Research was conducted into enhancing current landmine detection methods using novel and intuitive techniques. The failures of previous technological methods for detecting landmines were noted and based on these two principles of the research were concluded:
- Technological enhancements must integrate with the current demining toolkit available to deminers
- It is not realistic to build a single device which can be used in all situations to detect all types of ERW (Explosive Remnants of War). Instead, a device should be made that a deminer can learn to make judgements on when its use is appropriate and can therefore be trusted.
One such device adhering to these principles was already in development at ERA Technologies. The device was a Vallon VMM3 pulse-induction metal detector fitted with a GPR (Ground Penetrating Radar) sensor. The device operated in two, switchable modes; the standard aural feedback from the metal detector and aural feedback of the GPR.
It was demonstrated that ERW could be located using standard demining procedures and the metal detector. The detector was then be switched to GPR mode which aurally provided feedback about the shape and depth of the buried object. It was hoped that the GPR could be used to help discriminate between metallic objects and landmines due to their shape, thus saving time excavating every metallic object.
The device was trialled at an MOD test site by Kjell Bjork, a landmine expert who was working with the York research group. His main conclusions were that while the use of GPR showed great promise, there was still much needed progress; most notably a means of combining the feedback from the metal detector and the GPR was needed to be able to more easily relate metallic content with depth and size of the buried object.
Basis of research
Based on this conclusion, the aim of the research summarised here was to develop a way to combine in real-time, signals from both a metal detector and GPR to produce an aural feedback which would help a deminer to discriminate between buried metallic objects and landmines, as well as improving the detection of minimum metal landmines in non-ideal conditions. (For example, a landmine is not easily located in ground contaminated with metallic fragments such as bullet casings. Using GPR to identify the shape of buried objects would enable the deminer to discriminate between small pieces of shrapnel and a firing pin inside a landmine.)
It was decided to meet this aim using the principle of sonification which may be briefly demonstrated by considering a car. When a car is running normally, one becomes less aware of the noise of the engine, almost to the point where it is not consciously noticed. However, if a fault were to develop, such as engine knocking – the brain would immediately recognise this sound as abnormal, even if the driver was unaware as to what caused change in the sound, their attention would be focussed on the source of the abnormal sound. This can be demonstrated in all manner of situations, even when there is loud background noise (such as a car radio).
It was hypothesised that by sonifying metal detector and GPR data, the same effect of listening to a car engine and aurally detecting and diagnosing problems as mentioned above could be applied to the ground. A deminer would be able to interact with the detector and the ground using their brain as data processor for the aural feedback. It was hoped that a deminer would be able to learn how the sounds of the detector correspond to the constituency of the ground and therefore make better judgements as to detecting and identifying buried objects. As the method of operation would be almost identical to that of a metal detector, the new technology would integrate seamlessly into existing demining toolkits as an addition and not as a replacement.
New sonification techniques were developed to produce aural feedback from both detectors. Most importantly, no filtering of the signals was done, instead the signals were synthesised into frequency modulated audio streams using different wave shapes and mixed together. Test results demonstrated the ability of a user to:
- Identify discontinuities and abnormalities in the ground
- Distinguish between metallic and non-metallic buried objects
The results clearly demonstrated that signal combination of data from multiple sensors using sonification vastly improves the detection and discrimination of buried mine and mine-like objects.
Due to the noted adaptive filtering of sound by the brain it was then hypothesised that results equally conclusive could be made in non-ideal soils such as gravel, compacted stones, ferrous and metal-contaminated soil. Quantitative tests need to be performed, as well as trials with experienced deminers to evaluate not just the detection capabilities but also the impact on the existing demining toolkits and the training of new deminers.
The sonification methods need to be improved, paying particular attention to the fact that a deminer may be listening to these sounds for an hour or more at a time. They must not be monotonous nor fatiguing yet they must maintain a level of discretion as a deminer must always be aware of other sounds in the field.
There has been considerable research in other sensing technologies appropriate for use in mine detection, such as Nuclear Quadrupole Resonance which can be used to determine the type of materials buried in the ground. Investigation into how this could be incorporated and to the improvements in detection possible could prove advantageous.
Further work on interpreting data from the metal detector should also be undertaken. Phase information of pulse induction metal detectors can be used to indicate the shape and size of the metallic object.
By improving the methods of sonification and including new and appropriate sensors, it is hoped that a sensor can be made which can produce a detailed and accurate aural representation of the shape, depth, material and metallic content of buried objects.
If you’d like a copy of the full dissertation research, then please get in touch by leaving a comment below.