This paper presents an approach for explosive-landmine detection by designing, developing and integrating a Ground Penetrating Radar (GPR) onboard an autonomous aerial drone. Our goal is twofold: firstly, we have tackled the development of a custom-designed lightweight GPR by approaching an interplay between hardware and software radio. This technology is called Software-Defined Radio (SDR). Our SDR-based GPR system results on a much lighter sensing device compared against the conventional GPR systems found in the literature and with the capability of reconfiguration in real time. This means our system is capable of detecting landmines under terrains with different dielectric characteristics. Secondly, we introduce the integration of the aforementioned SDR-based GPR into an autonomous drone. To this purpose, we have developed an accurate navigation flight control system based on optical laser technology that enables the drone to operate steady at the corresponding altitude over the ground (50 cm). To this purpose, we propose a novel nonlinear model-dependent control strategy for regulating the altitude (z) and the attitude of the drone (roll phi, pitch theta, yaw psi) named backstepping+DAF. Experiments have been conducted aimed at characterizing how our SDR-based GPR system performs depending on different landmine morphology. Also, how steady flight is achieved by means of the proposed flight control system that consequently enhances the GPR performance. Results have shown that our SDR-based GPR system is able to detect buried landmines up to similar to 20 cm of depth for semi-wet terrains (humidity about 70%), however, the artifact must cover an area >60 cm(2) with at least 30% of the material made of metal. By calculating ROC data, the accuracy of the GPR is above 80% under the tested scenarios.
