Sensing the Unknown: Using Electromagnetism to Find Hidden Objects – Dr Liam Marsh

When I decided to study Electrical Engineering at university, I hadn’t thought much about what would come next. Spending my days as a first year undergraduate hearing about abstract electromagnetic theory seemed about as dull a subject as I could possibly imagine. Given the choice I would always have picked a soldering iron and a few hours in the lab over learning Biot-Savart Law or Gauss’ law. How on earth could Maxwell’s Equations, published circa 1861, possibly be of any use to me (or anyone?!) in the future?

Somehow, a few years later, I found myself once again caught up in that world of magnetic fields and electric charges. As a final year masters’ student, I was tasked with designing a next-generation walk-through metal detector (eugh, electromagnetics again). This wasn’t my choice, it was a random allocation; however, as happens from time to time the smallest of things can end up having the most unimaginable impact on our lives. The task was to make a metal detector that someone could walk through, without emptying their pockets, and for an operator to know right away whether you were carrying something you shouldn’t have been e.g. a knife. This would remove the need for an invasive pat-down. It took three years and a PhD to achieve this though.

What comes after a project you’ve dedicated yourself to for several years? Fortunately for me, I was working amongst some of the greatest minds in this field, now established as the EM Sensing Group at the University of Manchester, headed by Professor Tony Peyton. Tony knew this was the point to let momentum take you forwards, and onto something new without having to start all over again. So, we had successfully made a walk-through metal detector that could tell what metal object you were carrying. In doing so, we had a taken a system that previously went “beep” and needed human intervention, and given it a degree of intelligence to make that decision for us. In order to make this possible, we had to make wholesale changes to the design of the walk-through metal detector. Our research group theorized that we could better describe the relationship between the properties of metal objects (e.g. size, aspect ratio, and metallurgy) and the response of the detector. Thereby, deducing more information about the objects we were looking for. We call this an inverse problem; we know what the results look like, so can we deduce how the results came about. Thereby, we were able to determine not just what the object might be, but also pinpoint exactly where it is located in 3D space. The latter point alone could drastically change the nature of walk-through security devices. To realise this concept we needed to work with our industrial partners to design and build a completely new detector, from the core electronics and mechanical components, to the advanced algorithms that processed the sensor data to provide the overall output. In the end we could say that the inverse problem algorithms are what add the ‘intelligence’, but it is more accurate to say that these algorithms sit on top of a cutting edge sensor system.

Could we apply this elsewhere? As it turns out, yes you can. You can turn the system on its side, shrink it down and build it to run on batteries. You then have a hand-held metal detector that works similarly, but can be used to hunt for landmines instead. Could we know that a mine was under the ground, rather than scrap metal…without having to dig the item up and see for ourselves?

In World War II, when landmines saw their first large-scale deployment, the main methods to find them were: prod the ground gently with a stick, or use a metal detector. We were astonished to learn that as of the start of the 21st century, these methods were still preferred. Metal detectors were (and are) still on the front-line of landmine detection.

There are a few problems though…Firstly, it turns out that landmines from the last 30-40 years contain almost no metal, perhaps even just a fraction of a gram. This means that you have to build a really sensitive metal detector. Sadly, life is rarely that simple. This ignores the fact that landmines are not the only metal in the ground; in fact quite the opposite is true. Generally speaking, landmines are planted in former conflict zones, urban areas or areas with a history of human activity – where any of these things have occurred your very sensitive mine detector will potentially find hundreds of metal particles (these are often much bigger than a landmine) for every landmine you find, assuming that there are landmines where you are looking in the first place. As a deminer, every time your detector beeps, you must perform a careful excavation to identify the target as if your life depends on it, after all you might be one step from a landmine.

There is another equally prohibitive problem, that led to me making two trips a year to Croatian minefields for testing. The fact is, that if we put a number on the size of the response that a typical landmine gives to a metal detector, we would put this at about 0.04. Many areas of the world have iron-rich deposits within the soil e.g. bauxite/laterite soils. This means that our very sensitive metal detector could be set off by stray metal particles, and the very ground on which the detector is searching. If we were to assign a number to the response size of the ground, it would be 4; meaning the landmines we are looking for are 100 times less detectable than the ground they are buried in. In a lecture theatre I liken this to a situation whereby everyone in the room could talk at an elevated volume, with someone on the back row talking quietly – how could I be expected to determine this quiet voice amongst the noise?

Fortunately, our knowledge of physics can give us an advantage here. A traditional metal detector works on the principle of electromagnetic induction. Where an alternating current is passed through a coil of wire, and this produces a magnetic field. We look for tiny changes in this magnetic field to find metal objects. In a basic metal detector, we set a threshold on how big this change has to be, and then “beep” at a particular point. Using our approach, however, we apply several different alternating currents, at different frequencies, through that coil of wire. As a result we get several simultaneous measurements; this is called spectroscopy. So, instead of that single threshold change we were looking for previously, we now have a continuum of values which together provide a ‘fingerprint’ of the objects that we are trying to find. And guess what, the ground, the mines, and the scrap metals make different patterns.

Of course it isn’t as simple as it sounds. We are 10 years into this research with a team of a dozen specialists, and only now do we have prototype detectors that stand a decent chance of finding landmines. Getting this far has seen me and my team take many field trials, return to the drawing board on more than one occasion, and question whether this was possible on a monthly basis for a number of years.

It also turns out that our mundane electromagnetic theory can take us further than detecting weapons and threats. Over the last decade, whilst we have been designing and building landmine detectors, we have also managed to identify a number of different applications that our technology can also apply to. Some of our research group have built prototype systems that can classify and sort scrap metals to recover specific metals or alloys autonomously. This removes the need for human intervention and reduces the chance of mis-classification, thereby improving recycling rates. You can also adapt our spectroscopic metal detector, and make it in a different geometry, that instead of scanning over the ground, to allow us to pass food through it, to perhaps determine the internal quality of the food without the need for a destructive test. To do this, we need to increase the frequency of our metal detector significantly to to be able detect the weak conductivities that exist in biological tissues.

Perhaps the biggest challenge though, came from my mathematical colleague Dr Geoff Evatt, who knocked on my door one day and asked if I could scan a rock under a metal detector. Upon doing so I saw that it gave an enormous response, thousands of times larger than a landmine. “No problem,”, I told him, “but what is it?” He told me it was an iron rich meteorite. By iron-rich, we were talking about 100 g of iron, being scanned by a system that was designed to find 1 g of metal in a landmine. It would have all been so easy if it wasn’t for the unusual list of requirements that followed…”Would this detector work at –40 degrees?”, “Could you drag it across a sheet of ice behind a ski-doo?”, “Could you make it robust enough so that it can travel at 15 km/h to allow us to scan hundreds of square kilometres?” The list went on and on. So began my involvement in the “Lost Meteorites of Antarctica” which has previously featured on “LIVE with Scientists”. Suffice to say, building a metal detector the size of a tennis court, that can fit onto aircraft the size of Land Rovers, and survive the Antarctic climate searching for buried meteorites was not something I ever thought I’d do! After many hours in the lab, two trips to the northernmost settlement in the world, and two campaigns in Antarctica we found out first hand just how much effort is involved in designing and deploying equipment to one of the most hostile natural environments on earth.

From the -20 degree field conditions of a day testing in Ny-Alesund (Svalbard), to the 35 degree field conditions in a Croatian minefield, those early days in the lecture theatre seem very far away!

Want to find out more?

More about the writer

Dr Liam Marsh (PhD, CEng, MEng (Hons), MIEEE, MIET, AFHEA) gained a first class MEng degree (with honours) in Electrical Engineering and Electronics from the University of Manchester Institute of Science and Technology (UMIST) in 2007. He studied for his PhD ‘Electromagnetic Tomography and People Screening’ at the University of Manchester. In 2011 he took up the post of Research Associate at the University of Manchester. In 2018 Liam was appointed as a Lecturer in Embedded Systems in the Department of Electrical and Electronic Engineering at the University of Manchester. Liam’s research includes many aspects of metal detection and characterisation, magnetic induction spectroscopy/tomography, signal processing, remote sensing and security. In the past he has also worked in bio-impedance, an area in which he maintains an active interest. In recent years Liam’s research has focused on the detection and characterisation of landmines (including explosive remnants of warfare), and the development of magnetic sensing systems for polar regions. Liam currently teaches two second year undergraduate courses, Microcontroller Engineering II, and the Embedded Systems Project. He also supervises undergraduate and postgraduate project students working within his research area.