The duty author | Utada
| Tiger Sniff Technology Group
The header image is from Twitter
A 7-minute video uploaded in November 2017 by Stuart Russell, a computer vision professor at the University of California, Berkeley, is now more of an inevitable "military technology development prophecy."
The virtual video, made entirely by the creative team, is very similar in style to Black Mirror: a drone device smaller than a fist, all equipped with a facial recognition system and concentrated explosives, programmed in the background to find and kill specific targets.
This "drone" is called a "slaughter robot".
The GIF is from the video "Slaughter Robot" produced by Stuart Russell's team: after scanning a face, the drone finds a specific target and accurately hits the target's forehead.
And now, in this 2022 Russian-Ukrainian war, drones that claim to be targeted through artificial intelligence have appeared on social platforms and military experts' analysis reports.
For example, several "suicide" drones produced by the subsidiary of Russian arms supplier Kalashnikov, KUB-BLA (pictured below), were shot down or crashed unidentified at the Federal Savings Bank near Kiev, Ukraine.
The drone was shot down, and the green column was a battery
This white drone is actually more like a "cruise missile" (using existing weapons to launch and patrol and fly in the target area is the product of the organic combination of drone technology and ammunition technology). According to public information, it is capable of deliberately hitting a specific target within 30 minutes at speeds of up to 130 kilometers per hour and detonating 3 kilograms of explosives it carries at the same time.
KUB-BLA was shown to the outside world for the first time at the 2019 Russian Air Show. The reason why it has attracted the attention of the industry is precisely because the manufacturer claims that "it can do real-time intelligent detection, active identification and classification of objects".
Therefore, media outlets including Wired speculate that artificial intelligence may play an extremely important role in the process of "making deadly decisions".
It is believed that the drone crashed on the road without exploding, the image is from Twitter
In fact, these "scrapped" drones for no reason have aroused widespread discussion among foreign users. Because it obviously did not succeed in detonating explosives.
Therefore, some netizens commented that "Russian technology is not very good, very cheap"; others pointed out that "since these drones only explode when they attack a specific target, they will not detonate the bomb if they are shot down in the middle or do not reach the destination for other reasons, which is worth learning from the United States (below). ”
Comments from Twitter users
In fact, so far, there is no clear evidence that Russia is widely using such drones that integrate artificial intelligence technology, but the emergence of these drones has indeed raised concerns among military experts and the public. Wired quoted Zachary Kallenborn, a researcher at the National Coalition of Terrorism Research Institute, as saying what he said:
"The concept of killing robots – in particular the 'fusion of artificial intelligence and weapons'. Such technologies already exist and are being used. ”
So the question is, why these drones will crash or be shot down without completing the mission, a technical expert told Tiger Sniff, the technology has not yet developed in the harsh and complex battlefield, to complete the order completely independently at any time.
Turkey's TB2 drone
Interestingly, one of Ukraine's counter-attack weapons, also has drones.
Among them, the Bayraktar TB2 UAV designed and manufactured by Turkey, although it looks extremely bulky, slow to fly and has little defense capability, it has played an extremely important role in the battle.
In the two weeks since the Russian-Ukrainian conflict, the relatively bulky Bird has been launching several attacks on Russian missile launchers, tanks and supply trains; it has also been used to reconnoiter ground forces and search for targets.
Even NATO's high-end control systems, which can shoot down fighters, bombers and ballistic missiles, cannot be defended against such slow-flying drones.
The Wall Street Journal quoted Arthur Holland Michel, a senior fellow at the Carnegie Council, as saying that because "drones are very small, maneuverable, and fly at low altitudes," it is difficult to identify and track them; many of the "anti-drone systems" deployed today have relatively short ranges to protect specific facilities.
A light anti-tank missile that requires shoulder-to-shoulder NLAW
In addition to TB2, among the various weapons that the United States and other NATO countries have continuously sent to Ukraine, including the "short-range anti-aircraft missile Javelin", "light anti-tank missile NLAW", "drone" and other weapons, a killing drone called "Switchblade" (unmanned bomber) is about to be sent to the Ukrainian battlefield.
According to official information, depending on the model, it is capable of accurately striking people, tanks and other armored vehicles from miles away.
In other words, it should be called a "disposable robot smart bomb" equipped with cameras, navigation systems and explosives, which can be programmed to automatically hit targets, and can even be manipulated to bypass the target and strike at a suitable point in time.
This sounds very similar in function to Russia's KUB-BLA drone. In addition, the "spring knife" has been secretly sent to Afghanistan and Iraq since 2010, and has been in service on the battlefield for more than 10 years, and should be the most widely used drone weapon.
Spring knife 600, can accurately hit the tank
Despite Russia's high-level secret weapons, it looks menacing. In fact, from the current point of view, there is no evidence that Russia is deploying "artificially intelligently controlled drones" on a large scale. In addition, their combat performance in air combat has raised some questions from military experts.
But even so, through this war, the growing relationship and impact between artificial intelligence and sensors, drones and weapons, as well as a series of chain reactions in the military and technology industries, have once again been confirmed on a global scale.
Military thinking subverted by new technologies
In fact, the source of global attention from the application of artificial intelligence in the military field should go back 8 years.
In 2014, the Computer Systems and Logistics Information System (ALIS) of the U.S. F-35 fighter jet was reported to be loaded with artificial intelligence technology. This technology can help the "brain" to make assessments, check lists, organize information, and make automated decisions.
Then in 2016, senior U.S. Air Force officials publicly issued a statement that the F-22, F-35 and other fighters would be equipped with the latest artificial intelligence technology: "Soon, these fighters will control nearby wingmen who may carry weapons through AI systems." ”
In addition, they can test enemy air defense measures and conduct intelligence, surveillance and reconnaissance missions in high-risk areas. Zacharias, a former chief scientist in the U.S. Air Force, predicted at the time that in the future, pilots could directly maneuver more than one drone in the cockpit of a high-end fighter jet like the F35 or F-22.
"As these aircraft become more automated and smarter, they will become very powerful in combat; and pilots will change from executors to managers of drone teams."
F35 fighter
In 2019, we traveled to Switzerland, the "land of drones", to visit top university laboratories and drone technology companies there. Among them, the Laboratory of Unmanned Aerial Vehicles and Automation Systems (ASL) of the Federal University of Technology (ETH) zurich, led by Roland Siegwart, a professional expert in robotics and automation, has long been applying artificial intelligence, new sensors and other technologies to drones.
In addition, they have received R&D demands from the aerospace and automotive industries for autonomous driving systems, computer vision and 3D mapping.
"In fact, from the perspective of the necessary technologies of drones, many of them are common to self-driving cars and even mobile phones." This is a technical expert who plays an important role in the Swiss drone and robotics industry,
"The basic technologies needed for cameras, radar, mapping, and localization are essentially the same."
He told us at the time that cheapening, softwareization, precision and miniaturization are indeed a clear trend in the application of the drone industry. Many things that are not new, such as the parts in mobile phones, once applied to different carriers, will collide with completely different sparks.
As recently as October 2021, ETH lab researchers have given drones the ability to "automatically avoid obstacles and traverse the woods." Relying solely on-board sensing and computing, they allow drones to fly autonomously at high speeds in unknown and complex environments, such as construction sites and dense jungle environments.
"This application is not limited to quadcopter machines," the researchers note, noting that the same approach could also help improve the performance of self-driving cars, or perhaps even open the door to the even less likely — for AI systems to operate in areas where it's impossible to collect data.
Drones navigate the forest independently at a speed of 40 kilometers per hour
The increasingly important role of the computer in war also means that the communication systems it controls, sensors such as radar, and weapons such as drones are all "outlets" in a highly networked war – data that strengthens their interaction with each other while also throwing bombs harder, more accurately, and farther away.
Two very obvious examples.
In September 2020, Azerbaijan launched a war against neighboring Armenia, a brutal war that cost 7,000 people, allowing Azerbaijan to seize three-quarters of the area controlled by Armenian forces in just 44 days.
A key point that led to this outcome was the "drone force".
Azerbaijan purchased TB2 drones and Harop (Harrop suicide drones) from Turkey and Israel, respectively. The latter is smaller, more stealthy, more automated, and specializes in launching suicide attacks against radar systems. The result of their contribution was the destruction of more than 20 air defense systems, hundreds of armored vehicles and countless artillery.
In addition, in the second Libyan civil war launched in 2020, according to a United Nations Security Council analysis of the war, one of the three forces in Libya, the "Government of National Unity", won a decisive victory in the western region after receiving reinforcements and weapons support from Turkey.
Among them, from the weapons side, turkey-supplied drones and cruise missiles have some "new battlefield technology capabilities", and the programmed system really has the ability to "fire, forget and find" (below).
I have to say that cheap and difficult can also give birth to the most suitable technical force. Cheap and common hardware and technology, spreading from the mass commercial market to the battlefield, has become the key to victory.
Image from a Security Council document
The Economist notes that highly networked warfare has been studied for decades. Among them, the concept of a modern "kill chain" (transmitted from a series of sensors to a series of shooters) is generally believed to have originated in the 1970s.
In this networked system, for example, a fighter might transmit data on targets it cannot directly attack to a carrier-based cruise missile capable of launching the attack.
Just like playing chess to find the best path, after the drone fleet becomes huge, the "platooning" between systems and systems, and between drones and drones, is one of the jobs that can make artificial intelligence try.
The U.S. Defense Advanced Research Projects Agency (DARPA) noted in 2020 that while real air combat is extremely chaotic, it has "a clear goal, measurable results, and the physical limitations inherent in aircraft dynamics," so that "air combat" can be a testable case for advanced tactical automation.
In other words, since the laws of physics and aerodynamics are consistent, computers have the ability to handle most flight situations.
"This is bound to be part of future air operations." Human pilots work with swarms of combat drones, with pilots issuing basic instructions to the AI fleet, who takes care of the details. ”
Behind all this is the fact that machines react much faster than humans.
Aim under the drone perspective
Artificial intelligence and sensor technology will not only give a different war outcome, but also bring about a subversive transformation of traditional military combat tactics.
"In military tactics, the shift to a transparent combat space may lead to a re-examination of basic military principles. The tactics of "raids" that have been used since the time of Sun Tzu will be more difficult to implement on a large scale. Attacks like Egypt's attack on Israel in October 1973 will be almost impossible. ”
The Economist notes that officers have been training to emphasize the importance of rallying forces to concentrate firepower, and now they will learn to fight in smaller, more dispersed units.
However, just like in the opening video "Killing Robot", where humans are invisible in a world full of "sky eyes", how to hide themselves in a war space full of surveillance instruments, radar and data?
While this will become more difficult, there will always be "stronger defense systems that need to be iteratively upgraded."