

easy to assemble an external module; easy to transport; easy to operate; flexible;
good maneuverability; adjustable to the row size; low power consumption;
operate on soft soil; operate in all stages of culture; uses renewable energy; light
weight; small size; low noise.

Navigation: efficient; easy operator training; automatically data acquisition;

reduces repetitive tasks; avoids damage to humans; animals; obstacles; etc.;
minimum culture damage; minimal damage in the culture and soil; low operating
cost; low purchase price; profitable; fast payback.
Autonomy: works without supervision; low maintenance; easy to start a task;
operates without breaks; safety operation; easy to maintain; able to upgrade.

Customer requirements were converted into some design parameters that have
potential to fulfill the customer requirements.

Mobility: operates without illumination; dimension; configurable by the operator;

speed for transport; adjustable gauge; omnidirectional; wheel dimension; use
commercial parts; system for join modules

Navigation: custom configurable; manual operation mode; semiautonomous;

controlled by external modules; automatic stabilization; local positioning system;
satellite navigation system; remote control.

Autonomy: susceptible to receive modular tools; remote surveillance; security

system; easy maintenance.

The results were subjected to a functional decomposition and them, splited into
main functions and alternative available techniques.

Mobility
Application: sensing; agricultural tasks
Environment: indoor; outdoor
Operational mode: autonomous; teleoperated; hybrid
Operation area: small (below 1 ha); medium (1 to 10 ha); large (up 10 ha)
Operation speed: slow (below 1 km/h); medium (1 to 10 km/h); fast (up 10 km/h)
Autonomy: small (below 30 min); medium (30 min to 2 hours); large (up 2 hours)
Payload: small (below 5 kg); medium (5 to 25 kg); large (up 25 kg)
Frame: gantry; rectangular; down; narrow
Energy demand: small (below 1 kw/h); medium (1 to 15 kw/h); large (up 15
kw/h)
Power source: gas/alcohol; diesel; battery; environment
Traction system: wheel; track; leg; hybrid
Actuators traction system: electric; hydraulic; pneumatic; gearbox
Steering system: differential steering; articulated steering; directional wheels
Actuators of the steering system: electric; hydraulic; pneumatic; gearbox
Suspension system: none; spring; compressed air; rubber bumper
Structural material: structural steel; structural aluminum; polymers; composite

Navigation
Position system: ultrasonic; GNSS; radio frequency; vision; odometry
Guidance system: digital compass; GNSS; gyroscope; odometry
Navigation system: GNSS; vision; touch sensor; ultrasonic sensor; optical sensor
Short-range obstacle detection: touch sensor; infrared; ultrasonic; laser
Long -range obstacle detection: laser; vision; ultrasonic; infrared
User input interface: radio joystick; cable joystick; wireless pc; mobile; web
Response robot interface: lcd display; pc monitor; speaker; SMS; web.
Autonomy
Mission control system: software; hardware; hybrid

Description of the electro-mechanical system

The agricultural mobile robot designed was named Agribot, with the aim to be
a platform for the development of experimental control, navigation technology
and data acquisition in the agricultural environment. The main application of the
robot is to perform remote sensing of the most important agronomic parameters of
Brazilian crops. The application will be executed in large areas and it does not
require actions with high power, as in farming operations, but only to move
effectively in this environment. Figure 1 shows an isometric view of the robotic
platform with all the mechanical parts.

01 01

02
02 12
03
04
09
05

11
03
08
04 07
05 10
06 09
06
08

07

Figure 1. a) Isometric view of the robotic platform; b) General view of the main
components of the robot

In Figure 1a: 01-Diesel engine and pumps system; 02-floor; 03-back wheel
module; 04-secundary frame; 05- main frame; 06-propulsion system; 07-
suspension system; 08-front wheel module; 09-steering system.
In Figure 1b: 01-cooler system; 02-Diesel engine; 03-fuel tank; 04-secundary
frame; 05-hydraulic system support; 06-front wheel module; 07-propulsion
hydraulic motor; 08-gauge adjust system; 09-hydraulic pumps; 10-ladder; 11-
hydraulic fluid tank; 12-batteries pack.

Dimension and frame

As the agricultural mobile robot is designed to operate on the main crops in

Brazilian agriculture, for almost the entire growth cycle and post-harvest, it
requires structure versatility in order to attend all the situations. It was established
that the mechanical structure would be in portico, and developed into independent
modules called: main frame and wheel module. For this application, a platform
was designed with four wheels and the movement in the field will be done in the
space between crop rows, thus avoiding damage in the culture.
In the main frame the engine is fixed, which provides power to the steering and
propulsion systems, hydraulic fluid and fuel tanks. Above the main frame there is
a second structure called secondary frame, used to fix the floor of the platform.
 The wheel module comprises an agricultural tire with dimensions of 9.5" x
24", the hydraulic motor of the propulsion system, the suspension system
mounted with an air shock absorber and the steering system, which has an
hydraulic cylinder as actuator connected to a rack, which drives a sprocket. The
steering system allows to the wheel an orientation from -133 ° until 133 ° in
relation to the origin of the system (wheel position where the robot goes straight).
This range is enough to position the instantaneous center of rotation of the robot
united with the center of mass, which enables the robot to turn around its own
axis.
To allow the robot to operate in cultures with different intra row distance, a
main frame with capability to adjust the gauge was developed. This process is
accomplished by a manually activated sprocket-rack system, which connects the
left and right wheel modules in the main frame. There is one system for the front
axle and one for the back axle, and they are operated independently. The
minimum distance between the wheels is 2.25 m and the maximum aperture is
2.40 m, the ground clearance is 1.80 m, and the height of the center of mass is 1.5
m. With the settings chosen, the platform has the maximum transverse stability of
33 and longitudinal stability of 40 .

Power system

Diesel engines are the most common mechanisms used in agriculture. Its main
feature is the high torque generated even at low speeds. The power supply system
comprises a Diesel-cycle engine model QSB 3.3 manufactured by Cummins Inc.
It has four cylinders, 3.3 cubic liters, turbocharged and intercooled system which
develops 56 kW (75 HP). Eco-friendly alternatives such as adoption of batteries
or solar power would not provide sufficient energy to ensure the desired
autonomy to the robot.

Propulsion system

The propulsion system is performed by hydrostatic transmission and consists

of two variable axial piston pumps (swashplate), with electronic proportional
control performed by solenoid. Those pumps, manufactured by Bosch Rexroth
AG, are attached directly to the Diesel engine. Each hydraulic pump works in a
closed circuit and is responsible for feeding two hydraulic motors, which are fixed
directly to the rim of the wheel. The pump 1 supplies the front-right and rear-left
motors through a tee-connector. Similarly, the pump 2 supplies the front-left and
rear-right motors. The X form connection with the engines were positioned and
connected to the pumps and ensured a differential hydraulic system, eliminating
the requirement of an individual rotation control for each motor. In total, there are
four radial piston hydraulic motors installed with two modes of operation (high
and low speed), also manufactured by Bosch Rexroth AG.

Steering system

The hydraulic steering system consists of variable axial piston pumps

(swashplate) with electronic proportional control actuate by solenoid
manufactured by Bosch Rexroth AG, and fixed directly after the pumps of the
propulsion system. The pump feeds four double-acting hydraulic cylinders, with
through rod manufactured specifically for this application, which are controlled
by a load sensing hydraulic control block. The hydraulic control block is
manufactured by Bosch Rexroth AG and has four proportional valves with two
ways, activated by solenoid. The system also presents a pressure control and relief
valve. Each hydraulic cylinder is connected directly to a sprocket-rack system.
The feedback of the position of each cylinder is performed by a linear
potentiometer manufactured by Gefran SpA positioned parallel to the hydraulic
cylinder.

Reserve system

A reserve hydraulic system is available, which is composed of a gear pump

manufactured by Bosch Rexroth AG. The pump feeds a hydraulic control block
manufactured by Hydraulic Designers Ltd. with one proportional valve with two
ways, activated by solenoid. This circuit can be hereafter used to supply several
additional components that may be inserted in the platform. The robotic platform
also has a set of batteries connected in series to supply electricity power for the
control systems, computers, sensors and other components that make up the
structure. This system currently has three batteries of 12 volts and 170 Ah
connected in parallel providing 510 Ah.

Suspension system

The suspension system is pneumatic and individual for each wheel module,
with a stroke length of 0.3 meters. The suspension system is designed to absorb
vibrations from rolling over rough terrain, and its main function keep all four
wheels in contact to the ground. Although the system is passive, if necessary, the
system can be update to an active model by adapting a control valve linked with a
source of compressed air.

Kinematic Model

The Agribot kinematic model is based on the wheel dimension and position in
relation to the center of mass (CM) of the robotic platform. To determine the CM,
four scales are placed under the four wheels of the robot. Equivalent masses are
calculated for all sides, and the proportion between them gives the position of the
center of mass. It is assumed for the kinematic model that the orientation of all
wheels is perpendicular to the instantaneous center of rotation (ICR), and that
there is no lateral sliding during the movement. Figure 2 presents the position of
the variables of the kinematic model in relation to the robotic platform frame.
The inputs of the system are: Turn radius (TR); Orientation of the TR in
relation to the frame ( ), which assumes values to /2 until /2; Scalar
velocity of the platform ( ).
The outputs are: Angular velocity of the platform ( ); Orientation ( ),
speed of displacement of the hydraulic cylinder ( , ) and the angular velocity
(rot ) of the four wheels.
 T
ICR X ICR , YICR

RWi
TR m

i
n Wi

T T
Wj X W j , YW j Wi X Wi , YWi

vCM

CM

Figure 2. Variables in the Agribot kinematic model

It is assumed for the kinematic model that the CM is the origin of the
coordinated system. The ICR can be calculated by the Equations (1) e (2).

= . ( + 2) (1)
= . ( + 2) (2)

With the ICR position it is possible to determine the steering angle of the
wheels. Two vectors for each wheel (Figure 2) are used in this calculation. The
vector has its origin joined with the position of the wheel that desires to find
the steering angle ( ) and finish in the position of the ICR. The vector has its
origin in the same point of and is oriented parallelly to the frame ending in the
opposite wheel ( ). The signal of the determinant of the matrix in Equation
(3), which is formed by the position of the two vectors, is used to determine the
orientation of the angle.

= (3)

The angle ( ) between the vectors and can be calculated using one of
the properties of vector product as presented in Equation (4).

. | [ ]|
= .
. [ ]
(4)

The next step is to convert the angle between the vectors ( ) to the steering
angle of the wheels ( ). For this purpose, some logic notations are made in
function of the TR and the angle . The steering angle of the wheels ( ) are
given by the Equation (5) until (16).
For the wheel W1 :
if (TR 0 and W1 2 ) 1 W1
3
2 (5)
if (TR 0 and 2 W1 ) 1 W1 2 (6)
if (TR 0 and W1 0) 1 W1 2 (7)

For the wheel W2 :

if (TR 0 and W2 2 ) 2 W2 2 (8)
if (TR 0 and 2 W2 ) 2 W2
3
2 (9)
if (TR 0 and W2 0) 2 W2 2 (10)

For the wheel W3 :

if (TR 0 and W3 2 ) 3 W3
3
2 (11)
if (TR 0 and 2 W3 ) 3 W3 2 (12)
if (TR 0 and W3 0) 3 W3 2 (13)

For the wheel W4 :

if (TR 0 and W4 2 ) 4 W4 2 (14)
if (TR 0 and 2 W4 ) 4 W4
3
2 (15)
if (TR 0 and W4 0) 4 W4 2 (16)

In order to ensure the synchronized movement of the four hydraulic cylinders,

according to the kinematic rules, the speed of displacement ( , ) was calculated
using the equation (17).

(17)

The angular velocity of the center of mass ( ) in rad/s is calculated by the

Equation (18).

= (18)

The maximum angular velocity allowed to the platform is 0.8 rad/s. The scalar
velocity of the platform will be automatically reduced for values bigger than this
maximum. With the position of each wheel in relation to the CM and the position
of the ICR, the radius of the patch realized for the wheel can be calculated using
the Equation (19).

= + (19)

Using the angular velocity ( ), the radius of the patch of the wheel ( )
and the diameter of the tire ( ), the speed of the wheel, in RPM, can be
calculate by the Equation (20).
 . .
= . 60 (20)
. .

All the calculations are made for the four wheels. Finally, with the Agribot
kinematic model, the desired angles ( ) for the four wheels are used as the
setpoints for the steering control system.

Control system

Some of the recent applications of mobile robots use a distributed architecture

based in fieldbus networks to attend the requirements of control and robustness
(Blackmore and Griepentrog 2006, Blackmore et al. 2007). Fieldbus control
systems replaced the traditional centralized control systems due to various
benefits, such as reduced costs of implantation and in the number of wires,
increased reliability and interoperability, improving the ability to reconfigure the
system and ease of maintenance. Although the distributed fieldbus control
systems offers several advantages over the traditional centralized control systems,
the existence of communication networks results in more complex solutions for
the design and implementation process. Networked control systems had additional
problems inherent in control applications with fieldbus, such as delays, delay
variation, limitations of bandwidth and data loss (Baillieul and Antsaklis 2007).
Aiming to deal with the robotic platform requirements and control problems, a
networked control systems architecture was adopted.

Power system: The control system of the Diesel engine is owned by Cummins Inc.
and uses a SAE J1939 high layer communication protocol, based on CAN, with
data transmission rate of 250 Kbit/s. The electronic control system requires the
transmission of some periodic messages (by the user), otherwise the engine is
turned off automatically. The input data of the system is the RPM of operation
and the output are fault alarms and some engine operating parameters.

Propulsion system: The propulsion control system of the engines is owned by

Bosch Rexroth AG. The electronic control system of the hydraulic system uses the
CAN ISO11898 protocol with data transmission rate of 250 Kbit/s and 29-bit ID.
The input data of the system are the displacement direction, motors speed, static
brake status, and the output data are motor speed and transmission fault alarms.

Steering System: A steering system must ensure the synchronism between the
wheels as a function of the maneuver performed and the vehicle geometry.There
are some problems related to the robot wheel steering system, which challenges
the development of the control system. The first issue is due to the hydraulic
system delay. Differently of electric and pneumatic actuators that usually provides
fast actuation on the controlled process, the hydraulic system used in the robot
guidance shows a slow response time and high inertia. These characteristics
influences the system performance and consequently the controller’s choice and
design. The nonlinearities in the steering actuators are another important problem.
The spool of the electrical valves are controlled by a solenoid that is controlled by
a PWM signal. The relationship between the displacement of the spool, the fluid
flow through the valve and the PWM signal are nonlinear and will depend
constructive parameters of the valve, the fluid conditions, etc.
Another problem is the inertia in the steering system due to the friction
between the wheel and different terrains such as dirt, pasture and asphalt. The
minimum value needed for the beginning of the steering movement is not constant
and depends on the amount of inertia, which is being submitted to each wheel,
and also depends on the robot mass distribution among the wheels. Moreover,
there is a difference between the inertia related to static (when the robot is fixed)
and dynamic friction (when the robot is moving).
The control system of the guidance hydraulic cylinders is done by an
electronic control unit (ECU) model MC050 manufactured by Sauer-Danfoss.
The system operates the solenoid of the Bosch Rexroth AG. load sensing control
block. This control block has constructive features that allow precise control of
the flow of hydraulic fluid, regardless of the pressure of the transmission line.
This eliminates problems caused by variation of force required to the wheels
movement, and hence the pressure in the line. This is caused by surface changes,
external forces, weight distribution, among others. The use of a double-acting
through-rod cylinder eliminates the problem of difference of flow and force
existing between expansion and retraction of the cylinder.
The input data of the system are the PWM values (0-100%) that commands the
opening and closing of the valve of each hydraulic cylinder. The output data are
the analog values read from the linear potentiometers. The electronic control unit
of the steering system is able to communicate in the CAN ISO11898 and SAE
J1939 protocol.

CONCLUSION

The basic idea of the complexity of an autonomous mobile robot is illustrated

in this paper, with particular focus on the design challenges. Initially, the possible
areas of activity and the main consumer markets were identified. The operations
that could be performed were identified and the most important features that make
up the agricultural environment were defined. Considering these data, technical
options were selected, in view of the set parameters of operation, and among
those, the one that best fits the prerequisites of the project. The computational
modeling was performed. The manufacture of this platform enabled the
knowledge that the methodology used to develop the agricultural robot was
efficient, and met all the needs.

REFERENCES

Ahamed, T., L. Tian, Y. Zhang, and K. C. Ting. 2011. "A review of remote
sensing methods for biomass feedstock production." Biomass and
Bioenergy no. 35 (7):2455-2469. doi: 10.1016/j.biombioe.2011.02.028.
Auernhammer, H. 2004. Off-Road Automation Technology in European
Agriculture - State of the Art and expected Trends. Paper read at
Automation Technology for Off-Road Equipment, ASAE International
Conference, at Kyoto - Japan.
Auernhammer, H., and H. Speckmann. 2006. "Dedicated Communication
Systems and Standards for Agricultural Applications." In Handbook of
Agricultural Engineering., 435-452. ASAE.
Baillieul, J., and P. J. Antsaklis. 2007. "Control and Communication Challenges
in Networked Real Time Systems." Proceedings of IEEE Technology of
Networked Control Systems no. 95 (1):09-28.
Blackmore, S., S. Fountas, and H. Have. 2004. "System Requirements For a
Small Autonomous Tractor." Agricultural Engineering International: the
CIGR EJournal (Manuscript PM 04 001.).
Blackmore, S., and H. W. Griepentrog. 2006. "Autonomous Vehicles and
Robotics." In Munack, A. (eds.). CIGR Handbook of Agricultural
Engineering, edited by ASABE, 204-215. Michigan.
Blackmore, Simon, H. W. Griepentrog, Spyros Fountas, and T.A. Gemtos. 2007.
"A Specification for an Autonomous Crop Production Mechanization
System." Agricultural Engineering International: the CIGR Ejournal no.
2007 (PM 06 032. Vol. IX).
Blackmore, Simon, Bill Stout, Maohua Wang, and Boris Runov. 2005. Robotic
agriculture the future of agricultural mechanisation. Paper read at 5th
European Conference on Precision Agriculture, at Uppsala, Sweden.
Chan, L.K., and M.L. Wu. 2005. "A Systematic approach to quality function
deployment with a full illustrative example." The international journal of
management science of Computer Programming no. 33 (2):119-139.
Deng, Y. -M., S. B. Tor, and G. A. Britton. 2000. "A dual-stage functional
modelling framework with multi-level design knowledge for conceptual
mechanical design." Journal of Engineering Design no. 11 (4):347 - 375.
Fountas, S., B. S. Blackmore, S. Vougioukas, L. Tang, C. G. Sørensen, and R.
Jørgensen. 2007. "Decomposition of Agricultural tasks into Robotic
Behaviours." Agricultural Engineering International: the CIGR Ejournal
no. 9 (Manuscript PM 07 006).
Gozdowsk, D., and S. Samborski. 2007. "Book Reviews - Precision agriculture."
Journal: Communications in Biometry and Crop Science no. 2:90-94.
Keicher, R., and H. Seufert. 2000. "Automatic guidance for agricultural vehicles
in Europe." computers and electronics in agriculture no. 25 (1-2):169-
194.
Lee, W. S., V. Alchanatis, C. Yang, M. Hirafuji, D. Moshou, and C. Li. 2010.
"Sensing technologies for precision specialty crop production." computers
and electronics in agriculture no. 74 (1):2-33. doi:
10.1016/j.compag.2010.08.005.
McBratney, Alex, Brett Whelan, Tihomir Ancev, and Johan Bouma. 2005.
"Future Directions of Precision Agriculture." Precision Agriculture no. 6
(1):7-23. doi: 10.1007/s11119-005-0681-8.
Mogensen, L. V., N. A. Andersen, O. Ravn, and N. K. Poulsen. 2007. Using
Kalmtool in Navigation of Mobile Robots. Paper read at European Control
Conference, at Kos, Greece.
Pedersen, S.M., S. Fountas, H. Have, and B.S. Blackmore. 2005. "Agricultural
robots: an economic feasibility study." Precision Agriculture:589-596.
Reid, John F., Qin Zhang, Noboru Noguchi, and Monte Dickson. 2000.
"Agricultural automatic guidance research in North America." computers
 and electronics in agriculture no. 25 (1-2):155-167.
Reske-Nielsen, A., A. Mejnertsen, N. Andersen, O. Ravn, M. Nørremark, and
H.W. Griepentrog. 2006. Multilayer controller for outdoor vehicle. Paper
read at Automation Technology for Off-Road Equipment 1 e 2 Set, at
Bonn, Germany.
Sørensen, C. G., R. N. Jørgensen, J. Maagaard, K. K. Bertelsen, L. Dalgaard, and
M. Nørremark. 2010. "Conceptual and user-centric design guidelines for a
plant nursing robot." Biosystems Engineering no. 105 (1):119-129. doi:
DOI: 10.1016/j.biosystemseng.2009.10.002.
Sørensen, Claus G., Rasmus N. Jørgensen, Jørgen M. Pedersen, and Michael
Nørremark. 2006. HortiBot: Application of Quality Function Deployment
(QFD) Method for Horticultural Robotic Tool Carrier Design Planning -
Part II. Paper read at 2006 ASABE Annual International Meeting, 9 - 12
July, at Portland, Oregon.
Sørensen, Claus G., Michael Nørremark, Rasmus Nyholm Jørgensen, Kjeld
Jensen, Jørgen Maagaard, and Lars Aalkjaer Jensen. 2007. Hortibot:
Feasibility study of a plant nursing robot performing weeding operations -
part IV. Paper read at 2007 ASABE Annual International Meeting, 17 - 20
June, at Minneapolis - Minnesota - USA
Sørensen, Claus Grøn, Rasmus Nyholm Jørgensen, Jørgen Maagaard, Keld
Kjærhus Bertelsen, Lars Dalgaard, and Michael Nørremark. 2008. User-
centered and conceptual technical guidelines of a plant nursing robot. In
2008 ASABE Annual International Meeting, edited by ASABE.
Providence, Rhode Island: ASABE.
Srinivasan, Ancha. 2006. Handbook of precision agriculture : principles and
applications. Bringhamton, NY: Food Products Press.
Tabile, Rubens Andre, Eduardo Paciencia Godoy, Robson Rogerio Dutra Pereira,
Giovana Tripoloni Tangerino, Arthur Jose Vieira Porto, and Ricardo
Yassushi Inamasu. 2011. "Design and development of the architecture of
an agricultural mobile robot." Revista Brasileira de engenharia Agrícola
no. 31 (1).
Yavuz, Hakan. 1999. Conceptual design and development of an autonomous
mobile robot. [electronic resource]. Thesis (Ph.D.), University of
Lancaster.
Yavuz, Hakan. 2007. "An integrated approach to the conceptual design and
development of an intelligent autonomous mobile robot." Robotics and
Autonomous Systems no. 55 (6):498-512.

© 2014 ISPA. All rights reserved.