Review of AI-augmented multisensor architectures for detecting and neutralizing UAV threats

Abstract

The unconventional proliferation of unmanned aerial vehicles (UAVs) has led to an urgent demand for advanced counter-unmanned aerial system (C-UAS) technologies capable of accurately detecting, classifying, and mitigating these threats. This paper offers a comprehensive overview of current detection methodologies, including radio frequency (RF) signal analysis, deep learning-based visual recognition (e.g., YOLOv5), thermal imaging, acoustic pattern classification using convolutional neural networks (CNNs), and integrated sensor systems utilizing attention mechanisms. A comparative analysis is conducted based on key performance indicators such as precision rates, mean average precision (mAP), operational range, response time, and robustness to environmental noise. These performance metrics are organized into a summarizing table for clarity. Additionally, several real-world C-UAS platforms such as DedroneTracker, AUDS, and Fortem DroneHunter are examined to illustrate approaches to full system integration. The discussion also encompasses the legal and ethical considerations, the implications of autonomous UAV swarms, and emerging trends in C-UAS strategies, especially those leveraging edge computing and cognitive modeling. The findings support the effectiveness of adaptable, modular, and interpretable counter-drone frameworks suited for dynamic and high-threat environments.