Chinese air force applying artificial intelligence to air combat training

by Andrew Tate Jun 17, 2021, 08:10 AM

The adoption of artificial intelligence (AI)-based technology in pilot training by China's People's Liberation Army Air Force (PLAAF) has progressed to use in simulator...

The adoption of artificial intelligence (AI)-based technology in pilot training by China's People's Liberation Army Air Force (PLAAF) has progressed to use in simulator training, according to a 12 June report published in the PLA Daily newspaper.

The paper wrote that pilot Fang Guoyu, who was described as a “group leader in a brigade under the PLA Central Theatre Command Air Force”, was pitted in the simulator against an opponent driven by an AI-equipped system.

Fang, an experienced pilot and the winner of an “air combat assessment exercise”, was initially able to win against the AI opponent, but the ability of the AI system to learn from experience made it progressively harder for the pilot, who ultimately was unable to defeat the machine, according to the paper.

In November 2020 Janes reported that the designer of China's Hongdu JL-10 advanced jet trainer anticipated that AI would be applied to monitor the performance of trainee pilots in the air and could also be used to combine real-world and simulated training.

The recent progress, which according to the PLA Daily represents “the latest advances in simulation training” for the PLAAF, is the result of development work by pilots from an undisclosed brigade under the Central Theatre Command as well as from unnamed scientific research institutes.

The PLA Daily report gave no details about the type of scenarios involved in the work but the context suggests that they were primarily close-range air combat manoeuvres. If so, the weapons simulated may have been short-range air-to-air missiles, such as the PL-10, or may have been limited to engagement with cannons. Among the PLAAF's current combat aircraft, only the fifth-generation J-20 is not equipped with a cannon.

Already a Janes subscriber? Read the full article via the Client Login
Interested in subscribing, see What we do