The increasing presence of unmanned aerial vehicles (UAVs) in urban, industrial, and critical infrastructure environments has created a need for detection systems capable of operating reliably under diverse environmental conditions. Approaches based on a single detection technique typically exhibit limitations associated with environmental noise, lighting variations, or communication signal loss, thereby affecting their identification and tracking capabilities.
This paper presents a low-cost, multisensor architecture for UAV detection by integrating data from acoustic, radio frequency (RF), and computer vision sensors. The proposed approach combines the strengths of each technique to increase system robustness and reduce the probability of false detections. Data captured by the different sensors is processed and fused to generate a unified representation of UAV activity, enabling identification even when one of the data sources experiences degradation or uncertainty.
The architecture is implemented on embedded systems, promoting local data processing and real-time operation. Acoustic detection identifies characteristic signatures produced by UAV propulsion systems, while RF monitoring facilitates the detection of communication patterns associated with UAVs. Additionally, computer vision algorithms contribute to the visual validation of detected targets.
The experimental evaluation considers various operational scenarios to analyze system performance in terms of detection accuracy, false alarm rate, response time, and computational resource consumption. Furthermore, the individual contribution of each sensing modality and the effect of the fusion strategy on overall system performance are investigated. The results allow for a comparison between single-sensor approaches and the proposed multisensor system, identifying the advantages and limitations of each configuration for real-time aerial monitoring applications.