Combined Sensors for Landmine Detection
Gwangju Institute of Science and Technology, Gwangju 500-712 South Korea
Popular version of paper 1pPAc1
"Combined seismic, radar, and induction sensor for landmine detection"
Presented Monday afternoon, June 30, 2008
The United Nations estimates that the more than 110 million landmines buried around the world kill or maim more than 2,000 people per month. Landmines are relatively cheap to manufacture but, once deployed, are very expensive and dangerous to remove. Although many sensors and systems have been developed for detection and remediation of these landmines, no single sensor can reliably detect all types of buried landmines with a reasonable false alarm rate in all environmental conditions. The primary reason for this is that the soil is a very complex and inhomogeneous medium -- rocks, roots, moisture variations, scrap metal, etc. can all cause a sensor to give false alarms.
Sensor diversity will be the key to new landmine detection systems, so an experiment to investigate the potential for multiple sensors was set up at Georgia Tech. In the experiment, co-located ground penetrating radar (GPR), electromagnetic induction (EMI), and seismic data were taken. These three sensors were chosen because, collectively, they can sense a wide range of physical properties and can be operated simultaneously. The seismic sensor is sensitive to the differences between the mechanical properties of a landmine and the surrounding soil, while the GPR is sensitive to the dielectric properties, and the EMI sensor to conductivity. All of these sensors are capable of detecting most mines; however, their response to specific mines and clutter objects will be quite different since they respond to very different physical properties. For example, a small piece of metal clutter could cause a false alarm for an EMI system, but it would be unlikely to cause a false alarm for a GPR or seismic system. Likewise, a buried rock could cause a false alarm for a GPR system, but it would be unlikely to cause a false alarm for an EMI or seismic system. By combining these diverse systems, it should be possible to create a system with a better detection rate and a lower false alarm rate.
The EMI sensor probes the subsurface with time-varying magnetic fields that mostly sense magnetic and conductive materials such as metal. Simple EMI sensors (metal detectors) are capable of detecting most mines; however, they will also detect every buried metal object such as bottle tops, nails, shrapnel, bullets, etc. This results in an unacceptably high false alarm rate. In recent years, advanced EMI sensors that use a broad range of frequencies or a broad range of measurement times along with advanced signal processing have been developed. These systems are capable of discriminating between buried land mines and many types of buried metal clutter. At Georgia Tech, an advanced EMI system was developed that operates over the frequency range 300 Hz to 90 kHz. Sample measurements made with the system are shown in figure 1 for a TS-50 landmine, a MAI-75 landmine, and a patio stone. The data are plotted as Argand diagrams where the imaginary part of the EMI response is graphed as a function of the real part with frequency as a parameter. The shapes of the curves on the Argand diagrams are indicative of the type and distribution of metal in a target. The shapes are quite different in figure 1 indicating that these EMI signatures can be used to discriminate between the targets.
Fig. 1: EMI response plotted on an Argand diagram over the center of a) eight different instances of TS-50 anti-personnel landmines buried 0 to 5 cm deep, b) six different instances of MAI-75 anti-personnel landmines buried 0 to 5 cm deep, and c )three different instances of patio stones buried
A GPR is a radar system that probes the subsurface using electromagnetic waves which are reflected from objects buried in the ground. The GPR sensor developed for the experiment uses resistively loaded vee antennas to transmit and receive the electromagnetic waves. These antennas are very “clean” in that they have very little self clutter and a very low radar cross section to lessen the reflections between the ground and the antennas. Consequently, the GPR has a very “clean” response to buried targets which allows better detection of landmines and improved clutter reduction. To demonstrate the capabilities of the GPR sensor, landmines and typical clutter objects shown in figure 2a were buried in the experimental model as shown in figure 2b (burial depths are indicated in parentheses). A three-dimensional iso-surface graph of the GPR measurements of this scenario is shown in Figure 2c. The image clearly indicates the locations of all of the landmines and some of the clutter targets.
Fig. 2: (a) Photograph, (b) burial map, and (c) three-dimensional iso-surface (-19 dB) of the resulting GPR data indicating the location of the buried landmines and some clutter objects
A seismic landmine detection system probes the subsurface using seismic surface waves which excite mechanical resonances of the buried landmines, resulting in increased surface displacements directly above the landmines; the seismic waves are also scattered by the presence of buried objects. An animation of the surface motion caused by a seismic wave traveling across a minefield is shown in figure 3. The waves travel left to right and can be seen interacting with a small anti-personnel (AP) landmine buried in the center. These data were measured using a custom built, non-contact, radar sensor; however, ultrasound, acoustic, and ground-contacting sensors have been demonstrated in other experiments. The seismic system has been shown to be very resistant to most types of clutter. An image formed from another set of seismic data is shown in figure 4 along with a photograph showing the locations of the buried targets. The landmines are evident in the image while the other targets are not.
Sound 1. Animation of the interaction of propagating seismic surface waves with an AP landmine
Fig-3: (a) Burial locations and photograph of the targets and (b) image formed from the seismic data.
Fig. 4: (a) Burial map for the targets. Images formed from the sensor data: (b) EMI data on a 90 dB scale, (c) GPR data on a 20 dB scale, and (d) seismic data on a 30 dB scale.
In the three-sensor experiment, a range of mines and clutter objects were buried at various depths in the experimental model at Georgia Tech. Multiple burial scenarios were investigated with a variety of AP and anti-tank (AT) landmines and typical clutter objects. The scenario depicted in figure 5a has six buried mines (two AT landmines: VS-1.6 and VS-2.2; four AP landmines: TS-50, M-14, VS-50, and PFM-1) and 21 clutter objects such as gun shells, nails, rocks, cans, etc. All three of the sensors were scanned over the simulated minefield and their output recorded.
Figure 5b is an image formed from the measured EMI data by calculating the energy in the imaginary part of the response of the EMI sensor. The response of all of the landmines and metal clutter objects can be seen in the image; however, the response of the nonmetallic items, the rocks and the bag of sand, can not be seen. Thus, using the EMI sensor with an “energy detector” will produce many false alarms, as is evident from the image in figure 5b. An image formed from the GPR data is shown in figure 5c on a 20dB scale. The image is formed by first back-projecting the data and then summing the energy over all depths. All of the landmines can be seen in the image while the smaller clutter items on the left of the image are not apparent. The larger clutter items on the right side of the image, including the rocks and the bag of dry sand, are apparent. An image formed from the seismic data is shown in figure 5d on a 30 dB scale. The image indicates the location of the two AT landmines, the four AP landmines, and the cans. While the cans show up quite well in this image, the other clutter items such as the rocks and pieces of metal do not appear.
All three of the sensors detected the buried landmines and some of the clutter objects. By using advanced signal processing and the diversity of sensors the number of false alarms can be greatly reduced.