Abstract

Landmines pose a critical threat to human and animal lives in post-conflict regions and continue to hinder economic recovery by affecting agriculture and infrastructure development. The process of discovering and clearing ordnance involving landmines using human or animal forces presents extreme risks and demands a considerable labour force and time. Landmines are made from a variety of materials, including wood, glass, metal, and plastic, and they vary in size. To improve detection accuracy, multiple sensors with different capabilities can be used simultaneously, allowing for data fusion and more informed decision-making. The use of uninhabited aerial vehicles (UAVs) equipped with diverse remote sensing technologies offers a safe and efficient means of accelerating humanitarian demining operations. In this study, two integrated remote sensing modalities - Ground-Penetrating Radar (GPR) and magnetometers - are mounted on an advanced autonomous UAV to enhance detection probability and reduce false alarm rates. A custom Android-based tablet application is used to analyse the fused data in real time. The performance of the individual sensing modalities was evaluated through field tests conducted in Latvia, Croatia, and Cambodia. The complete integrated system, incorporating sensor data fusion, was tested in a landmine field in the UK. Results from these outdoor trials confirm the effectiveness of the proposed fusion techniques in detecting legacy landmines, Unexploded Ordnance (UXO) and Improvised Explosive Devices (IEDs) with high accuracy.