Content uploaded by Juan Agüera Vega
All content in this area was uploaded by Juan Agüera Vega on Mar 31, 2015
Content may be subject to copyright.
Applied Engineering in Agriculture
Vol. 30(3): 361-366 © 2014 American Society of Agricultural and Biological Engineers ISSN 0883-8542 DOI 10.13031/aea.30.10342 361
COMPARISON OF POSITIONAL ACCURACY BETWEEN
RTK AND RTX GNSS BASED ON THE AUTONOMOUS
AGRICULTURAL VEHICLES UNDER FIELD CONDITIONS
J. Carballido, M. Perez-Ruiz, L. Emmi, J. Agüera
ABSTRACT. Currently, many systems (machine vision, high resolution remote sensing, global positioning systems, and
odometry techniques) have been integrated into agricultural equipment to increase the efficiency, productivity, and safety
of the individual in all field activities. This study focused upon assessing a satellite-based localization solution used in
straight path guidance of an autonomous vehicle developed for agricultural applications. The autonomous agricultural
vehicle was designed and constructed under RHEA (Robot fleets for highly effective agriculture and forestry management)
project and is part of a three-unit fleet of similar vehicles. Static tests showed that 99% of all positions are placed within a
circle with a 2.9 cm radius centered at the geo-position using real-time satellite corrections (RTX). Dynamic tests between
rows demonstrated a mean (N=610) of the standard deviation for real-time base station corrections (RTK) of 1.43 cm and
for real-time satellite corrections (RTX) of 2.55 cm. These results demonstrate that the tractor was able to track each
straight line with high degree of accuracy. The integration of a Global Navigation Satellite System (GNSS) with sensors
(e.g., inertial sensor, altimeters, odometers, etc.) within the vehicle showed the potential of autonomous tractors for
expanding agricultural applications utilizing this technology.
Keywords. Autonomous tractor, GNSS, Precision agriculture, RTK-GPS, Agricultural machinery.
nnovative technologies (i.e., GNSS, GIS, machine
vision, sensors, agricultural machinery controller, and
high resolution remote sensing) are beginning to play a
vital role in agroforestry systems, as they aid in
compliance with current regulations while improving system
cost and efficiency (Fountas et al., 2006; Sørensen and
Bochtis, 2010). In recent years, several studies have emerged
suggesting that fleet of automated agricultural machinery can
increase sustainability and competitiveness in agricultural
production (Blackmore et al., 2005; Peleg, 2005; Bakker et
al., 2011). There are, however, important challenges that
must be overcome in these fleet automated systems. These
challenges include lowering the control system cost,
increasing production flexibility, and reducing the number of
devices aboard each fleet to avoid the failure of one vehicle
causing the entire fleet to be out of order. Meanwhile, new
systems must also have affordable automation systems and
comply with health and safety regulations.
The placement of fleet automated technology in the
agroforestry sector may provide a number of benefits,
including 1) reducing environmental contamination from
excessive agrochemical applications by adopting Global
Navigation Satellite System (GNSS) based site-specific
application techniques, 2) increasing yields by optimizing
site-specific input application levels and 3) decreasing
necessity of skilled farm laborers required to perform
An autonomous agricultural vehicle requires a combina-
tion of several techniques (sensors, machine vision
techniques, etc.) including GNSS. For real-time
applications that require on-the-go corrections, a
differential GNSS technique (DGNSS) is preferred to
achieve very high location accuracy. As the resolution at
which the geoposition improves, it increases the number of
plant-specific management tasks suited for automation. A
straightforward method to achieve accurate geopositioning
is to use two GNSS receivers (a rover and a base) that track
the same satellites. In this case, the position of the base (a
stationary unit) can be accurately determined using satellite
signals. The location information from the base can be used
to correct the location of the rover, and this correction
information can be communicated to the field GNSS
receiver by a radio link (Heraud and Lange, 2009; Perez-
Ruiz and Upadhyaya, 2012). This method allows for
minimization of error and higher real-time accuracy (Leica
Geosystems AG, 1999).
In today’s agricultural processes, RTK-DGNSS (Real
Time Kinematic-Differential GNSS) based auto steering
provides substantial savings in agro-chemicals and reduced
Submitted for review in July 2013 as manuscript number PM 10342;
approved for publication by the Power & Machinery Division of ASABE
in April 2014.
The authors are Jacob Carballido del Rey, PhD. Student, and Juan
Agüera Vega, ASABE Member, Professor, Department of Rural
Engineering, ETSI Agrónomos y Montes, University of Córdoba, Spain;
Manuel Pérez-Ruiz, Professor, Department of Aerospace Engineering
and Fluid Mechanics, Agro-forest Engineering, University of Seville,
Spain; Luis Emmi, PhD. Student, Center for Automation and Robotic
(UPM-CSIC), Madrid, Spain. Corresponding author: Manuel Pérez-
Ruiz, Department of Aerospace Engineering and Fluid Mechanics, Agro-
forest Engineering, University of Seville, Ctra. Sevilla-Utrera km 1,
Seville 41013, Spain; phone: +34-955481389; e-mail:
362 APPLIED ENGINEERING IN AGRICULTURE
hand-weeding requirements, with the associated
environmental and economic advantages (Griepentrog
et al., 2004; Blackmore et al., 2005; Fennimore et al.,
2010). Although the use of two GNSS receivers requires a
significant financial investment, RTK-GNSS systems are
becoming increasingly common among commercial
farming operations for automatic steering of tractors and
other types of field equipment.
One disadvantage of using RTK-GNSS solutions in
agriculture is the requirement that a base station be located
within 10 km at all times, and this results in high capital
cost. Multiple reference station RTK trials have been on-
going since the late 1990’s (Hu et al., 2003; Ong Kim Sun
and Gibbings, 2005). For example, both Leica Geo-systems
and Trimble have provided such Network RTK services for
the whole Great Britain since early 2006 (Edwards et al.,
2010). Likewise, some government institutions are working
to mitigating this challenge by developing a network of
base stations, which provide access to the RTK correction
signal over a wide geographic region via cellular or radio
modem (Mesas and Torrecillas, 2007). In the future, this
network may provide coverage to all farmers with RTK-
GNNS receivers, eliminating the need for multiple base
stations on each farm. However, another factor that must be
considered, due to the increased use of GNSS base stations,
is the lack of knowledge as to how the base station
coordinates are influenced by the movement of tectonic
plates (Prawirodirdjo and Bock, 2004).
Recently, a real-time positioning products has been
released (i.e., RTX), claiming to bridge the gap between
real-time RTK-PPP (Real Time Kinematic-Precise Point
Positioning) and Network RTK-GNSS. These develop-
ments are a combination of real-time data and innovative
positioning algorithms to deliver centimeter accuracy
around the world and allow satellite correction to be
delivered directly to the GNSS rover receiver, with no need
for additional equipment such as radios and antennas. Rizos
et al. (2012) reported that RTX is capable of providing real-
time positioning at 4 cm level horizontally (95%), with
initialization times of less than 1 min.
The aim of this study was to determine the GNSS
centimeter-level accuracy, through RTK (from base station)
and RTX (from satellite) signals, of the straight path
provided for an autonomous vehicle developed for
MATERIAL AND METHODS
GLOBAL NAVIGATION SATELLITE SYSTEM (GNSS)
Real-Time Differential GNSS Correction
With 2 cm accuracy, RTK systems are the most accurate
solution for GNSS (Global Navigation Satellite System)
applications. An RTK system requires two receivers, a
radio link, and an embedded navigation controller that
integrates rover sensors and GNSS data to compute the
final position of the rover receiver (Misra and Enge, 2006).
In this study, an RTK-GNSS receiver (BX982, Trimble
Navigation Ltd., Sunnyvale, Calif.) was used to accurately
locate the autonomous tractor for all field trials. The
GNSS-based navigation system included:
• a rover RTK-GNSS receiver with two GPS antennas
mounted on top of the tractor’s cabin 2 m above the
soil surface and 1.5 m apart,
• vehicle steering actuators,
• manual override sensors,
• steering angle sensors,
• controller that implement steering correction
• terrain compensation sensing (i.e., pitch, roll, and
The system utilized an RTK-GNSS correction signal
from a local (located ~0.3 km from the test site) GNSS base
station (Trimble Model BX982) to obtain RTK Fixed
quality accuracy. The rover was set to output the
“NMEA-0183 PTNL, AVR” string containing the
geographic coordinates (latitude and longitude) and yaw
angle in degree and range (m) between primary and
secondary antennas at 1 Hz rate via an RS-232 serial
Real-Time Extended GNSS Correction
The real-time extended (RTX) positioning is a new
technology that provides users with centimeter-level real-
time position accuracy. The correction signal is based on
satellite information generated at processing centers and
broadcast to users through satellites. Horizontal position
error obtained in real-time, via a receiver acquiring the
RTX correction data through the satellite link in North
America (Ames, Iowa), was RMS 1.4 cm, with a 95%
horizontal error of 2.4 cm (Leandro et al., 2011). Using an
RTX signal is advantageous because it does not require a
local base station for signal correction.
AUTONOMOUS AGRICULTURAL TRACTOR
The autonomous agricultural tractor was designed and
constructed under a European research project and is part
of a three-unit fleet of similar vehicles (RHEA, 2012). The
platform of the autonomous vehicle was a conventional
38 kW tractor (New Holland model Boomer T3050, 3-point
hitch, Zedelgem, Belgium) that was retrofitted for
autonomous agricultural operations. Figure 1 shows the
Figure 1. Autonomous tractor unit configuration.
30(3): 361-366 363
equipment setup used in the field experiments, and figure 2
shows how the GNSS correction signals were captured and
transmitted to the receiver through an external port.
A specially fabricated frame was located in the retrofit-
ted tractor and used to mount the most necessary
equipment, including the on-board computers, inertial
measurement unit, modems for navigation, connector
boxes, etc. The motion of the autonomous tractor had three
primary degrees of freedom (longitudinal, lateral, and yaw).
The tractor controller was responsible for sensing the
vehicle location and heading angle.
To configure a fully autonomous agricultural system
capable of ensuring precise navigation (navigation system),
it is necessary to configure a framework (hardware and
software) to merge perception (accurate vehicle
positioning) and action (steering and speed control). The
hardware framework should be modular, flexible, and
robust, exhibiting real-time multitasking features and
integrating modern standard communication protocols.
Specifically, the vehicle controller used in this part of the
experiment was based on a cRIO 9082 NI computer, and
the control algorithms were developed using the LabVIEW
graphical programming environment (Emmi and Gonzalez-
Field tests were performed over a 1 week period during
the winter of 2013 at the Center of Automatic and Robotic
field experiment site, at the Spanish National Research
Council (CISC), Madrid (latitude: 38.53894946 N,
longitude: 121.7751468 W). Three criteria used for
choosing the test plot were the following: (i) a plot that was
almost flat, (ii) a plot large enough for five 20 m rows, and
(iii) a plot that was within range of the correction base
station used in the experiment.
A static test was carried out on a building of approxi-
mately 20 m in length where an open sky was visible. In
this first test the RTX calculation was performed for the
rover receiver and provided an accurate position of the new
European correction signal using a GNSS navigation
receiver. The correction signal was tested for 30 minutes on
three different days, at different times of the day, within the
same week as a dynamic test. Based on manufacturer
recommendations (Lemmon and Wetherbee, 2005), this
testing procedure would provide enough satellite
constellation averaging to estimate the GNSS system
Each dynamic test consisted of five passes of 20 m
following a straight line (fig. 3). Two points (“AB”) for
each straight line were generated as an actual geospatial
location by an RTK-GNSS receiver using a handheld
surveying system interfaced to a rover RTK-GPS (Trimble
model Bx982). The geographic coordinates for points “A”
and “B” were obtained by placing the bottom tip of the 2 m
GPS antenna survey pole against the soil surface and
holding the pole vertically with the aid of a bubble level.
Points A and B were established for a dual-purpose: a)
the straight-line mission planning for the autonomous
tractor and b) a straight-line marked on the ground for
accuracy measurements. All passes were travelled at a
travel speed of 2.5 km h-1.
Figure 4 shows the small tillage steel piece that was
attached under the autonomous tractor, in the central axis,
to mark the ground with the actual path of the autonomous
In this experiment, two types of GNSS correction
signals were used: (1) RTK-GNSS signal provided by the
base station and (2) RTX-GNSS based on satellite
correction trough a satellite link.
The following raw GNSS data were recorded for all the
dynamic tests on the autonomous tractor: UTC time,
longitude, latitude, height, velocity, signal quality indicator,
PDOP, heading, and number of satellites. Only the time,
longitude, latitude, and heading were utilized for the
accuracy analysis. A program was created in LabVIEW
(National Instruments, Austin, Tex.) to convert geographic
coordinates to UTM coordinates.
To determine the accuracy of the autonomous tractor
path compared to the prescribed path, the single point
cross-track error (XTE) was defined as the perpendicular
distance from the straight-line “AB” to each error
measurements on the ground. Measurements were taken
every 0.2 m between the ideal straight-line and the
autonomous tractor path.
Total XTE was calculated using the root mean squared
(RMS) value of all the single point XTEs along the full
length of the straight-line (Taylor and Schrock, 2003).
Cross-track error is an important variable that affects the
potential skip or overlap.
For the t th pass, the RMS error was then calculated with
the following equation:
Nt = total number of measurement point for the t th pass,
eit = distance from the point i to the t th pass.
For the statistical analysis, the errors were calculated for
each measurement. The SAS general linear models
Figure 2. Flowchart of the location system on autonomous tractor.
procedure (SAS, 2008) was used to test for significant
differences between both treatments (RTK vs. RTX) using
ANOVA. Statistics for the GNSS receiver (RTX satellite
correction) position accuracy values in static tests were
calculated using JMP (SAS Institute, Cary, N.C.).
In total, 4970 GNSS data points were logged on three
different days in the same week: day 1 (1220 data points),
day 2 (1800 data points), and day 3 (1950 data points).
Figure 5 shows the visibility of the GNSS satellite during
test day 2 (10 GPS + 7 GLONASS); these conditions were
similar to other static and dynamic test days. Table 1 shows
the mean, standard deviation, maximum, minimum, and
RMS values for the GNSS receiver error when using the
RTX correction signals. The small RMS error for this test
with RTX correction indicates that RTX has the potential to
be used in an autonomous tractor. The magnitudes of the
average circular error probable (CEP) was 2.9 cm at 99%,
Figure 3. Straight mission for the autonomous tractor.
Figure 4. Implemented steel tillage bar on the autonomous tractor.
Figure 5. Plot of the visible GNSS satellites.
30(3): 361-366 365
which means that 99% of all positions are placed within a
circle with a 2.9 cm radius centered at a real position.
RTX-GNSS AND RTK-GNSS DYNAMIC TEST
The GNSS antennas mounting location on the autono-
mous tractor enabled an unobstructed view of the sky
during the entire trial. This allowed for optimal signal
reception regardless of satellite geometry, and the RTK and
RTX-GNSS fixed quality was obtained for the recording of
all data. The data in table 2 shows the RMS and standard
deviation values for the GNSS receiver error mounted on
the autonomous tractor when using RTK and RTX-GNSS
correction signals. The rover receiver had a 2.5 cm
horizontal accuracy and a 3.7 cm vertical accuracy on a
continuous real-time basis. This level of accuracy was
expected because the RTK technique can determine the
sensor position within a few centimeters (Trimble, 2007).
The average 2.4 cm for RMS cross-track error in the RTK
correction signal system indicates that the passes were very
straight, as was desired for the autonomous tractor. The
9.8 cm average error in the RTX correction signal system
could limit the use of RTX-based autonomous tractor
application in some horticultural crops and agricultural
operations that require a high degree of accuracy. This
unfortunate RTX accuracy was coincident with a
significantly large heading error. However, fully automatic
vehicles could be used for automated precision farming in
many other applications such as site-specific management
of weed control on extensive crops, variable rate
application in orchards and vineyards using the appropriate
implement, and variable-rate application of fertilizer based
on yield maps. Between rows there was an error with a
constant standard deviation, the average of these for RTK
was 1.43 cm and for RTX 2.55 cm. These results
demonstrate that the tractor controller was able to track
each straight line with a standard deviation of better than
3.5 cm; the vehicle lateral position error never deviated by
more than 4 cm for RTK and 10 cm for RTX.
There is a base of scientific research focused on
achieving accurate geopositioning information through
RTK-GNSS equipment mounted on an autonomous tractor
using a dedicated reference station for signal correction
(e.g., Nørremark et al., 2007; Sun et al., 2010; Griepentrog
et al., 2005). To the best of our knowledge, however, an
autonomous tractor using a DGNSS system has not been
fully implemented. This study demonstrated the feasibility
using a real-time RTX based on GNSS correction signal
from an autonomous tractor where extreme accuracy is not
required. The following conclusions were drawn based
upon the results of this research:
• RMS Easting and Northing for the static tests with
RTX correction showed values of 0.90 and 1.13 cm,
respectively. This indicates that RTX has the poten-
tial to be used to get the location of autonomous
tractors for applications that require a high degree of
• The RMS error of the autonomous tractor using the
base station (RTK-GNSS) signals was approximately
four times less than the RMS using the RTX correc-
tion signals. However, a fully automatic vehicle
could be used for automated precision farming in
many applications where a very high level of accura-
cy is not required, such as, site-specific management
of weed control on extensive crops, variable rate
application in orchards and vineyards using the ap-
propriate implement, and variable-rate fertilizer
application based on yield maps.
• The study has shown that the real-time extended
GNSS signal could be used on an autonomous trac-
tor, which greatly reducing the total equipment cost
of the system without a large performance penalty.
This research was funded by the European Union’s
Seventh Framework Program [FP7/2007-2013], under
Grant Agreement 245986 in the Theme NMP-2009-3.4-1
(Automation and robotic for sustainable crop and forestry
management). The authors thank Alberto Jarduo, Pablo
Agüera, and Soluciones Agrícolas de Precision S.L in
Cordoba/Sevilla, Spain for their technical assistance.
Table 1. Statistics for GNSS receiver using RTX correction signal on static, i.e. the autonomous tractor without motion.
Statistics for RTX Position Accuracy Values
Easting (cm) Northing (cm)
Day GNSS Data Mean S.D. RMS Max. Min. Mean S.D. RMS Max. Min.
1 1220 0.00 0.80 0.80 2.10 -2.50 0.00 0.90 0.90 2.80 -2.30
2 1800 0.00 1.00 1.00 2.80 -3.60 0.10 1.00 1.00 3.70 -3.30
3 1950 0.00 0.92 0.90 2.50 -3.20 0.10 1.00 1.50 3.20 -3.50
All data 4970 0.00 0.91 0.90 2.47 -3.10 0.07 0.97 1.13 3.23 -3.03
Table 2. Statistics for the GNSS receiver using the RTK and RTX correction signals on the autonomous tractor with motion.
RTK-GNSS (cm) RTX-GNSS (cm)
Row Measurements RMS S.D. RMS S.D.
1 122 2.71 2.10 7.89 2.94
2 122 3.36 1.52 7.02 3.44
3 122 1.36 0.87 11.73 2.13
4 122 2.93 1.61 9.91 2.88
5 122 1.65 1.05 12.73 1.34
All Rows 610.00 2.40 1.43 9.86 2.55
366 APPLIED ENGINEERING IN AGRICULTURE
Bakker, T., Asselt, K. V., Bontsema, J., Muller, J., & Straten, G.
(2011). Autonomous navigation using a robot platform in sugar
beet field. Biosyst. Eng., 109(4), 357-368.
Blackmore, S. B., Stout, B., Wang, M., & Runov, B. (2005).
Robotic agriculture. In: J. Stafford (Ed.), Proc. of the 5th
European Conference on Precision Agriculture. Uppsala,
Sweden: .Wageningen Academic Publisher
Edwards, S. J., Clarke, P. J., Penna, N. T., & Goebell, S. (2010). An
examination of network RTK GPS services in Great Britain.
Survey Review, 42(316), 107-121.
Emmi, L., & Gonzalez-De-Santos, P. (2012). Hardware architecture
design for navigation and precision control in autonomous
agricultural vehicles. Proc. of the First International Conference
on Robotics and Associated High-technologies and Equipment
for Agriculture (RHEA-2012), (pp. 217-221). Pisa, Italy:
University of Pisa.
Fennimore, S. A., Tourte, L., Rachuy, J. S., Smith, R. F., & George,
C. (2010). Evaluation and economics of a machine-vision
guided cultivation program in broccoli and lettuce. Weed Tech.,
24(1), 33–38. doi:http://dx.doi.org/10.1614/WT-09-022.1
Fountas, S., Wulfsohn, D., Blackmore, B. S., Jacobsen, H. L., &
Perdersen, S. M. (2006). A model of decision making and
information flows for information-intensive agriculture. Agric.
Syst., 87(1), 192-210.
Griepentrog, H. W., Christensen, S., Søgaard, S., Nørremark, H. T.,
Lund, M., & Graglia, E. (2004). Robotic weeding. Proc. of
AgEng 2004 Engineering the Future. Leuven, Belgium:
Technologisch Instituut vzw, 2004
Griepentrog, H. W., Nørremark, M., Nielsen, H., & Blackmore, B.
S. (2005). Seed mapping of sugar beet. Precision Agric., 6(2),
Heraud, J. A., & Lange, A. F. (2009). Agricultural automatic
vehicle guidance from horses to GPS: How we got here, and
where we are going. ASABE Distinguished Lecture Series
No.33. St. Joseph, Mich.: ASABE.
Hu, G., Khoo, H. S., Goh, P. C., & Law, C. L. (2003). Development
and assessment of GPS virtual reference stations for RTK
positioning. J. of Geodesy, 77(5-6), 292-302.
Leandro, R., Landau, H., Nitschke, M., Glocker, M., Seeger, S.,
Ghen, X., Deking, A., Tahar, M. Ben, Zhang, F., Ferguson, K.,
Stolz, R., Talbot, N., Lu, G., Allison, T., Brandl, M., Gomez, V.,
Cao, W., & Kipka, A. (2011). RTX Positioning: the next
generation of cm-accurate real-time GNSS postioning.
Proceeding of the ION GNSS, (pp. 1460-1475). Portland, Ore.:
The Institute of Navigation
Leica Geosystems AG. (1999). Introduction to GPS (Global
Positioning System). Heerbrugg, Switzerland: Leica Geosystems
Lemmon, T., & Wetherbee, L. (2005). Trimble Integrated
Surveying Techniques (White Paper). Westiminster, Colo.:
Trimble Geomatics and Engineering Division.
Mesas, F. J., & Torrecillas, C. (2007). Andalusian network
positioning. Mapping Interactivo. Retrieved from
Misra, P., & Enge, P. (2006). Global Position Systems: Signals,
Measurements and Performance (2nd ed.). Lincoln, Mass.:
Nørremark, M., Søgaard, H. T., Griepentrog, H. W., & Nielsen, H.
(2007). Instrumentation and method for high accuracy
georeferencing of sugar beet plants. Computers and Electronics
in Agric., 56(2), 130-146.
Ong Kim Sun, G., & Gibbings, P. (2005). How well does the virtual
reference station (VRS) system of GPS base station perform in
comparison to conventional RTK? J. of Spatial Sci., 50(1), 59-
Peleg, D. (2005). Distributed coordination algorithms for mobile
swarms: new directions and challenges. In A. Pal, A.
Kshemkalyani, R. Kumar, & A. Gupta (Eds.), Distributed
Computing-IWDC 2005 (pp. 1-12). Springer-Verlag Berlin
Pérez-Ruiz, M., & Upadhyaya, S. K. (2012). Chapter 1: GNSS in
precision agricultural operation. In F. Elbahhar (Ed.), New
Approach of Indoor and Outdoor Localization Systems (pp. 1-
26). InTech, under CCBY 3.0 license.
Prawirodirdjo, L., & Bock, Y. (2004). Instantaneous global plate
motion model from 12 years of continuous GPS observations. J.
of Geophysical Res., 109(B8), 1-15.
RHEA. (2012). A robot fleet for highly effective agriculture and
foresty management. Retrieved from http://www.rhea-
Rizos, C., Janssen, V., Robert, C., & Grinter, T. (2012). PPP versus
DGNSS. Geomatic World, 20(6), 18-20.
SAS Institute . (2008). SAS/STAT 9.2 User's Guide. Cary, N.C.:
SAS Institute, Inc.
Sørensen, C. G., & Bochtis, D. D. (2010). Conceptual model of
fleet management in agriculture. Biosyst. Eng., 105(1), 41-50.
Sun, H., Slaughter, D. C., Perez-Ruiz, M., Gliever, C., Upadhyaya,
S. K., & Smith, R. F. (2010). RTK GPS mapping of transplanted
row crops. Computers and Electronics in Agric., 71(1), 32-37.
Taylor, R. K., & Schrock, M. D. (2003). Dynamic testing of GPS
receivers. ASAE Paper No. 031013. St. Joseph, Mich.: ASAE.
Trimble Navigation Limited. (2007). Form 10-k annual report.