Because I had expressed some doubts about the value of the HSTAMID system as long ago as 2003, I was invited to Newcastle,
UK, for an informal viewing of the detector which was a restricted technology. While the detector was 'classified', those showing it to me did not
put any constraints on my freedom to comment.
HSTAMID
combines a hand-held GPR and a metal-detector - an approach
that was also being worked on elsewhere (two programmes in UK,
for example). What made it 'different' was that it was already
being used by US soldiers in Afghanistan - so was considered
to be in a field-deployable stage of development.
I
took some targets and hoped to be able to place a few on the
riverbank outside the Pearson factory where the brief meeting
took place. Pearson had a large lawned area, but I was only going
to use the mowed dandelions on the riverbank. The Pearson
people would not let me dig there even when I promised to lift
turf carefully and return it so that no one would know. So
my targets had to be placed above ground.
Surprisingly,
I could still learn something from that. The system
is heavy with a small search-head
which would mean an advance in small increments. It was foldable and had a few
design bugs (a wimpy looking head-hinge, exposed wire to the head,
low-volume - but nothing that could not be addressed). Also,
the thickness of the head-plastic looked too thin so that it would quickly wear through. It was designed to work below water (as a
metal-detector only) so was presumably well sealed against moisture but that can cause a problem when condensation forms inside the sealed unit in hot climates.
Some
models had the metal-detector and GPR separately operated but on this model you could not turn-on the GPR on its own.
When the MineLab F3 metal detector signalled, the volume of the
GPR signal instantly kicked in unless you had turned the GPR
off (which could be done if you want to use it as a stand-alone
and rather cumbersome metal-detector). If the GPR "found" an
object under the ground and around the metal reading,
the GPR gave an audio-signal that was very different from the
metal-detector indication. So the user-interface involved interpreting
varying sounds only. The difference in sound was easy to hear.
In that configuration, it could not be used to find mines
with no metal content - but they are more myth than reality
in any case.
It
was power-hungry and the large battery was worn on your belt. All
MineLabs are battery guzzlers but the GPR is a much greater glutton. The manufacturers assured me that the battery type could be varied (a US military battery with a weird voltage was being used).
The
operator said that, with experience, you can begin to tell (very
crudely) the size and shape of the GPR detection - partly from
the speed of the signal and partly by moving the search-head
towards it from all sides as you would with a metal-detector - as long as the target is fairly big (bigger than many AP blast mines).
There was a different GPR audio-signal when passing over a [Plasticine filled] PMN and a GYATA-64 placed on the surface. The signal variation reflected the different cavities
inside these mines (both have a very similar outward appearance).
However, he said that normal use would be that any GPR signal
accompanying a metal-detection should always be investigated. Would it have 'seen' this mine below the surface? The fact that I was not allowed to find out may say something.
Ground
compensation and set-up were automatic, with a female voice
announcing when the detector was ready to use. Cute, if a little
toe-curling.
Above
surface, the metal-detector and GPR combination worked
- with the GPR kicking in on everything with bulk that the metal-detector
"found". It even kicked in on an MUV fuze - with cavities inside
the metal - while staying silent when the metal was a solid
lump of frag, etc. Well, well, but the metal-detector had already signalled on this of course, so it was not telling me anything new.
It all sounds
very simple - and it was. I was thrown a little by the 40hrs
training time recommended, but was told it was necessary to break existing
habits and reinforce new ones. This was the reason given to explain why I was not allowed to use it myself. In retrospect, I have little doubt that I was being 'conned' and the GPR signals were in response to something the operator was doing. They believed that it would be able to do things soon, but it was not doing it yet.
The
questions over reliability in a real context could not begin
to be addressed. I could not begin to assess how reliable it would be when the operator could not see the target on the surface, How would it have performed if the target
were buried among stones, roots, on dry or wet
ground, flush with surface or at varying depths? These are critical questions and should be assessed with a range of
targets. I could not begin to assess how deep the GPR or the metal-detector
was capable of searching - or how similar its performance was
to the stand alone model of the Minelab F3.
If
the GPR element could be made reliable - or if it is easy to tell in what
conditions is may not be reliable (this could be what takes so
long in training) I can think of places where it might be useful.
I can also think of many where it would never be any more useful
than a metal-detector on its own - such as a typical Afghan
hillside of stones (stones would make the GPR signal). The ability to turn off the GPR would allow
the detector to still be used there (but only if no other metal-detector were
available because it is so heavy).
If
the GPR and operator are less than 100% reliable at either identifying
a mine-like object or indicating that they cannot determine
whether a mine-like object is there - the system reduces 'safety'
over the use of the MineLab on its own (when all metal indications
would be investigated as potential mines). Safety in Humanitarian
Demining refers to both the safety of the operator/deminer and
(crucially) the safety of the eventual end-users of the land that will be declared searched and cleared, so safe to use.
In
all detection systems there are three typical false-alarms:
1)
False indication - where nothing is present.
2) False-positive - where the article present is not an explosive hazard.
3) False-negative - where an explosive hazard is present but the
detection system fails to indicate this.
Of
these, the False-indication and False-positive result in extra
work but do not have a direct impact on safety. If they make
the operator 'careless', they may have an indirect impact on
safety but this can be controlled by adequate procedures and supervision.
A False-negative is a major safety issue, with missed items
resulting in deminer and civilian injury. Any new technology should not be allowed to risk increasing
the number of these events.
Because
safety is supposed to be the prime concern in HD, concerns to
limit "False-positives" which slow the work down must start with a requirement that "False-negatives"
be reduced to zero (or as close to zero as is possible). With
existing equipment, this often comes down to devising and enforcing
safe operating procedures - and that would be the same for HSTAMIDS.
I
should stress that, even assuming that the HSTAMIDS system (with
operator) could ever be 100% effective (no False-negatives),
the addition of the GPR to the MineLab F3 does not increase
the thoroughness of current methods of mine detection using
metal-detectors. In other words, it does not increase safety
by finding more than a metal-detector would on its own. It is
not intended to. It is intended to add increased speed - but
that is achieved at the expense of thoroughness (safety). It is not an appropriate tool outside a military context (where
speed may be more important than occasional False-negatives
and missed devices).
Before
releasing this for use in HD, I recommended that the following
should be field evaluated using genuinely experienced actual
deminers under the supervision of independent
assessors:
a)
The level of operator training, base intelligence and experience
that is realistically required for the user to confidently discriminate
signals.
b)
The ability of the GPR to give meaningful signals on irregular
dry or wet ground with mixed soils, and ways for the user to
assess the parameters where its use may be unsafe in the field.
c)
Realistic assessment of detection depth - and how to do this
in-and-for each area of use.
d)
Realistic assessment of the likelihood of false-negative indications.
e)
Field durability and reliability.
f)
Cost-benefit analysis over current methods.
This has never been done, to my knowledge. Almost twenty years on, the only 'test' I know about was done by HALO Trust, receiving US money to use HSTAMIDS in a known mine-belt and declaring it a success when more than 90% of the mines were found. The other 10% had made the metal-detector signal so they were dug up aggressively - and there were accidents, the details of which have never been shared. The declaration of success was made at the 2007 ICBL shindig in Geneva which I attended as UNDP Chief Technical Advisor to Tajikistan. Leaving mines behind is only a success if you are paid for it and money is your only goal. I said something like that - to the amusement of many present.
The
cost of HSTAMIDS is many times that of conventional Minelab F3 metal detector. The HSTAMIDS would have to find at least as many mines and have amazing
speed advantages over the solo metal-detectors to be anything like cost-effective.
It does not, so it does not sell in HD.
I saw the UK's dual GPR and metal-detector sensor being used by Ghadaffi's army in Libya when I was UNDP Chief Technical Advisor there in 2009. They did not use the GPR element because the results (shown on a screen) were too confusing. Still, there were proud of it because it was expensive and sold by UK, so must be good. I did not tell them how many of the Vallon metal-detectors (made in Germany) they could have bought for the same money without the GPR add-on.
But I have heard of one place where a combined GPR and metal-detector was really being used in 2018. The Sahara desert - looking for sporadically placed large anti-tank mines. It may be good for that - but whether it is better than a conventional detector I really cannot say.