Destroy a swarm of drones automatically: a laser weapon of the US Navy will get a function of AI

This is a very time-consuming job for a man-operator, especially when it comes to an attack of a whole switch. In view of this, naval postgraduate researchers (NPS), the US Navy Navy Center in Dalgren, Lockheed Martin, Boeing and the US Air Force Research Laboratory have decided to automate the critical parts of the laser weapon tracking system.

By improving the classification of goals, assessing the position, choosing a aiming point and supporting the aiming point, ship laser systems will be able to evaluate and neutralize drones faster and more accurately. In order to reach this level of automation, researchers generated two data sets that contained thousands of images from drones and then applied to these sets of training data. A series of three scenarios was proposed to teach the system.

The first used only synthetic data, the second - a combination of synthetic and real data, and the third - only real data. According to the US Navy, the third scenario was the most effective and had the slightest error. The next step will be field tests using radar and optical support for real goals with the help of a semi-autonomous system, in which the man-man will control some aspects of support. "We now have a real -time model within our tracking system.

This calendar year we plan to demonstrate an automatic selection of aiming at a tracking structure to simply confirm the concept Montag. Recall that scientists at the University of Geriot-Wotta in the United Kingdom have developed a new Lidar system that can "see" through fog and smoke, identifying hidden objects and giving advanced remote sensing opportunities.