AI-BASED UXO DETECTION USING SUAS EQUIPPED WITH A SINGLE- OR MULTI-SPECTRUM EO SENSOR

Abstract

Unexploded ordnance (UXO) poses a threat to soldiers operating in mission areas, but current UXO detection systems do not provide the required safety and efficiency to protect soldiers from this hazard. Recent technological advancements in artificial intelligence (AI) and small unmanned aerial systems (sUAS) present an opportunity to explore a novel concept for a UXO detection system. The system proposed in this study integrates a sUAS with an onboard single- or multiple-spectrum (MS) electro-optical (EO) sensor. The major contributions of this thesis include the development of an AI-based algorithm for reliable UXO detection using a Deep Learning Convolutional Neural Network, execution of experiments to validate the proposed system’s performance, and analysis of the proposed system’s feasibility. To that end, the thesis describes the development of the UXO detector for a single-spectrum sensor, followed by the development and integration of five UXO detectors for the MS sensor. The field experiment conducted using a commercial-off-the-shelf (COTS) sUAS equipped with a standard EO sensor is also described. This thesis concludes that AI-based UXO detection using a single-spectrum or MS sensor flown on a COTS sUAS is a feasible solution. The thesis also proposes the steps for further enhancement and improvement of the developed system and lays out additional test and evaluation strategies to fully test the developed capability.