Central Research Institute Cyclone: Neural network for detecting drones

2024: Product Announcement

On July 11, 2024, the Roselectronics holding, part of the Rostec State Corporation, announced the launch of a new neural network capable of increasing the detection range of unmanned aerial vehicles (UAVs) by 40%. The development is designed to automate the operation of optical drone detection facilities monitoring airspace around protected facilities.

According to the press service of Rostec, the innovative technology was created by specialists from the Central Research Institute "Cyclone," which is part of Roselectronics. The neural network uses advanced methods of computer vision and artificial intelligence for early detection of aircraft, their classification and, if a threat is identified, prompt informing the operator to take measures to neutralize unmanned aerial vehicles.

Mil.ru

A feature of the development is the use of an ensemble of several optimized neural networks, which made it possible to achieve a synergistic effect and significantly increase the efficiency of the system. The technology also provides the ability to transfer the anti-drone complex to a completely autonomous mode of operation.

Yury Koval, technical director of the Central Research Institute "Cyclone," noted the high potential of the created neural network for further development in the field of security. According to him, the developed solution demonstrates a significant advantage over similar IT products, increasing the range of unmanned aerial vehicle detection systems by about 40%.

The development of the neural network was carried out within the framework of the large-scale hackathon "Leaders of Digital Transformation," aimed at creating innovative digital solutions for government agencies, business and regions. The team of the Central Research Institute "Cyclone" successfully coped with the task of creating an air object detector capable of detecting and classifying flying objects for a potential threat. The project was highly appreciated by the expert community and was awarded a prize following the results of the competition.^[1]

Notes