Drone Swarm Detection Using Artificial Intelligence Based on Ultrafast Neural Networks

R-DEX Systems, Inc. (R-DEX), in collaboration with its strategic partner, proposes to leverage its extensive expertise in Artificial Intelligence/Machine Learning (AI/ML), Radio-Frequency (RF) simulation, RF processing, real-time sensor processing, automatic target recognition (ATR), target tracking, and neuromorphic software and hardware to develop an innovative solution for detecting multiple drones using ultrafast neural networks with extremely low power consumption. The proposed solution will capitalize on R-DEX's vast experience in RF simulations and our partner’s cutting-edge neuromorphic technology.In Phase 1, R-DEX plans to employ 3D simulation techniques, modeling the blades drones, helicopters, and wind of birds as polygons to simulate radar scatter based on their respective locations in space relative to the simulated transmitted signal, thereby computing micro-Doppler signatures for the identification of the scene. These micro-Doppler signatures will serve as input for an ultrafast spiking neural network developed by our partners. Together, R-DEX and its strategic partner will investigate various methods for detecting multiple drones within a swarm while distinguishing them from non-drone objects. Additionally, the project will explore different approaches for constructing effective spiking neural networks and assess the potential of using Evolutionary Optimization for Neuromorphic Systems (EONS) to develop a highly efficient minimal size network.Specific attention will be paid to 1) computing and energy resources, 2) speed of the network, 3) accuracy of detection and percentage of false alarm.