Metrics and Performance Analysis of Next Generation Handheld Mine Detectors
Date
Authors
Talbott, Marie
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Soldiers in theater face the threat of mines and improvised explosive devices (IEDs) which grow more sophisticated and more dangerous each year. In current conflicts, insurgents regularly use IEDs along well-traveled roads or even in urban environments to disrupt and delay soldiers during missions, as soldiers must first clear a safe path forward before proceeding. Mine and IED detection systems in theater can be integrated onto a vehicle, robotic platform, or a handheld system and use a variety of sensors to detect potential threats. In this thesis, we will analyze data collected by three handheld mine detectors that utilize Ground Penetrating Radar (GPR) to detect mines and IEDs. These detectors are prototypes intended to improve upon the handheld detector currently fielded by the US Army. We will assess each GPR system’s capabilities for detecting targets as well as the system’s limitations, particularly targets with little or no metal content. Data used in this thesis is from a collection that occurred in April and May 2016 at a US Army test site over relevant targets and threats. Data was collected with all three prototype systems and the currently fielded handheld detector over a short time period so that any environmental or weather issues that may impact system performance would be experienced by all systems. These systems are prototypes and thus a signal over background metric will be applied to the data collected with each system to calculate detections of targets (GPR response at a target location) and false alarms (GPR response not due to a target). Several methods for selecting the background for the signal over background will be investigated as there are varying levels of position information available for each detector. Once the signal over background metric is calculated for all data collected with each system, alarms will be created using a simple peak picking algorithm and using a blobbing technique. These alarms will then be scored using methodology identical to the analysis the author performs at her place of employment (The Institute for Defense Analyses) in support of the Countermine Division of Night Visions and Electronic Sensors Directorate (NVESD) based in Fort Belvoir, to assess a variety of mine and IED detection systems. Results will be shown for each system broken down into target type and target depth. Because there was rainfall during the data collection, results will also be compared for dry and wet conditions.
Description
Keywords
Ground Penetrating Radar, IED Detection, Mine detection, Improvised Explosive Device (IED), Probability of Detection (Pd)