Extendable Sensor-software Platform for Detection, Classification, and Tracking of UAS (DeteCT )

The rapid proliferation of unmanned aerial systems (UAS) has created significant security challenges across military, government, and civilian sectors. These systems, ranging from small commercial drones to sophisticated military-grade platforms, pose increasing threats to critical infrastructure, military operations, and public safety. Current counter-UAS (C-UAS) solutions often rely on single-sensor approaches that suffer from high false-positive rates and limited effectiveness across different operating environments. Additionally, existing systems typically require significant human intervention for decision-making and threat response, creating operational bottlenecks and increasing the risk of delayed or inadequate responses to UAS threats.Assured Information Security, Inc. (AIS), proposes Phase II of the Extendable Sensor-software Platform for Detection, Classification, and Tracking of UAS (DeteCT) effort for developing an innovative C-UAS system based on a Mixture-of-Experts (MoE) architecture that leverages specialized machine learning models for each sensor modality (acoustic, passive radar, RF, EO/IR, etc.). Each expert model will be specifically designed and optimized to exploit the unique characteristics and strengths of its respective sensor domain, enabling more accurate and reliable detection within that modality. These individual expert systems will feed into a sophisticated weighting mechanism that dynamically adjusts the contribution of each expert based on environmental conditions, sensor health, and threat characteristics. This approach will be implemented within an open software platform, allowing for continuous improvement of individual expert models and easy integration of new sensor types or detection algorithms. The system will operate autonomously, using the weighted consensus of multiple experts to make high-confidence decisions about threat classification and response, while maintaining near-zero false positive rates. By combining specialized expertise across multiple sensing domains with dynamic weighting and an open architecture, the solution will provide superior detection capabilities, robust performance in degraded conditions, and the flexibility to evolve with emerging threats.