Deep learning based anti-drone technology

The development of drone technology is very rapid. From the use of the U.S. military drones to the popularization of drones in many fields such as civilian use, drones have become a new trend. It also brought with it a lot of new problems. The incidents such as drone insurance crashing fighter planes are even more alarming for people.

Current research on anti-drone technology

Establish a sound anti-UAV system, introduce deep learning to identify UAVs, detect black flying UAVs in time, and take appropriate alarm and countermeasures to achieve complete supervision of UAVs. The development direction of UAV technology.

Conventional methods of countering drone threats include destroying launch platforms, camouflaging fraud, electromagnetic interference, and direct fire strikes. If UAVs perform their missions alone or only in a few formations, conventional countermeasures should be used. However, when a large fleet of unmanned drones using colony tactics struck, the reaction time left for combat personnel and systems was extremely short. Conventional air defense weapon systems are the most commonly used anti-UAV weapons and can be deployed on space-based, sea-based and land-based platforms. But these weapons are an overkill for micro-drones and there is a great deal of cost asymmetry. And these systems are bulky and unable to withstand the invasion of small, inexpensive drone clusters.

Through exploration and research, and Puwei as independent research and development of low-altitude early-warning tracking photoelectric system and laser identification of anti-UAV systems. It is designed for early warning and pre-prevention of low-altitude dangerous aircraft in cities, especially the more widely used currently, and there are special security products developed for the monitoring and early warning of potentially dangerous low-cost micro-drones. Both are based on high-definition laser imaging and infrared thermography detection and sensing technology, combined with servo photoelectric rotary table and image recognition tracking technology, based on the city's high point, a single device can cover a radius of 10km. It plays a crucial role in the safe and healthy development of the city.

The system integrates advanced DOE optical infrared thermal image point target tracking detection technology, high-definition laser scanning surface target image recognition algorithm technology, and million micropulse high-precision servo-driven photoelectric turntable technology, making it possible to download monitoring in the normal mode, while Low-flying, low-velocity small drones perform functions such as detection, classification, and tracking. It can carry out anti-drone missions by autonomous scanning, radar joint detection, manual scanning and detection in multiple ways. The distance to the micro-drone can reach 2 km, and the detection distance for light aircraft in the 720-degree space can reach 10 km. The micro-drone can reach up to 2km, tracking angle speed up to 200 degrees/second, and can capture micro-drones tracking 30 meters/second and supersonic airplanes above 340 meters/second. Exceeds the 1km level of the UK's anti-nuclear aircraft defense system (AUDS), reaching the international advanced level.

The development prospects of anti-drone technology

The current technical methods still have limitations, lack of flexibility, and accuracy must be improved. Therefore, a deep-learning-based UAV identification algorithm is the key to the future development of anti-UAV technology, such as the establishment of a vision sensor network for image capture and information storage of UAVs; The learning network of the network has come to an efficient identification model to realize the classification between drones and non-UAVs.