A swarm of small, fast, cockpitless fighter jets are flying in a tight circle around an F-22 Raptor, reacting to its every move, waiting for a command. They will scout ahead, attack or sacrifice themselves if needed, relying on their superhuman reaction time and precision to execute manoeuvres that human pilots would never manage to do. This is the way many nations envision the air combat of the future. But why?
There are several problems with the current generation of combat aircraft. First of all, they are incredibly expensive. One F-35 costs between $77 million and $102 million, that is almost 200 times more than state-of-the-art P-51 Mustang cost in 1945, adjusting its price to modern dollars. Even “cheap” stealth fighters like J-20 or Su-57 are estimated to have a price tag of at least $50 million, not including stupendous operating costs and program expenditures in their billions.
The second problem is the increasing cost of human (and especially pilot) life. Economically speaking, a life often has actual (although a bit vague and fluctuating) dollar rate attached to it – meaning, how much a nation is willing to pay to avoid the death of one of its citizens. Not only is that cost steadily rising, but political consequences of losing a soldier are becoming more and more acute for many governments, especially democratic ones, and the ones pretending to be that way. On top of that, pilot training costs are on the rise too, and creeping into the tens of millions for the newest stealth fighters, that require exceptional skills to operate.
The answer to both of these problems is obvious: drones. As many conflicts show, they are excellent in situations where reconnaissance or ground attack is needed, mostly due to long endurance capabilities. Their pilots can take more risks too – that is why, for example, the United States are losing their UAVs in rows performing missions in hostile environments and not even batting an eye.
But there is a difference between long-endurance missions, controlled by remotely located operators, and real combat. Coordinated ground attack and support, even more – air to air combat – pose a number of challenges for current drones.
One of those challenges lies in ethics, as many national and international restrictions require a human to take a decision to kill. Another one is a risk of electronic attacks, connection issues, and other communication interference. Lastly, most of current drones – even the stealth ones – are not exactly built for pitched combat. Pretty much anything that flies can carry air-to-air missiles, but attaching a Sidewinder to MQ-9 Reaper does not make a slow, clunky drone a fighter.
Nevertheless, combat drones can not only cut costs and save pilot lives, they also have all the prerequisites to be better than humans at fighting. They have much faster reaction time than even the best pilots and can execute complex procedures with unrivalled precision. Complicated tasks, such as air combat, rely on both of those skills. Psychology also plays a part: pilots can be scared, intimidated, or get sloppy after not practising for several months; computers don’t. Add an ability to endure G-forces that would crush a human into a pancake, and it comes as no surprise that artificial intelligence (AI) will easily beat the best fighter pilots in a dogfight.
So, the ideal here is to have combat-capable drones controlled by an advanced AI with a pilot somewhere nearby overseeing and directing their actions to avoid mistakes in judgement and legal issues. UAVs have to function along manned fighters, in one system with them, utilizing the computer’s reaction time and precision, but relying on humans for judgement.
That is exactly the premise of a concept nicknamed “loyal wingman”: one of the trending buzzwords in the world of military aviation.
Teaming up systems
The first steps towards the concept were made in 2014, when the Dassault nEUROn stealth drone performed a much-publicized formation flight with a Rafale fighter and a Falcon 7X business jet. Although the flight was human-controlled, one of the aims of the nEUROn program, launched in early 2000s, is to test algorithms necessary for autonomous combat performance and the place of human factor in the mission loop – that is, to develop a framework for actions of AI-controlled drones in a combat environment.
The task is difficult. If the concept of “loyal wingman” is to be realized in full potential, we are talking not only about cooperation between numerous drones and their ability to take initiative, perform attacks, and react to threats, but also to rely on one pilot for approval of decisions and control in critical situations. An algorithm for multiple machines performing in a team with a human has to be developed.
In 2018, Airbus announced that they had successfully tested just that – a manned-unmanned teaming technology, where five target drones performed as a team while being controlled by an airborne mission group commander. Special human-machine interface, teaming and swarming algorithms, and mission management systems were demonstrated, allowing for a ballet of jet-powered drones in the air over the Baltic Sea. It was a demonstration of a “system of systems”, another buzzword, meaning the ability of different systems to pool their strenghts to reach a performance exceeding the sum of its parts.
Although the mission of the “loyal wingman” software is rather clear, there are several possible approaches to hardware. The most obvious one is to develop unmanned versions of true combat jets, something a number of 5++ generation and 6th generation fighter projects will supposedly have. Both the Mikoyan MiG-41 program and the (seemingly discontinued) F/A-XX program came with promises to have unmanned versions for especially dangerous missions, a noble pursuit in theory, but still not an answer to the problem of rising aircraft costs.
Another approach is the idea to have a swarm of purpose-built disposable drones. Airbus’ test was part of the development of the Future Combat Air System, pan-European 6th-generation fighter project intended to capitalize on nEUROn’s research. From the ground up, it was designed as a team of the new manned fighter jet and several “remote carriers” – small and medium parasite UAVs. Their exact functions and capabilities may vary, with at least Intelligence, surveillance, target acquisition, and reconnaissance (ISTAR), as well as shielding manned aircraft from enemy fire being the priority. Something similar, lovingly nicknamed “project mosquito”, was announced by Britain – a swarm of carrier-launched drones to accompany their Eurofighter Typhoons, F-35s and Tempests.
Yet another way is to combine capabilities of both approaches by having cheaper, smaller versions of fighter jets, or rather small cheap drones with a capability to carry weapons. This kind of “loyal wingmen” is by far the most developed, or at least the most publicized at the present time.
Flight of the Valkyrie-ish
The poster child for the mixed approach is XQ-58 Valkyrie. Often described as an “unmanned fighter jet”, it is a technology demonstrator developed by U.S. aerospace industry newcomer Kratos Defence & Security Solutions and test-flown in early 2019. According to the manufacturer, the aircraft can reach high-subsonic speeds, has decent maneuverability and stealth characteristics, and can carry up to 250 kilograms of ordinance in internal weapon bays. The final production version is supposed to cost between $2 and $3 million, on par with some anti-air missiles. The company also offers even cheaper non-stealthy UTAP-22 Mako “loyal wingman” based on their aerial target drone.
XQ-58 Valkyrie (U.S. Department of Defense / Wikipedia)
In 2020, Kratos received a contract for the U.S. Air Force’s Skyborg program, intended to further develop the “loyal wingman” concept. The program has an emphasis on AI systems that could both control drones and act as co-pilots on manned aircraft. Boeing, General Atomics, and Northrop Grumman have been awarded contracts as well. While the latter ones have not announced their intended testbeds, Boeing’s bid may have been revealed before.
The Airpower Teaming System (ATS), intended as a supersonic drone with fighter-like capabilities, was developed by Boeing Australia for the Royal Australian Air Force and showcased in mid-2020, with a clear intention to perform flight tests later that year. Three prototypes are reportedly completed and one was already rolled out and paraded in front of cameras.
There is yet another American “loyal wingman” program. It was developed by Sierra Technical Services and designated Fifth Generation Aerial target, or 5GAT, its name bearing the legacy of the drone’s previous purpose. Decidedly larger and heavier than other projects, 5GAT cannibalizes parts of T-38, F-5 and F/A-18 jets in order to save costs and – at least according to the manufacturer – has somewhat of dogfighting capability, initially being designed as an aerial target.
5GAT Prototype (Photo: Sierra Technical Services)
5GAT and Valkyrie are the two prime contenders to participate in the first simulated human-versus-AI dogfight, announced by the U.S. Air Force in June 2020 and scheduled for July 2021. The showdown was initially promised to happen in late 2020, but got delayed as the focus shifted from using an AI-controlled fighter jet (most likely F-16) to a purpose-built machine for the fight.
Americans and Australians are not the only ones working on the idea, though. When the Russian Sukhoi S-70 Okhotnik was first revealed, its capability to function in conjunction with a Su-57 fighter jet was heavily emphasized. So far, its role extends only to reconnaissance and ground attack, as nothing has been said of Okhotnik’s air-to-air potential.
In August 2020, Russian company Kronshtadt, manufacturer of the Orion drone, revealed a mockup of Grom – a Valkyrie-like loyal wingman, referred to as an unmanned stealth attack aircraft. According to the company, it could both protect Russian Su-35 and Su-57 fighter jets from anti-air fire and carry ground attack and reconnaissance missions on its own, although neither production deadlines nor information on its flight characteristics or AI systems have been disclosed.
Japan is another nation which currently dabbles in the concept. They have combined European and American approaches and announced the development of Combat Support Unmanned Aircraft (CSUAs): medium-sized and large-sized combat drones made for, respectively, scouting and air-to-air combat. Both kinds of UAVs are intended to accompany Japan’s F-X, the 6th generation fighter jet about to be developed by Mitsubishi Heavy Industries. Although the design phase of the CSUA is reportedly well under way, no characteristics or visualisations have been released yet.
China is even more secretive. There is no doubt it is working on its own “loyal wingmen”, but the only breadcrumbs are offered by several publications of state newspaper Global Times, where Chinese next-generation fighter jets are described as being able to command drones. Size and capabilities of those drones will probably be unknown until one day they appear on a National Day parade, prompting even more speculation.
This article was originally published on September 5, 2020.