Multi-sensory system for UAVs detection using Bayesian inference

Unmanned Aerial Vehicles UAVs have revolutionized a wide range of activities and businesses, creating new opportunities for commercial and military applications. However, they also pose potential risks for terrorist and criminal activities. Therefore, researchers have been examining potential drone threats and considering how to address security and privacy concerns related to this technology. This research topic has gained significant attention in recent years due to the rapid proliferation of commercial and recreational drones, as well as the associated risks to airspace safety and security. Various detection technologies have been developed, including radar, optical, and acoustic sensing systems. However, each detector has its own limitations, such as reduced effectiveness in low-light conditions, fog, and noise. To address these limitations, we have developed a drone detection technique that utilizes multiple detectors, including visual, acoustic, and magnetic field sensors applying artificial intelligence, to compensate for their shortcomings. In this approach, each detector makes an independent decision, which may be either consistent or conflicting with the other detectors. In our study, we employed the Bayesian Inference technique to optimize decision-making in cases where there was conflict among the decisions made by the multi-sensors. We used indicators such as the Ephemeris indicator (EI) and the Acoustic ambiance indicator (AI) to generate settings to determine the degree of conflict. The drone detection process was fully automated, and the conflict decision was optimized using this approach. Our results indicated that the automatic drone detection with Bayesian inference had the best performance for drone identification in terms of accuracy, specificity, and sensitivity, as well as the highest accuracy in preventing unwanted drone interventions.