Opportunities for Robotic
Landmine Clearance
James Trevelyan
Note: This paper was written before
we gained a full appreciation of the demining problem. We have significantly
changed our research direction by visiting real minefields.
Robots are not likely to
be useful in minefields for a long time yet. Why
Not? - Click for More
Landmines Working Paper 1: Robotic
Landmine Clearance
James Trevelyan
May 1995
Department of Mechanical and Materials Engineering
The University of Western Australia
Nedlands 6907, Western Australia.
E-mail: jamest@mech.uwa.edu.au
Back to Demininmg Page
Summary
This paper is based in information we have been able to obtain since
we started research on this problem in December 1995. The technical solutions
proposed later in the paper are based on certain assumptions which we are
unable to confirm at the time of writing. First that an acceptable proportion
of landmines can be detonated by firmly probing the ground in the way a
human foot would strike the ground in walking. Second that a mechanical
probe can be designed to withstand the repeated mine explosions which it
would set off.
This paper provides background information for a research project which
aims to find improved ways of removing abandoned land mines. Later papers
will present more technical details on detection and eradication techniques.
 The Landmines Problem
The current solutions
Robots
Operating Costs and Finance
Acknowledgements
References
The Landmines Problem
According to current estimates, about 100,000,000 anti-personnel and
other landmines have been laid in different parts of the world. (This estimate
is quoted by numerous sources, but it is only an indication which is based
on a wide range of information gathered in different countries. No one appears
to have detailed accurate estimates. The number of mines which have to be
cleared to restore economically useful land and resources could be much
less than this estimate. ) A similar number exist in stockpiles and about
2,000,000 new ones are being laid each year. These mines, and other unexploded
ammunition, are causing between 500 and 800 deaths and maiming 2000 per
month (Report on UN International Meeting on Mine Clearance, Geneva July
5-7th 1995 by J. Nicoud, from Mines Advisory Group), almost entirely innocent
civilians who had no part in the conflicts for which the mines were laid.
Anti-personnel mines are usually designed not to kill, but to inflict horrible
injuries instead (McGrath 1994 gives comprehensive information on landmine
types and their effects). However many victims eventually die of their injuries,
and suffer a long and agonising death, usually with scant medical attention
(MAG 1994).
Some countries have banned the use of landmines and others are supportive
of a complete ban. (Australian Government 1996, moratorium also announced
in April 1996 by UK Government.) However, their low cost ($3 - $30 --- dollar
values are based on United States currency throughout) and the large numbers
in existing stockpiles make them an attractive weapon for insurgency groups
which operate in many countries with internal conflicts---the most common
cause of wars today. They are often used for self-defence by villages and
groups of people travelling in many districts where civil law and order
provides little effective protection. Angola, Afghanistan, Rwanda, Bosnia,
Cambodia, Laos, Kuwait, Iraq, Chechnya, Kashmir, Somalia, Sudan, Ethiopia,
Mozambique, Falkland Islands. A familiar and incomplete list of recent conflicts
and landmine proliferation.
Apart from deaths and injuries to people, the major effect is to deny
access to land and its resources, causing deprivation among the affected
populations. Regardless of any efforts to stop the further use of landmines,
the safe restoration of productive land is an urgent issue in affected countries.
The UN is clearing about 85,000 mines each year though funding is being
cut back as the UN works through its current funding crisis. Other non government
organizations (NGO) such as the Mines Advisory Group (MAG) are clearing
mines (MAG Annual Report 1994/95) but they measure their work in terms of
a few square kilometres per year.
At this rate, according to UN sources, it will take about $33,000,000,000
and 1,000 years to eliminate the currently known landmines (UN Mine Clearance
Operations reported in Red Cross 1995). Since similar statistics have been
reported in many different publications, they may all be based on a single
original survey, and we have not yet been able to obtain this data. Nor
do we know how reliable this information is, and we do not know which mines
have been used in different affected areas and countries. Cost estimates
vary as well and it is not clear what measures they are based upon.
The aim of this paper is to review current initiatives to solve this
problem and to propose new directions where robotics and automation technology
might provide a useful contribution. It is clear from the literature that
current technologies are not effective (See Bachground paper No. 5 for International
Meeting on Mine Clearance 1995).
It is important to understand that humanitarian demining (removing all
mines from large areas used by people and animals) is not a priority for
military engineering efforts and it would not be realistic to expect military
research and development to provide solutions. When confronted with a minefield,
armed forces need a path to reach their next objective. They can afford
to clear a narrow path or simply plough a track through the minefield. 100%
removal is not essential: the aim is to reduce the risk (mainly to vehicles)
to an "acceptable" level.
Mines have been designed to make clearance tedious
and difficult. Any improvement on military detection and removal techniques
tends to lead to design changes which defeat the improvement. Thus many
modern mines are made almost entirely from plastic and trigger mechanisms
have been refined to defeat clearance devices such as fuel/air explosions
above a minefield. Plastic mines are therefore almost impossible to
detect with current detectors --- though dogs can be trained to detect
the presence of explosive. 
While dogs can help to identify areas to be demined, they are not reliable
and consistent enough for locating individual mines. Modern plastic blast
mines are cheap and easy to manufacture. However, fragmentation mines still
contain large amounts of metal and are therefore easier to detect.
If a technical solution for large scale landmine clearance is to be effective,
it should not be seen as useful for tactical purposes because this would
almost certainly trigger design changes to defeat it. If, for example, we
developed a hyper-sensitive artificial "nose" to smell the minute
traces of gas leaking from explosive charges, this would encourage mine
manufacturers to seal the explosive charges better or to use explosives
which release less detectable gas. Alternatively, a force commander could
lay small pieces of plastic explosive (without detonators) simply to confuse
such a detector.
While some assistance may come from aid organisations, it seems unlikely
that wealthy countries will provide money for clearing landmines which typically
cost $300 - $600 each to remove. (Current estimates from Afghanistan - demining
costs vary widely and depend on the types of mine encountered, density of
mines, and environmental factors.) The people most affected by mines have
the greatest need and incentive, but the least resources.
If the cost of demining can be reduced with better techniques to a level
which relates to the economic output of the land to be cleared, financial
institutions such as the World Bank and the IMF might be persuaded to provide
low interest long term loans to help people buy and maintain the equipment
to restore their land and economic potential. Since loans are available
to help with irrigation works, it seems reasonable to think that comparable
loans might be available to restore mine-affected land.
However, this means that we should look for low cost solutions based
on simple equipment which can be operated without costly technical experts
from "western" countries.
Conventions on warfare require minefields to be marked on maps when they
are created so they can be removed after hostilities end. However, even
reasonably well-organised armies can ignore this simple requirement in the
heat of battle, especially when they withdraw or are in desperate circumstances.
Most mines laid by Argentine forces in the Falkland Islands were unrecorded.
Guerrilla groups seldom have the administrative support to record their
minefields. Therefore, most of the mines causing problems now were not recorded
when they were laid.
Government agencies provide little protection or security to their civilian
populations in many countries where mines have been laid. Therefore civilians,
travellers and traders make use of mines for self defence by laying their
own minefields. Travellers have been known to protect their campsites with
mines, and to collect the mines each morning before moving on. Therefore,
some of the mines are no longer in their original positions, even if they
were originally recorded.
Older children and adults have been known to use mines for fishing because
they are deprived of agricultural land and are desperate for food. Typically
they will try to detonate a mine in shallow water by throwing rocks at it
(McGrath 1994, p43 on mines as tools). The fish die in the shock wave from
the explosion. Many people trying this also receive injuries, or are killed
by flying rock fragments.
Mines have been laid in all kinds of terrain for some time now, so local
changes will have affected the landscape. In fertile areas, or arid land
relieved of grazing animals, vegetation will have grown. Mines which were
originally on the surface may have been overgrown by bushes or grasses or
buried by eroded sand or soil. Buried mines may now lie under tree roots
or even under stems.
The current solutions
The first step is to educate local people about mines (MAG 1994/95 ).
Children are particularly at risk because they are not careful when walking
along unfamiliar paths, and cannot see as far ahead (because they are not
as tall as adults). Children are naturally curious, and will often collect
mines and other unexploded ordnance to use as toys or to sell, even as scrap
metal.
The next step is to mark known minefields and preferably fence them to
discourage people and large animals from entering. It is important to realise
that animals often have far more value than we realise. A sheep which may
sell for $10 in Australia may be worth the equivalent of $200 in Afghanistan,
not allowing for war-induced scarcity. When one considers that the average
income is often more than 20 times less, the real value of a sheep may be
as much as $4,000 in our terms. Sending animals to trigger mines is not
an economic (or ethical) solution.
Once local people have been educated and minefields have been marked,
work can commence on eliminating them. If possible, vegetation is removed
by burning beforehand.
A flail can eliminat many mines from flat, open areas (Red Cross 1995).
This is an armoured vehicle which carries a spinning drum with chains which
beat the ground ahead of the vehicle to trigger surface or buried mines.
Unfortunately these vehicles are expensive ($350,000) and usually eliminate
only about 90% of mines. Some mines can be thrown by the flails and they
cannot operate on boggy or rough terrain. Therefore, a more thorough method
is needed as well.
Mines with a high metal content (such as fragmentation mines and some
home-made blast mines using metal fuses) can be detected with metal detectors.
Deminers dressed in protective vests and face shields carefully sweep sensitive
metal detectors side-to-side above the ground. They need to watch carefully
for trip-wires which activate many types of mine. When metal is discovered,
the position is marked. Then, lying face down on the ground, the deminer
carefully probes the ground with a rod or bayonet to confirm that it is
a mine-like object. The rod or prodder (preferably fibreglass) needs to
be inserted at a shallow angle, about 30°, to avoid triggering the mine.
Then the deminer removes the earth from around the suspected mine until
enough has been exposed to confirm that it is a mine. More often than not,
this will simply reveal a drink can, a discarded metal part or a metal fragment
from an earlier explosion. False alarm rates from typical metal detectors
are typically 10,000:1 (from sources with field experience in Afghanistan).
Plastic mines can only be detected by gently probing the ground with
a prodder. Once again, when a worker feels a suspect object, the surrounding
earth is removed by hand.
Mines discovered in this way are preferably destroyed in situ
by a shaped charge explosion detonated remotely. Some mines have internal
sensors which trigger an explosion if they are removed or tilted. Mines
which can be made safe to handle can be accumulated in a dump and detonated
more economically in a single explosion. Mines which are not immediately
destroyed may be stolen and sold.
Robots
Robots have limited use in this scenario. See Bachground paper No. 5
for International Meeting on Mine Clearance 1995 for specific comments on
robotics. Apart from their cost and the difficulties of operating them without
well-trained people to support and maintain them, robots are usually delicate
machines which only work well in a clean, predictable, production line environment.
However, it is easy to ignore the fact that satellite TV antennas and
video cassette players are more delicate, and are built to far more precise
specifications, and are now widely used in developing countries, even where
landmines are common. Some groups use highly sophisticated weapons technology
in regional conflicts where landmines proliferate. We should not judge whether
technology is appropriate on the basis of apparent wealth and education.
What then, can robots contribute?
We need to keep the attributes of robots. They come in many different
forms, but they all have important common characteristics. If we work within
their known limitations, we are more likely to end up with a practical device.
It is difficult to design extremely robust mechanisms which are capable
of delicate manipulation. Such mechanisms tend to have delicate parts which
need protection from mechanical overloads.
Figure 1. Two possible ways of using robots to help with
landmine clearance. Robots can work well without close human supervision
if they perform simple repetitive tasks which can be well defined (by prior
programming).
Robots work well with:
- simple, repetitive, well-defined movements and operations;
- electronic sensors to measure process characteristics on-line;
- high precision tasks;
- occasional human interaction or supervision for optimum efficiency;
- environments free of hazards such as dust, dirt, corrosion, and temperature
extremes;
- complex movements provided sufficient information is available to calculate
required movement;
- electronic sensors for modifying the robot's movements in response
to task variations, provided the robot is rigid and quick to respond.
Robots have often failed to work well when taken outside these situations.
For example, much research has been directed towards autonomous vehicles
(mobile robots) for use outdoors, but finding a combination of sensors which
provide reliable long-term operation has been extremely difficult. This
problem is not yet solved.
Robots can perform tedious and monotonous tasks.
One task which a robot could do quite well is scanning the ground to
locate and classify buried objects with electronic sensors. Some sensing
systems, such as ground penetrating radar, require the sensor antenna to
be located at many well-defined positions in a grid pattern. ( See papers
by Pichot & Dourthe, Daniels & Dittmer, Chigenll, Azevedo et
al, Garriott et al, Fritzsche and Trinkhaus in WAPM'95 Proceedings.
) Robots may be essential for using this kind of sensor.
Robots can perform tasks without a person present.
We need to keep people away from mines. Therefore robots can help with
this by performing mine-removal operations remotely.
We could use robots as teleoperators, thereby using the operator to control
all the robot's movements. TV cameras provide remote "eyes" for
the operator. This technique has been used for bomb-disposal robots and
is in common use for tethered and free-ranging underwater robots. This is
an effective combination where the task is difficult, does not need to be
done quickly, and close human interaction (or supervision) is essential.
Unfortunately the equipment is expensive and is often unsuited for tedious
and repetitive tasks.
Robots can work unattended performing simple repetitive tasks, possibly
quite slowly. Probing the ground in a predetermined pattern is one example.
Another is scanning the ground with a sensor device. The key point here
is that the task must be simple enough for a robot to perform without manual
intervention. Some form of automatic monitoring may be needed to stop the
task and call for assistance if something goes wrong.
Robots need protection from environmental extremes.
Mine clearance poses an environmental problem for any robotic device.
Even if we assume, for a moment, that the equipment will not be exposed
to the risk of accidental explosions, there are still some basic requirements
for a field operable robot system:
reliable power supply, and internal protection from power supply interruptions,
temperature moderation (avoiding direct sun exposure, heating in extreme
cold), protection from dust, corrosion, vibration, protection for delicate
mechanical parts from excessive force or vibration loads, particularly in
transport between sites, and adequate reliability, and a secure, clean maintenance
depot with trained support staff
Since one of the main reasons for considering a robot is to remove people
from the mine explosion hazard, we need to consider how to safeguard the
robot itself.
Here we can adopt some standard engineering solutions. First, we can
design the robot to be mechanically robust and provide adequate shielding
from blast and high velocity fragments. Additional dust and shock protection
will be essential. Second, we can arrange the robot so that the components
which may be expensive to protect are placed out of danger.
Figure 2. Two recently proposed concepts for robotic mine clearance. 
Figure 2 shows two types of device which have been proposed recently.
The "mine-rat" is a small light-weight autonomous vehicle which
carries a sensor to detect landmines (Nicoud and Mächler 1995).. It
is light enough not to trigger surface or buried mines. An alternative is
a larger and heavier vehicle which can withstand anti-personnel mine explosions
(Velez and Thomas 1995). Both have serious disadvantages.
The "mine-rat" can still disturb trip-wires and cannot be "hardened"
against accidental explosions. Also, some surface mines (such as the Russian
butterfly mine) can be triggered by successive applications of lower force
loads, so a "mine-rat" would be at risk from these devices.
The heavier vehicle would almost certainly be susceptible to anti-tank
mines.
Both vehicles, however, require a reliable means for navigating themselves
across the terrain. While GPS can provide a position reference, it does
not solve the problem of locating and identifying obstacles, and selecting
an appropriate path to follow towards a desired position. So both vehicles
require "human" drivers (in addition to GPS for position measurement
and recording). This, in turn, necessitates a "drivers station"
in a support vehicle nearby, much like the bomb disposal robots mentioned
before.
Both these devices attempt to access the minefield by traversing the
ground, and expose themselves to severe control difficulties and explosion
hazards as a result. If we approach the minefield from above, instead, we
can almost completely eliminate these problems.
Figure 3 shows another approach, though it does not look much like a
robot. Here a tool (a simple probe is shown) is suspended above the minefield
by a system of steel cables controlled by winches on three or more poles
or towers at the corners of the minefield.
Figure 3 - A minefield
can be accessed more safely from above.
|