In the field of military technology, artificial intelligence allowed analyzing variables such as speed, air temperature, ammunition, weight … everything that could contribute to giving an advantage when facing an enemy. What undoubted He could not do was predicting what movements a hunting pilot would do surprisingly, with the aim of avoiding who was chasing him. At least until now.
In an advance that, according to those responsible, will transform the future of air combat, Chinese scientists claim to have annulled the last tactical advantage of the pilots over artificial intelligence in air fighting: The ability to overcome algorithms by unpredictable and high intensity aerial acrobatics.
In a study published in Journal of Gun Launch & Control, a method is described that combines advanced infrared images with predictive models based in the for anticipate opponent movements by detectioneithern of subtle wing and tail movements.
It is a development that I could virtually defenseless even the most agile combat planessuch as the US F-15, according to the team of scientists from the Northwest Institute of Mechanical and Electrical Engineering, led by Lin Zhiwei.
The study addresses a critical failure in existing air combat systems: its dependence on predictions based on trajectoriesthat hinder the interpretation of sudden and non -linear maneuvers executed by human pilots. The Zhiwei team exceeded this limitation, focusing on the physical mechanics of enemy aircraft.
Through a modified Yolov8 neuronal network, the system analyzes infrared images for detect millimeter deformations on opponent control surfacessuch as the direction rudder of 1.5 meters or the two-meter elevator of the F-15, during the flight.
The interesting thing is that the Chinese team has managed to carry out this thanks to An American computer program, such as Yolov, acronym by You Only Look Once (you just look once). Its objective is the detection of objects in images or videos. It is an important component of many applications, such as autonomous vehicles, robotics and video surveillance. It is basically responsible for identifying a traffic light, a traffic signal or a pedestrian. And he does it in less than a blink.
That is precisely what allows you Identify movements as subtle as the change of direction of a tail rudder that can be found a hundred meters and thus anticipate the maneuver that will be carried out.