[Paper Review] An Experimental Study on Airborne Landmine Detection Using a Circular Synthetic Aperture Radar

Markus Schartel, Ralf Burr+3 more|arXiv (Cornell University)|May 6, 2020

Geophysical Methods and Applications18 citations

TL;DR

This study demonstrates airborne circular synthetic aperture radar (CSAR) using a UAS-mounted FMCW ground-penetrating radar to detect buried anti-personnel landmines with centimeter-level accuracy. Using 3D repeat-pass CSAR imaging and depth profiling, 11 of 15 test landmines—including shallow and deeply buried ones—were unambiguously detected, proving the method's potential for stand-off, high-precision mine detection in humanitarian demining.

ABSTRACT

Many countries in the world are contaminated with landmines. Several thousand casualties occur every year. Although there are certain types of mines that can be detected from a safe stand-off position with tools, humanitarian demining is still mostly done by hand. As a new approach, an unmanned aerial system (UAS) equipped with a ground penetrating synthetic aperture radar (GPSAR) was developed, which is used to detect landmines, cluster munition, grenades, and improvised explosive devices (IEDs). The measurement system consists of a multicopter, a total station, an inertial measurement unit (IMU), and a frequency-modulated continuous-wave (FMCW) radar operating from 1 GHz to 4 GHz. The highly accurate localization of the measurement system and the full flexibility of the UAS are used to generate 3D-repeat-pass circular SAR images of buried antipersonnel landmines. In order to demonstrate the functionality of the system, 15 different dummy landmines were buried in a sandbox. The measurement results show the high potential of circular SAR for the detection of minimum metal mines. 11 out of 15 different test objects could be detected unambiguously with cm-level accuracy by examining depth profiles showing the amplitude of the targets response over the processing depth.

Motivation & Objective

• To develop and validate a novel airborne UAS-based ground-penetrating SAR system for stand-off detection of buried anti-personnel landmines.
• To overcome limitations of linear down-looking SAR by employing circular synthetic aperture radar (CSAR) to improve depth resolution and reduce surface clutter.
• To achieve centimeter-level localization accuracy in 3D space for buried explosive threats using coherent data processing and motion compensation.
• To evaluate the performance of side-looking radar geometry in detecting both surface and deeply buried mines across diverse mine types.
• To demonstrate the feasibility of using FMCW radar with full-motion flexibility on a UAS for high-resolution subsurface imaging in real-world conditions.

Proposed method

• A multicopter UAS equipped with a frequency-modulated continuous-wave (FMCW) radar operating from 1–4 GHz was used to collect coherent radar data over a circular flight trajectory.
• The system integrated a total station and inertial measurement unit (IMU) for high-accuracy real-time localization and motion compensation during flight.
• Back-projection signal processing with Hann-windowed FFT and 16-fold zero-padding enabled range compression and phase correction for 3D-SAR image formation.
• The phase model accounted for signal propagation through air and soil using the law of refraction, with soil permittivity assumed constant at an empirical value.
• Depth profiles were generated by focusing the SAR on different planes (z₀), and amplitude variations across these planes were used to estimate target depth.
• A 2D-cell-averaging constant-false-alarm-rate (CA-CFAR) algorithm was applied on each focus plane to detect targets, with results visualized in 3D with color-coded depth.

Experimental results

Research questions

• RQ1 Can a UAS-mounted circular SAR system detect buried anti-personnel landmines with cm-level accuracy in a controlled environment?
• RQ2 How does the performance of side-looking CSAR compare to linear SAR in detecting mines near the surface and at depth?
• RQ3 To what extent does target depth and scattering center distribution affect the amplitude response in depth profiles across different focus planes?
• RQ4 Can a simple CA-CFAR algorithm reliably detect multiple mine types when combined with 3D-CSAR imaging and motion-compensated data?
• RQ5 What are the limitations of single-sensor radar detection when applied to diverse mine types, especially minimum-metal or non-metallic variants?

Key findings

• 11 out of 15 different dummy landmines were unambiguously detected using vertical polarization, with detection accuracy at the centimeter level.
• Depth profiles showed amplitude variations exceeding 6 dB across different focus planes, confirming the ability to estimate target depth with cm precision despite simplified soil permittivity modeling.
• Targets with complex scattering structures (e.g., pressure plates, VS50) exhibited asymmetric amplitude responses, indicating multiple scattering centers influence detection visibility.
• The highest detection performance was achieved when focusing on the plane where the target's maximum reflection occurred, with target 7 best visible from the northeast direction.
• False alarms occurred primarily in cluttered regions (e.g., near target 15), and target 6 remained undetected in all configurations, highlighting limitations of single-sensor approaches.
• Horizontal polarization detected 9 of 15 targets, confirming that sensor fusion across polarizations and geometries is essential for comprehensive mine detection.