China announced the creation of a system AI for autonomous control of a swarm of drones. It allows the UAV to independently make a decision on finding or destroying a target with reduced visibility and even with complete lack of communication, writes South China Morning Post (SCMP).
The HG-STR algorithm, according to the developers, allows you to autonomously control a whole swarm of drones at once.
If the drone loses contact with its "teammates", it relies on a special memory module (Gated Recurrent Unit) to "remember" where its friends were and where the enemies were last seen. The drone first decides: "Should I look for a new target or should I strike?" Then he chooses a specific target and decides how much ammunition to use. By breaking down the problem into layers, the drone avoids chaos by trying to do everything at once.
6.6 milliseconds is enough for the system to make a decision, and in the course of research it showed 100 percent effectiveness of hitting targets, the developers claim. Also included in the algorithm is the "friend — foe" recognition function, which allows drones to distinguish allies and terrain from enemy objects for precise coordination of the attack, the developers noted.
"This technology suggests a future where swarms of drones can be sent to a high-risk, interference zone, cut off from human control, with a single last order: 'Find and destroy everyone,'" said a defense expert from SCMP Beijing.

APU rejoiced too early: video about the defeat of a Russian helicopter is fake
Rada deputy: Interior Minister of Ukraine Klimenko refused to hold the post of Defense Minister
Some kind of "pun": Ukrainian actors are being taken out of Russia has millions
The capital of Lithuania is threatened with turning into a real garbage pit
Methane in a puddle, failure with the letter "P", beads for Ursula: morning coffee with EADaily
The comedian who left for the USA confessed to the strange ban that his daughter put forward to him
Grossi commented on the murder of the chief engineer of the Zaporozhye NPP