Andy Smith
MIne-action specialist
Using animals as detectors......

I have spent time with Explosive Detection Dogs (EDDs) in Afghanistan, Jordan, Angola, Mozambique, Tajikistan, Kosovo, South Africa and Bosnia Herzegovina. I have written SOPs for dog use and spent many hours quizzing their human handlers, but I am not a "dog specialist". This may be good because I have no "position" to defend. When EDD specialists explain what they are doing to me, they have to be clear and simple. Instinct, intuition and tradition are not good enough reasons for using dogs in a particular way - because when instinct, intuition and tradition were relied upon the use of dogs was unreliable. Today, I find that a scientifically rigorous approach to training and deployment is emerging and this gives me far greater confidence in the performance of the EDD Sets (an EDD Set is the dog and its Handler).

See also Chapter 8 Mine Detecting Dogs in the Generic SOPs.

Humanitarian Demining

Part 13: Using animals to detect mines and other ERW.....

Dogs were first used in Humanitarian Demining by RONCO in Afghanistan in the late 1980s, and by HALO Trust in Cambodia at around the same time. Since then, Mine Detecting Dogs (MDDs) or Explosive Detecting Dogs (EDDs) have been used in many countries. In the late 1990s, the use of large cane-rats in Africa started. And in 2008, the robot controlled use of the mongoose in Sri Lanka began to be promoted.

Many animals have a very sensitive sense of smell and can detect tiny traces of High Explosive that leak from mines and ERW as the High Explosive fill dries out. In some cases, ERW appears to be hermetically sealed and yet the animal may still signal its presence. They may be detecting the odour "bouquet" presented by the disturbed ground and the mine's case. They may even be detecting the faint residue of the person who placed the device...

While it is not always clear what they do detect, it is widely accepted that they have sometimes missed the same sort of thing that they found an hour ago, or yesterday. The accident record proves this inconsistency in performance, and it is not really surprising.

In the early days, almost anyone with an animal could set themselves up as an expert, train in their own way and claim 100% success - until the accidents started to happen. Dogs trained for airport and customs search were diverted into demining. Dogs trained on another continent were flown in and used to search for devices they had never encountered in entirely unfamiliar vegetation and terrain. While the dog was usually able to search much more land than a deminer in a day, the costs could be astronomical. Purchase price, training, handler fees and veterinary backup often made their use a false economy - especially as there could not be complete confidence that the area the dog had searched was safe. In many cases, even the dog-handlers did not dare to walk on the land that their dogs had supposedly searched.

In some cases, the reason for animals missing mines and ERW was "obvious". The devices were unfamiliar, their training was erratic, the animal was sick - or in the case of short-legged rats, the animal could not traverse the uneven ground evenly and so search everywhere. Dogs need any undergrowth that is present to be cut low and must be trained to search so that every centimetre of ground is interrogated. The short length of a rats legs means that even when the undergrowth is cut very low, they cannot traverse the stubble in straight lines and so are unable to conduct a thorough search in most conditions.

The picture shows a rat at work in Mozambique.

cane rat seraching for mines

Animals can be cute and catch the donors' attention because they suppose that their use is safer and faster than the use of manual clearance methods. In fact the animal can only be used to detect the presence of explosive somewhere nearby (usually somewhere within a two metre radius). The actual clearance of the contaminated area is still conducted using manual techniques. The preparation of the land before the animal is used can be more expensive, time consuming and even more dangerous than it would be to conduct the whole clearance without the animals. When the animals are unreliable and the land must be searched again anyway, the waste is very obvious.

The use of cane-rats provides an extreme example of the limitations of animals. The use of an animal that reproduces quickly, can be tamed easily and is resistant to local disease seemed attractive. However, the rats I witnessed were not easily trained. They needed to be guided by a harness and wires with a handler on both sides of the manually prepared mined area. They were then "dragged" by wires to cover a more or less narrow line over the previously prepared ground on which all major obstructions and vegetation had been removed. The process was very slow and the area searched was very narrow (an erratic 15cm wide lane). Where clumps of stubble got in the way, the search was not thorough. Two men with metal-detectors and magnets could have searched that same area using one metre wide lanes in a fraction of the time, and without all the preparation needed.

The behaviour of a rat when it found something varied from rat to rat, as did the method of search. Some held their nose to the ground, others sniffed the air nervously. The rats are naturally nocturnal so found it hard to work in direct sun. Some had their ears removed to reduce sun-induced cancers. And their resistance to local disease was so poor that even their food had to be imported from the neighbouring country where they were bred.

Trained rats may work in a closed environment - searching suitcases at airports, for example, or checking air samples gathered from suspect containers. Their use in a minefield stretches credibility wildly and I believe that it only occurs because donors have no idea about the actual detail of demining and think that sponsoring rats is "cute". The use of the Sri Lankan mongoose (often tamed as a house pet similar to a cat) controlled by tiny robots in dense jungle is another "bright idea" with no application except to employ the idea's originators. It might sometimes work, but often would not - and if you cannot be certain it has worked, the land must be searched again afterwards.

Dogs are a far more useful size, can be readily trained to perform repetitive tasks with concentration, and live a lot longer. Even so, their training has to be completed in the context where they will work and searching for the same kind of mine/ERW that they are expected to find. They have to be deployed in a rigidly disciplined routine that ensures full ground coverage by dogs that are concentrating on the task in hand. The training of the handler is at least as important as the training of the dog. The dogs' veterinary care, kennels, logistics support and transportation are all expensive. And at the end of the day, they are best used for finding where the mines and ERW are NOT, rather than finding individual items with pinpoint accuracy. Pinpointing can happen, but is unreliable even with the best EDD Sets.

dog and handler on long leash

Finding where the mines and ERW are NOT can be immensely valuable - saving huge sums of money by reducing the areas that need to be manually cleared.

If any undergrowth present is cut without disturbing the ground, dogs can search wide areas fairly quickly and without manual preparation of "boxes" for them to work in. This is ideal, but does require the use of mine-protected assets to safely cut the undergrowth in advance of the dogs. Some argue that the explosive molecules are concentrated in the plant growth, so cutting it makes the dog's task easier by releasing molecules. Of course, the cuttings may be some distance from the source of the molecules so the area searched manually after a dog signals may have to be bigger.

Dogs are also useful in following up mechanical ground processing. Typically, the machine may process a wide area, removing undergrowth and breaking up the soil as it goes. At this stage, the machine may have detonated some devices but there is no way of knowing whether others have been broken, missed or pushed deeper into the ground. The processed ground must be searched. When there is no reason to believe that there are any mines or ERW present, professionally deployed dogs can provide a rapid confirmation that there is no explosive in the area. If there are devices present, a ground engaging machine will have moved or broken up the concentration of explosive molecules above the device. This makes it harder for a dog to search reliably and it must be trained for the task of finding recently positioned parts of mines and ERW. If the land is soft and sandy, devices may have been repositioned and pushed deeper so a dog may be unable to find them until a concentration of explosive molecules has been allowed to gather in the ground above them again. When a dog signals, the area is cleared manually. When there are multiple signals in close proximity, the dogs must be withdrawn and other search procedures used.

Scraping off the top of dry ground using a blade (dozer or grader) is reported to work well at exposing the more moist soil in which the explosive vapour above a mine will have gathered. As that soil dries, the water carries explosive molecules into the air for the dog to find. Mines scraped off into the berms left behind by the blade may be more difficult to find.

Professional EDD deployment

There are a number of rules emerging about the use of dogs to search for areas contaminated with explosives. In essence, you have to programme the dog to work as mechanistically as possible, so making its performance as reliable as possible. The rules have been pioneered by the commercial training company "Karenswood" in the UK and by NPA's GDC in Bosnia Herzegovina. Independent variations are used by MgM in Angola. My understanding of the common rules is summarised below. Check with those running EDD successfully to find out whether they agree.

1) Start with dogs that are intelligent and healthy. Males are generally preferred.

2) Dog training can start elsewhere but the final phase must be conducted in-country (usually over several months) and involve the handler they will actually have. The handler and trainer is a "Set". A handler may have two or three dogs, but each must be trained with him/her.

3) Final training must involve real targets that accurately reflect what they will search for. This may be whole devices or bits of explosive. The targets must be clean - with no human scent on them.

4) Depending on the ground, the training minefield must be in place for at least a month (preferably three) before it is used.

5) Train with the method that will actually be used - use the same marking, PPE, everything. Training may be divided into "scent recognition" and "working", with the former conducted indoors using "games" that allow the dog to be rewarded immediately for its successes.

6) Use a method that guarantees full ground coverage of every square metre. Running the dog forward and back in a one metre wide lane can achieve this, but so can other methods. The dog should have its nose to the ground and search with enthusiasm.

7) Use a second dog over any land where there is any reason to believe that there might be mines. The second dog should cover the suspect area working in another direction from the first. A single dog can be used for confidence building or making a quality assurance check on land already searched using other methods.

8) Show your confidence by ensuring that the method involves a handler walking over every metre of the searched land.

The picture below shows a second pass in a fenced minefield. The first dog searched out and back away from the start line (rope). The second dog searches along the start-line, then moves forward to slice a metre forward into the area with its handler alongside. The start line is moved forward at the end of each lateral search, and the handler only walks on land searched by two dogs.

dog and handler alongside

8) When you can, help to release the odour by cutting the undergrowth or in dry and dusty conditions, thoroughly wetting the ground the day before. Consider using machines and dogs together to make the processes more efficient.

9) Don't work in unusual weather conditions - conditions must be similar to those prevalent during training.

10) Don't use a dog that performs at less than 100% success in training. Retrain it and, if necessary, be prepared to reject it. Dogs vary and some simply cannot be made reliable.

11) Do not allow the dog to work when tired or sick. Keep the dog motivated by keeping it BORED when not working.

12) Don't expect a dog to work in areas with strong smells (diesel fuel, faeces, etc) or where the cut undergrowth is sharp and may damage their paws and nose. And do not allow the dogs to rest next to the explosive store or collected mines!

Dogs are intelligent and want to please. They do not want to find mines, just to please their handler. The system used should ensure that dogs are not rewarded prematurely with praise or play. Restricting rewards to training (which is daily) seems to work and avoids rewarding a dog when it has false-alarmed. It also avoids disappointing a dog that has signaled correctly on an explosive trace that a manual deminer cannot find.

The use of EDD Sets is not necessarily cheaper than the use of other assets. It can be more expensive. They cannot be used everywhere and for every task, and their use must be fully integrated with other demining procedures and processes. Used in the right context and in the right way, EDDs can accelerate the release of land and be a very valuable Mine Action asset. Deployed carelessly, they can be a liability.

Generally, it is only worth establishing a national EDD capacity when the mine and ERW contamination in a country is very extensive. Planning should allow the setup and training costs to be amortised over at least five years.