First results: Robot mapping of areas contaminated by landmines and unexploded ordnance

Abstract

Landmines and unexploded ordnance are a serious threat to the life and livelihood in post conflict areas in many parts of the world. In addition to the many casualties each year, the inaccessible roads and loss of cultivated areas have a significant impact on the local economy. Many organisations are running humanitarian demining projects to clear the contaminated areas. But progress is slow since mine clearance is a very time-consuming process, and there is no room for error since most existing techniques involves an operator on site. A number of research projects have demonstrated various mine detection robot prototypes during the past decade, yet robots do not seem to be utilized in practical humanitarian demining projects. The Biosystems Engineering Group at the University of Southern Denmark collaborates with companies experienced in design of all-terrain vehicles and sensor technology to develop autonomous tool carriers for use in biological production applications. This article presents the first results applying this combined knowledge and experience to humanitarian demining. The aim is to develop a low-cost, reliable, efficient and user-friendly robot capable of detecting and mapping land-mines. It is hypothesized that with the exception of very inaccessible terrain an autonomous robot will be more efficient and reliable for mapping detected landmines than manual methods using the same sensor technology. At the same time it does not expose the operator to the risk of harm. This paper presents the first results from the project. The existing robot platform design has been simplified to lower cost and allow repair in the field with limited tools and spare parts. The robot will be able to utilize various mine detection implements and support different detection methods simultaneously. The FroboMind architecture based on Robot Operating System (ROS) is used for robot control. Software components will be released as open-source for others to build upon.

Full text

1
First results: Robot mapping of areas contaminated
by landmines and unexploded ordnance.
Kjeld Jensen1Leon B. Larsen1Kent S. Olsen1Jens Hansen2Rasmus N. Jørgensen1
Abstract—Landmines and unexploded ordnance are a se-
rious threat to the life and livelihood in post conflict areas
in many parts of the world. In addition to the many casual-
ties each year, the inaccessible roads and loss of cultivated
areas have a significant impact on the local economy. Many
organisations are running humanitarian demining projects
to clear the contaminated areas. But progress is slow since
mine clearance is a very time-consuming process, and there
is no room for error since most existing techniques involves
an operator on site. A number of research projects have
demonstrated various mine detection robot prototypes dur-
ing the past decade, yet robots do not seem to be utilized
in practical humanitarian demining projects.
The Biosystems Engineering Group at the University
of Southern Denmark collaborates with companies experi-
enced in design of all-terrain vehicles and sensor technology
to develop autonomous tool carriers for use in biological
production applications. This article presents the first re-
sults applying this combined knowledge and experience to
humanitarian demining.
The aim is to develop a low-cost, reliable, efficient and
user-friendly robot capable of detecting and mapping land-
mines. It is hypothesized that with the exception of very
inaccessible terrain an autonomous robot will be more ef-
ficient and reliable for mapping detected landmines than
manual methods using the same sensor technology. At the
same time it does not expose the operator to the risk of
harm.
This paper presents the first results from the project.
The existing robot platform design has been simplified to
lower cost and allow repair in the field with limited tools and
spare parts. The robot will be able to utilize various mine
detection implements and support different detection meth-
ods simultaneously. The FroboMind architecture based on
Robot Operating System (ROS) is used for robot control.
Software components will be released as open-source for
others to build upon.
Index Terms—humanitarian demining, mobile robotics,
WADS
I. Introduction
LANDMINES and explosive remnants of war (ERW)
are a threat to the life and livelihood of many thou-
sands of people in many parts of the world. ERW include
unexploded ordnances (UXO), which are explosives like
grenades, mortars, cluster munition etc. that have failed
to detonate as intended. Aside from the killing and injury
of people, the landmines and UXO have a significant im-
pact on the local economy due to hindered access to water
points, schools etc. and loss of fertile agricultural areas.
No one knows how many landmines and UXO remain un-
cleared, however The Landmine Monitor[1] recorded 3956
casualties from mines, victim activated improvised explo-
sive devices (IED) and other ERW in the year 2009. Civil-
1. Institute of Chem-, Bio- and Environmental Technology
University of Southern Denmark
Campusvej 55, 5230 Odense M, Denmark
Phone: (+45) 27781926, email: kjen@kbm.sdu.dk
2. Lynex, Aalsoevej 2, 8240 Risskov, Denmark
ians made up 70% of all casualties for which the civil-
ian/military status was known, and children made up 32%
of all casualties for whom the age was known. 66 states
and 7 areas not internationally recognized are confirmed
or suspected to have contaminated areas.
Humanitarian demining are activities leading to clear-
ance of landmine and UXO hazards that poses a threat
after the conflict has ended. The aim is the identification
and removal or destruction of all hazards, from a speci-
fied area to a specified depth to ensure the land is safe for
land users[2]. The removal or destruction of a landmine or
UXO hazard is a relatively simple process once the loca-
tion is known, however the critical problem is detecting the
precise location of a hazard and ensuring that all hazards
within the area have been identified.
Mine clearing is typically performed by a deminer seg-
menting the area into marked lanes. He is then working
his way along the lanes using a metal detector or prod-
der. When a possible target is detected, he excavates it
and in case of a landmine or UXO it is either removed or
destructed on site. The mine clearance speed appears to
be 3 to 20 square meters per deminer per day depending
on the terrain and level of metal contamination[3]. Newer
mine types contains little metal and hence require a much
more sensitive metal detector. This leads to a lot of false
positives which slows down the speed significantly as all
potential targets need to be inspected carefully[4].
The idea of using a mobile robot fitted with a mine de-
tection sensor to achieve a faster and more reliable coverage
of the area is not new, and a number of robots have been
developed and tested the past years [5]. Even though none
of them seem to have reached production on a larger scale,
many lessons may be learned from the projects.
[6] gives a thorough description of robotics for human-
itarian demining and related risky interventions and in-
cludes many references to related work within this field.
[7] describes the problems and challenges of robotics ap-
plied to demining and gives a number of recommendations
based on lessons learned. [8] analyses some most impor-
tant characteristics that should be taken into considera-
tion in building the robotic demining vehicle. [9] gives a
state of the art overview of mobile robotics for humanitar-
ian demining as well as recommendations on sensor tech-
nology, robot control, navigation etc. [10] surveyed the
robots and search methods for landmine detection over the
last decade and describes the problems involved, some of
the issues that have been overlooked, and stresses certain
guidelines for future robot design. [11] gives a review and
status summary of detection technologies that could be
applied to humanitarian demining operations. [12] made a

2
performance comparison between manual sweeping and a
teleoperated robotic system and concluded that remotely
operating a mine detector is technically very feasible, and
does not affect the detection rate negatively. [13][14] ex-
plore the idea of using common agricultural machines as
demining robots.
It is hypothesized that with the exception of very inac-
cessible terrain an autonomous robot will be more efficient
and reliable for mapping detected landmines than manual
methods using the same sensor technology. At the same
time it does not expose the operator to the risk of harm.
The aim of this project is to develop a low-cost, reliable,
efficient and user-friendly humanitarian demining robot ca-
pable of detecting and mapping as well as visually mark-
ing detected landmines and UXO within a bounded area.
The robot design is a tool carrier able to work with dif-
ferent types of mine detection tools simultaneously. The
estimated price for the tool carrier is 35.000 Euro.
II. Materials and methods
The landmine and UXO contaminated areas vary from
reasonable flat open areas to quite rough terrain with
slopes, rocks, trees, bushes and other obstacles. The sur-
face vary between soil, various types of sand, gravel and
stones which depending on the climate may be either dry
and dusty, muddy or even partially covered with water.
This terrain puts high demands on any vehicle operating
in the area, and some areas will be almost impassable by
a vehicle.
Many contaminated areas are located in countries having
a poor infrastructure with respect to logistics, availability
of materials, machine shops etc. Therefore challenges such
as transportation of the vehicle to the area of operation and
availability of fuel and spare parts needs to be taken into
account. It is also of great importance that any technology
introduced is accepted by both the authorities and the local
residents. The people and hence available labor often have
only minimal formal education, and skilled technicians may
be quite difficult to find.
A. Robot platform
Considering these challenges the robot must be rugged
and durable yet very simple in design to allow repair in
the field with limited tools and parts. It therefore makes
little sense to base a humanitarian demining robot on one
of the platforms traditionally utilized for robotics research
projects. The military has suitable platforms but these are
typically very expensive and difficult to obtain.
A better solution is to look at the agricultural and
forestry industry, where the development of tool carriers
have evolved for decades enabling some of them to work un-
der these tough conditions. Agricultural products include
relatively small and lightweight vehicles like [15] which may
be transported on a 4wd pickup truck or a regular sized
trailer. Making the mine detection module dismountable
from the tool carrier will allow the module to be utilized
for hand carried operation at locations inoperable by the
robot. It is important that the robot operation is very
Fig. 1. Casmobot platform
simple and intuitive allowing local deminers to perform
the operation with a minimum of training.
For this project it was chosen to base the robot design
on a slope mower vehicle (Fig.1) used in previous research.
Casmobot (Computer Assisted Slope Mowing Robot) was
developed by the University of Southern Denmark as part
of a biosystems engineering project with the purpose of
automating mowing of large sloped grass fields.
Casmobot is based on the commercially produced
tracked mower platform Lynex which due to its center of
gravity only 26 cm above ground is capable of working in
rough terrain including steep slopes up to 75 degrees in-
clination. The transmission is hydraulic driven by a 22 hp
gasoline engine. The width of the vehicle is 153 cm and it
weighs 295 kg. It can easily be transported on a standard
trailer or pickup.
Within the Casmobot project a robotic mower module
was developed. It enables the operator to perform the
mowing semi-autonomous as opposed to manual remote
control. A simple and intuitive user interface and two mow-
ing operation modes similar to autonomous mapping were
implemented.
The humanitarian demining robot design and develop-
ment builds upon the knowledge and experience obtained
in this project.
B. Mine detection implement
The organization DanChurchAid (DCA), became in-
volved in humanitarian mine action in the mid 1980s. In
1999 DCA started running operational demining projects
in 1999 and has carried out demining projects in in a num-
ber of countries since then. DCA typically uses a Wide
Area Detection System (WADS) and has implemented a
setup where measurements are geopositioned using GPS
precise positioning [16]. In 2010 the DCA implementation
of WADS was accredited as compliant with IMAS 03.40
Test and Evaluation of Mine Action Equipment by The
United Nations Mine Action Office (UNMAO) in Sudan
and can thus be used to mark areas as free of landmines
and UXO [17].
The DCA implementation of WADS has been used as
a portable setup (Fig.2) as well as mounted on a vehi-
cle (Fig.3) and uses commercially available metal detec-
tors widely used in the humanitarian demining community.
Output from the metal detector is continuously sampled
through an analog interface and saved on a laptop along

JENSEN ET AL. 2012: FIRST RESULTS: ROBOT MAPPING OF AREAS CONTAMINATED BY LANDMINES AND UNEXPLODED ORDNANCE.3
Fig. 2. WADS used for area clearance used by DCA in Sudan
Fig. 3. WADS used for road clearance by DCA in Angola
with the current position measured using precise position-
ing GPS (Fig.4).
After performing an area survey the saved data is post
processed to determine the location of potential targets.
A deminer then visits all potential target coordinates and
performs a manual sweep in a radius of 1.5 meter from the
target center.
The sensor used is an Ebinger UPEX 740M Large Loop
UXO Detector which uses the pulse induction principle to
detect metal components in the targets. Pulse induction
metal detection in general allows detection of deeper and
larger targets but has less discrimination between different
types of metal. The UPEX 740M is intended for fast search
of large areas, it allows an adjustable search head diameter,
it outputs an analog voltage corresponding to the amount
of metal detected. The UPEX 740M has been used in
humanitarian demining projects in a number of countries.
The demining robot may utilize various landmine and
Fig. 4. WADS setup implemented by DCA
Fig. 5. Demining robot operation modes
UXO detection implements using different detection tech-
nologies like metal detection, magnetometers, ground pen-
etrating radar, explosive vapour detection [4], mine raking
[18] etc. Due to the ability to carry multiple implements
the robot is also capable of utilizing multi-sensor tech-
nologies [11][19]. The theory of UXO detection methods
and principles is outside the scope of this project however.
Instead a WADS implement using a setup similar to the
DCA implementation the existing WADS setup designed
by DCA was developed for testing the demining robot.
An optional visual marking implement allows spray
paint marking of the surface where a potential target has
been detected. The spray paint is similar to what is used
for marking soccer fields. The visual marking serves as ad-
ditional safety to the deminer who performs manual sweeps
based on the list of target coordinates.
C. Robot computer
The robot computer runs the software architecture de-
veloped for the humanitarian demining robot. The com-
puter platform used is a FroboBox, which is a ruggedi-
zed and weatherproof computer that has been designed
for field robot platforms and applications. FroboBox runs
the Ubuntu linux distribution and provides USB, RS232
and Ethernet interfaces for sensors and actuators. The
FroboBox can be powered directly from an unregulated
battery power source, and can provide regulated voltages
of 5V and 12V to the attached sensors. It contains a WIFI
hotspot which enables wireless monitoring and program-
ming. Internet connection via 3G wireless broadband is
supported where available. The FroboBox technical doc-
umentation has been released under a Creative Commons
license [20].
D. Operator interface
Using a simple industrial remote control the operator
controls all operation modes and settings of the robot. The
robot has been programmed with three operation modes
viewed in Fig.5. Start up time is less than one minute
allowing the robot computer to boot and the GPS to obtain
a satellite fix.
Manual driving is used for driving the robot to and from
the area. The operator controls the robot speed and turn-
ing. Pressing a button on the remote control allows “preci-
sion driving” which causes the robot to throttle the speed
and respond slowly to operator control. This is ideal for
e.g. parking the robot on a trailer.

4
Fig. 6. Autonomous Mapping example
Assisted Mapping works like manual driving from the
operators point of view. However the robot actively ac-
quires data from the WADS implement and updates the
map with information about area covered and detected tar-
gets and obstacles. Attempts to remote control the robot
to drive fast over rough terrain causing the robot to vi-
brate, drive backwards over terrain not previously covered
by the WADS or drive across detected targets are pre-
vented.
Autonomous Mapping allows the operator to specify
boundaries of the area to be mapped by driving the robot
to the area edges using assisted mapping mode and mark-
ing their position. He then places the robot at the desired
start location and turns the robot to the desired direction
of driving. The robot will now autonomously cover the en-
tire area within the perimeter (Fig.6) and map all targets.
If overpassing is disabled the robot will replan the route to
avoid overpassing when detecting a potential target.
The operator has access to a continuously updated list
of potential target waypoints and a contour map show-
ing coverage and detected targets. The information can
be accessed through a built-in web server by connecting
to the robot WIFI network using a laptop, a tablet or a
smartphone. The map is formatted as an image and the
list of target waypoints is formatted as text (CSV), GPS
Exchange Format (GPX) and Keyhole Markup Language
(KML). Another way to access the information is to insert
a USB memory stick into the robot computer. Updated
versions of the map image and waypoint files will auto-
matically be written to a directory on the USB memory
stick.
The robot has a set of advanced settings which should
only be modified by a technician. They may be updated
by inserting a USB memory stick with an updated config-
uration file. The settings include:
•Maximum driving speed
•Maximum acceleration
•Minimum sensor overlap between two lanes
•Allow the robot to overpass detected targets
•Robot platform parameters
•Mine Detection Implement parameters
•Visual Marking Implement parameters
Fig. 7. Armadillo Demining Robot fitted with the WADS implement
Fig. 8. Armadillo Demining Robot system overview
III. Results
At the current state of development the robot acquires
data from the WADS implement and updates a map based
on WADS data and the estimated robot position and ori-
entation. The map contains information about the area
covered and detected potential targets. Fig.7 shows the
Armadillo demining robot with the WADS implement
mounted, and the illustration in Fig.8 gives an overview
of the modules and components of the system.
A. WADS implement
Adapting the UPEX 740M detector for use as a WADS
implement on the Armadillo Demining Robot requires only
a few modifications.
The large loop is very sensitive to nearby metal and elec-
tromagnetic interference, hence it must be placed at a dis-
tance from the robot. Practical tests showed that the loop
must be at least 60 cm away from the Armadillo platform.
A metal-free frame supporting the sensor loops constructed
with brackets for mounting in front of the robot was devel-
oped. The frame was made of wooden laths as this is easy
to repair almost anywhere. It is possible to dismount the
frame which allows hand carried operation of the WADS
implement in terrain inaccesible by the Armadillo.
A computer interface was developed for the analog out-
put of the UPEX 740M. The output Voltage range is -4V
to +3V which is converted and measured using a Robo-
Card board based on an Atmel ATmega microcontroller.
Sampled values with a resolution of 10 bits are transmitted

JENSEN ET AL. 2012: FIRST RESULTS: ROBOT MAPPING OF AREAS CONTAMINATED BY LANDMINES AND UNEXPLODED ORDNANCE.5
Fig. 9. Armadillo tool carrier design
at a rate of 20 Hz through a USB serial port. According
to Ebinger the UPEX 740M needs an independent power
supply due to the unit grounding design, so a 12V lead cell
battery therefore supplies the detector. The interface has
been released under a Creative Commons license [20].
B. Robot platform
The Armadillo Demining Robot has emphasis on mod-
ularity (Fig.9). Each of the two track modules can be re-
garded as self contained propulsion modules with on-board
electric motor, motor controller and gearbox. This means
a flexible platform where track modules and supporting
hardware such as battery pack and robot computer, can
be mounted in different configurations favoring e.g. weight
distribution or other requirements of the mine detection
implement. Each track module is powered by a 3.5kW
brushless DC motor which allows driving on slopes and in
rugged terrain.
The Armadillo carries two 48V 55Ah battery packs
based on deep-cycle lead-acid AGM technology support-
ing a 50% discharge. Tests have shown that during nor-
mal work in the field the Armadillo is capable of running
a normal working day without recharging. But naturally
that highly depends on the actual running time and ter-
rain etc. More advanced batteries such as Lithium ion
phosphate technologies will improve performance consid-
erably in terms of weight and operating time per charge.
But at the same time this will also make the robot more
expensive, and it will be more difficult to repair and obtain
spare parts in the field.
C. Software architecture
The Armadillo demining robot software architecture is
based on FroboMind [21] which is a conceptual architecture
for field robot software (Fig.10). It provides the means of
using the same generic architecture and components across
different field robots and hereby maximizing efficiency, re-
liability, modularity, extendability and code reuse. Frobo-
Mind is open source and is implemented in Robot Operat-
ing System (ROS)[22].
By utilising the ROS method of structuring the software,
each level of abstraction listed in the conceptual architec-
ture of FroboMind is created as a ROS package. Each
package contains ROS processes (nodes) as well as source
code and ROS specific files. Besides the packages defined
Fig. 10. FroboMind conceptual architecture fitted for Armadillo and
WADS
by the architecture, packages containing ROS core func-
tionalities and low-level interface drivers are implemented.
These packages is collected in a ROS stack named Frobo-
Mind. ROS stacks are collections of packages that provide
aggregate functionality, in this case the abstraction levels
of the FroboMind architecture as well as low-level interface
drivers.
The implementation of the low-level interface drivers en-
sures a reliable communication link between the FroboBox
and the sensors used by the WADS carrying Armadillo
demining robot. Furthermore the modularized setup, with
the well defined relations between the different compo-
nents, does make sure that it is possible to reuse com-
ponents in other applications.
D. Integration test
The system setup with the Armadillo platform and a
WADS implement has been tested in the field. The pur-
pose of this test was to validate the integration between
the Armadillo and WADS implement as well as obtain-
ing knowledge and experience with regards to the system
performance. The metal detector used for the WADS im-
plement is well tested in humanitarian demining projects,
so there was only a limited focus on the quality of the
measurements in this test.
A piece of farm land size 15 by 15 meter was selected
for the integration test. The UPEX 740M detector was
used to verify that no interfering metal objects were buried
there. Six pieces of 10 cm iron rod each weighing 920
gram were used as artificial targets. They were buried at a
depth of minimum 10 cm and geolocated using GPS precise
positioning.
The Armadillo with the WADS implement configured
as 1x2m compensating loop was then traversing the area
in lanes 75 cm wide by remote control. Fig.11 shows the
robot track and the true locations of the buried targets.
Measurements were recorded from the metal detector
interface at a rate of 20Hz and geotagged using the robot

6
Fig. 11. Map of the Armadillo track and true target locations
estimated state vector (SV). A point in polygon algorithm
was applied to dischard measurements outside the opera-
tion area. Measurements inside the area were preprocessed
using a threshold of 0V (recommended by Ebinger) fol-
lowed by a moving average (MA) filter:
yi=1
W
W−1
X
j=0
xi−j(1)
The filtered measurements were then fitted to a uni-
formly spaced grid across the area using bivariate inter-
polation. The software used for post processing of the
measurements has been designed to allow real time update
implementation:
BEGIN
WHILE recording DO
IF measurement available THEN
Import measurement
Geotag measurement using robot SV
IF measurement position within area THEN
IF measurement < 0 Volt THEN
set measurement = 0 Volt
Apply MA filter
Fit measurement to grid
Update contour plot and waypoint list
Wait until next timestep
END
Fig.12 shows a contour map viewing the post-processed
measurements from the WADS search integration test. A
MA window size of 5 were used. The shapes and position
inaccuracies of the local maxima can be explained by the
low 1 Hz update rate of the GPS receiver used in the test.
IV. Discussion
Based on tests of the Armadillo fitted with the WADS
implement in the field it is clear that the Armadillo robot
Fig. 12. Contour map of post-processed WADS measurements
design fits the application well. The electrical motores re-
duces platform vibrations to a minimum, and the tracks
gives the robot very good maneuvering capabilities. Al-
though the lead cell batteries makes the robot quite heavy
they are maintenance free and easy to handle in terms of
recharging.
The field tests also revealed that the platform could be
improved by lowering its center of gravity. A more simple
design could reduce the robot size and weight. This would
lower the complexity and reduce the cost as well.
The integration test revealed a potential problem with
the mechanical mount of the frame carrying the UPEX
740M large loop. Since the frame is mounted at a dis-
tance from the robot to avoid interference from metal ob-
jects, even small vibrations caused by robot movements
will transform into rather large vertical vibrations of the
frame. It needs to be investigated if these movements in-
fluence the measurements significantly and if so a damped
mechanical mount must be designed. Based on measure-
ments from the robot inertial measurement unit it is pos-
sible to determine the vibration amplitude as well as the
real time up or down position of the frame, so a software
filter might also be a solution.
Below is a list of some of the advantages and disadvan-
tages of using an autonomous robot for mapping detected
landmines and UXO within a bounded area:
Advantages
•Robotic mapping based on position estimation of the
mine detection implement allows accurate identifica-
tion of the location of potential targets to the extent
of the accuracy of the mine detection implement. The
implications of human errors are kept to a minimum.
•A robotic tool carrier allows simultaneous use of dif-
ferent mine detection sensor technologies which has a
great potential of increasing the accuracy and reliabil-
ity of the detection of targets.
•When using a robot the operator is able to keep the

JENSEN ET AL. 2012: FIRST RESULTS: ROBOT MAPPING OF AREAS CONTAMINATED BY LANDMINES AND UNEXPLODED ORDNANCE.7
required safety distance to the area of operation and
is not in danger in case of an uncontrolled explosion.
•The robot allows visual marking of detected targets.
This lowers the risk of the deminer accidentally trig-
gering an identified target when entering the area to
perform manual sweeps
Disadvantages
•Robotic mapping is costly compared to manual oper-
ation of the mine detection sensor. Additional costs
are the price of the robotic vehicle, spare parts, gener-
ator and fuel for recharging, logistic transport of extra
equipment etc. These additional costs needs to be bal-
anced with the higher efficiency, reliability and safety
of the operator.
•Some of the areas contaminated with landmines and
UXO are practically inaccessible by robotic vehicles
and definitely inoperable when using a mine detection
implement. This problem can to some extent be mit-
igated by the vehicle design, actuation of the mine
detection implement so it stays clear of the ground
when driving and designing for optional dismount of
the equipment for manual detection.
•A robotic vehicle suitable for driving in rough terrain
carrying a mine detection sensor exceeds the maxi-
mum ground pressure allowed for safe overpass of some
target types. This requires modification of the route
plan each time a new target is identified.
A. Conclusion
This article has presented the first results applying the
knowledge and experience from the agricultural robotics
domain to build a low-cost, reliable, efficient and user-
friendly humanitarian demining robot.
The robotic tool carrier Armadillo and a WADS imple-
ment were used for the integration test. The overall Ar-
madillo platform design fits the application well, but test-
ing in the field has revealed that it could be improved in
terms of lowering the center of gravity, using a more simple
design, and lowering the robot weight.
The next project milestones are a new prototype of
the Armadillo platform, migrating autonomous navigation
from the Casmobot project to this platform and designing
an additional implement for small search heads.
Steps further ahead are to include multi sensor support
to increase target detection reliability, and reduce depen-
dency on precise positioning GPS to increase stability and
reduce cost.
Design documents and source code available from the
project are available as open source for others to build
upon.
V. Acknowledgements
This research is linked to and partially funded by the
Danish Agency for Science, Technology and Innovation.
The authors wish to thank DanChurchAid, NORAD A/S
and Ebinger GmbH for supporting this work, and Rune
Bech Persson and Keld Bertelsen for providing images and
illustrations.
References
[1] Mines Action Canada, “Land mine monitor 2010,” October
2010, ISBN: 978-0-9738955-7-5.
[2] United Nations Mine Action Service (UNMAS), “Imas 09.10
clearance requirements second edition,” January 2003.
[3] Geneva International Centre for Humanitarian Demi-
ning (GICHD), “History, summary and conclusions of a
study of manual mine clearance,” August 2005, ISBN
2-88487-036-9.
[4] Geneva International Centre for Humanitarian Demi-
ning (GICHD), “A guide to mine action and explosive remnants
of war, third edition,” April 2007, ISBN 2-940369-02-X.
[5] M. Freese, T. Matsuzawa, Y. Oishi, P. Debenest, K. Takita,
E. F. Fukushima, and S. Hirose, “Development of a practical
landmine searching robot,” Proceedings of the l3th IASTED
International Conference ROBOTICS AND APPLICATIONS,
August 2007, W¨urzburg, Germany, August 2007.
[6] Y. Baudoin and Maki K. Habib, Eds., Using robots in hazardous
environments, vol. 1, Woodhead Publishing, 2011.
[7] Maki K. Habib, Humanitarian Demining: Innovative Solutions
and the Challenges of Technolog, I-Tech Education and Pub-
lishing, Vienna, Austria, 2008, ISBN 978-3-902613-11-0.
[8] S. Havlik, “A modular concept of the robotic vehicle for demi-
ning operations,” Autonomous Robots, vol. 18, no. 3, pp. 253–
262, May 2005.
[9] Yvan Baudoin et al., “Mobile robotic systems facing the human-
itarian demining problem state of the art (sota),” Tech. Rep.,
Royal Military Academy (RMA), 30 Av de la Renaissance, B
1000 Brussels, Belgium, December 2005.
[10] S. Rajasekharan and C. Kambhampati, “The current opinion on
the use of robots for landmine detection,” in Proceedings of the
2003 IEEE lnternational Conference on Robotics & Automa-
tion. September 2003, Proceedings of the 2003 IEEE lnterna-
tional Conference on Robotics & Automation, Taipei, Taiwan,
September 14-19. 2003.
[11] Geneva International Centre for Humanitarian Demi-
ning (GICHD), “Guidebook on detection technologies
and systems for humanitarian demining,” March 2006, ISBN
2-88487-045-8.
[12] H. Herman, T. Higgins, O. Falmier, J. S. Valois, and
J. McMahill, “Mine detection performance comparison between
manual sweeping and teleoperated robotic system,” 2010.
[13] M. Hemapala, V. Belotti, R. Michelini, and R. Razzoli, “Hu-
manitarian demining: path planning and remote robotic sweep-
ing,” Industrial Robot-an International Journal, vol. 36, no. 2,
pp. 146–156, 2009.
[14] V. Belotti, M. U. Hemapala, R. C. Michelini, and R. P. Razzoli,
“Lean robotics for humanitarian mine sweeping,” 2009.
[15] R.N. Jørgensen, C.G. Sørensen, J.M. Pedersen, I. Havn, H.T.
Jensen, K.and Søgaard, and Sørensen L.B., “Hortibot: A system
design of a robotic tool carrier for high-tech plant nursing,” 2007.
[16] Marcel E. Durocher and Anders Jansson, “Road verification
and clearance in eastern angola: The dca approach,” Journal of
Mine Action, Issue 10.1, August 2006.
[17] Rune Bech Persson, “Trial report handheld wide area detection
system danchurchaid sudan,” Tech. Rep., UNMAO, NMAC Su-
dan, February 2010.
[18] J. D. A. L. Silva, “Proposition and development of a robot
manipulator for humanitarian demining,” 2010.
[19] Motoyuki Sato, “Dual sensor alis for humanitarian demining
and its evaluation test in mine fields in croatia,” in IGARSS
2008, 2008.
[20] “Creative commons attribution-sharealike 3.0 unported license,”
http://creativecommons.org/licenses/by-sa/3.0/.
[21] Kjeld Jensen, Anders Bøgild, Søren H. Nielsen, Martin P. Chris-
tiansen, and Rasmus N. Jørgensen, “Frobomind, a conceptual
architecture for agricultural field robot navigation.,” in NJF
seminar 441, Automation and System Technology in Plant Pro-
duction CIGR section V & NJF section VII conference, 2011.
[22] Morgan Quigley, Ken Conley, Brian P. Gerkey, Josh Faust, Tully
Foote, Jeremy Leibs, Rob Wheeler, and Andrew Y. Ng, “Ros:
an open-source robot operating system,” in ICRA Workshop on
Open Source Software, 2009.