Chinese military researchers claimed to have conducted a dogfight between a human and an AI-controlled UAV in real-life conditions, in which the AI defeated the human opponent.
A research team led by Professor Huang Juntao of the China Aerodynamics Research and Development Center, a research institute of the People’s Liberation Army in southwest Sichuan province, published an article Feb. 27 in the peer-reviewed Chinese journal Acta Aeronautica et Astronautica Sinica. He stated that the AI had shown superior performance in state-of-the-art hand-to-hand combat.
The dogfight involved two small fixed-wing unmanned aerial vehicles, one piloted by an AI, while the other was remotely piloted by a human pilot on the ground.
As the battle began, the human took the first step to gain an advantage over the AI-controlled drone. However, the AI reportedly predicted his intent and managed to outmaneuver the opponent by making a countermove and staying close behind the opponent.
The human opponent tried to dive into the plane to lure the AI to crash. However, the latter reportedly went into an ambush position and waited for the enemy to advance.
China’s AI-controlled aircraft (in the red circle) defeated another remote-controlled aircraft operated by a human pilot in close combat. (China Aerodynamics Research and Development Center, Mianyang)
After that, the human pilot performed another tactic called “rolling scissors,” in which a plane suddenly slows down and changes course to trick the opponent into overshooting the target in the chase.
Likewise, the fight lasted about 90 seconds, after which it was canceled because the human pilot was unable to avoid the AI-controlled aircraft.
Artificial intelligence as the king of air combat?
“The era of air combat where artificial intelligence will be king is already emerging,” Huang and his colleagues said in the paper, noting that aircraft with autonomous decision-making capabilities “can completely surpass humans in terms of reaction speed.”
According to Huang’s team, this was also because the AI didn’t have to take human concerns into account during sharp cornering, such as blood loss from the brain due to excessive gravitational force or concerns about being injured.
“With superior calculation ability, it can more accurately predict the development of the battle to take the initiative in the confrontation,” said Huang’s team.
However, Huang’s team finds that getting AI into the air is very difficult compared to using it on the ground because the computing resources on an airplane are limited, which can severely affect the AI pilot’s performance.
The researchers explain that actual real-world conditions are more complex and unpredictable than those generated by mathematical models in a ground simulator. Engineers must also consider the costs and risks of an accident.
“The problem of aircraft air combat confrontation is very complex, with high dynamics, strong real-time characteristics, and a larger solution space,” said Huang’s team. “This poses a major challenge for the real implementation of intelligent decision-making.”
The researchers said there were few reports of taking the technology from theory to practice, and they built the AI air fighter pilot to make it easier for the Chinese military to use.
For example, the AI decision-making process does not depend on the hardware of a specific aircraft, meaning the AI pilot can operate on almost any fighter aircraft in the People’s Liberation Army (PLA) inventory.
US-made chips used in the development of Chinese artificial intelligence
In particular, the report on the dogfight between Chinese AI and human-piloted drones comes after the US military recently announced that a modified F-16 fighter jet has completed an artificial intelligence-guided test flight for the first time in history.
According to Huang’s team, research into pilot AI technology began around 60 years ago in the United States. However, China quickly caught up with the US even after a late start.
In 2020, a deep-learning AI system developed by Maryland-based company Heron Systems defeated some of the most experienced F-16 pilots in all five dogfights on a ground-based simulator. And within a year after that, Huang’s team claims to have achieved similar victories in simulations using AI that used a fraction of the computing resources used by its US opponent.
VISTA X-62 – modified F-16 used by the US Defense Advanced Research Projects Agency for an artificial intelligence guided test flight.
Ironically, Huang’s team used an American-built chip to build their AI pilot. The crew installed the Nvidia Jetson TX2 module as the AI brain on the plane.
The US government has banned California-based Nvidia Corp from exporting its most powerful computer chips for working with artificial intelligence to China. However, the Jetson series chips produced in China were not on the list.
The Nvidia chip is said to have enabled the AI pilot to make 1,000 decisions per second based on data collected from sensors onboard the plane or from the ground.
Are dogfights still relevant now?
What remains questionable, however, is the relevance of Huang’s team’s AI pilot and his simulated air combat to the current reality of air battles, where experts argue that combat missions are within visual range (WVR), or what is commonly referred to as “air combat”. , have taken place a back seat.
Air Marshal Anil Chopra (retired) of the Indian Air Force (IAF), who is currently Director General of the Center for Air Power Studies (CAPS) in New Delhi, recently wrote an article for the EurAsian Times examining air combat mission statistics, which showed that dogfights have lost relevance in modern times.
Similarly, a United States Air Force (USAF) study analyzing over 1,450 air-to-air sorties since 1965 also found that long-range weaponry and sensors have dramatically reduced instances of dogfights.
Huang’s team said in the paper that the proliferation of stealth and electronic warfare technologies has made air combat relevant, warranting continued research into hand-to-hand combat.
“Due to the advancement of stealth and electronic countermeasures technology, 25 to 40 percent of dogfights are at close range. The research on hand-to-hand combat has significant value in practice,” said Huang’s team.
In his article, however, Chopra notes that “both Russia and China are developing low-frequency radars to enable detection of stealth platforms,” meaning long-range air-to-air weapons aren’t going away anytime soon.
Also, EurAsian Times has discussed in the past how Russian-made radars like the AESA 1L119 NEBO SVU 3D VHF radar may be able to detect very poorly observable targets like the US-made F-35 from up to 152 kilometers away could.
This is a significant distance as it could allow a fighter like the Su-35 to conduct a targeted search and lock onto the American fifth-generation aircraft to launch its long-range air-to-air missiles like the Russian Vympel R-37M .