Dusseldorf Combat aircraft are to be accompanied by swarms of drones in the future. The Bundeswehr needs artificial intelligence (AI) for this. The Bundeswehr Planning Office has now commissioned Airbus Defense and Space with a second study on this.
As the Handelsblatt learned exclusively, the Munich-based company wants to test special swarm algorithms together with two start-ups.
Swarm technology gains military importance. China, the US and Russia have long been working on large swarms of drones. In the future, these could attack from the air, under water or on land and are difficult to fend off. In addition, as a networked system, they would be able to quickly reconnoiter a large area. The more autonomous the swarms become, the more powerful they become.
China in particular is driving the development forward. The People’s Republic had already shown in 2017 that it can coordinate more than a thousand smart drones in a swarm. “There are signs that China is likely to be ahead of the West in this technology,” says Lorenz Meier. He is co-founder of Auterion, a Swiss specialist in drone software. “From a Western perspective, we therefore have no choice as to whether we look at swarm technologies: we have to face the problem at least at the research level,” says Lorenz Meier.
Swarms of drones are also said to play an important role in the key future programs of the European armaments industry, such as the Future Combat Air System (FCAS). The next-generation fighter jet is to act in conjunction with drones that take on certain tasks, for example they could be sent forward for reconnaissance. Losing a drone would not be as expensive as losing the fighter jet. The systems should be able to be controlled by pilots in the fighter jet.
Drones need to become more autonomous so that fighter jet pilots can pilot them
But as long as a person has to determine which drone should do what, that seems unthinkable. Fighter jet pilots already have to monitor many systems in parallel today. You will hardly be able to carry several drones in addition. Therefore, the swarms must become largely autonomous.
The new study is therefore intended to test how swarms of drones can coordinate themselves with the help of AI. In a first study – called KITU 1 – Airbus had developed a theoretical concept for this. The acronym stands for “Artificial Intelligence for Tactical Unmanned Aerial Systems”. The next step is practical tests.
It is not possible for jet pilots to control drones during flight.
(Photo: Reuters)
“Research projects such as KITU 2 are significant with regard to the realization of large European defense programs such as the ‘Future Combat Air System’, in which aspects such as drone swarms, AI and cloud computing play an important role,” says Michael Schöllhorn, CEO of Airbus Defence and Space.
For KITU 2, Airbus Defense and Space is teaming up with two start-ups. The Munich drone company Quantum Systems supplies reconnaissance drones to the Ukraine. The Thuringian start-up Spleenlab has specialized in machine learning and autonomous systems in mobility and is already working closely with Quantum Systems.
AI: Drones should be able to find enemy vehicles
What the three partners want to show is called “order-based leadership.” Quantum Systems founder Florian Seibel explains: Instead of manually controlling individual drones, the swarm should be given a goal. “A task for the drone swarm could be: Find all enemy vehicles,” he says. The system would then have to decide for itself which drone to fly in which area and which sensors to use.
>> Now also read: drones from Boeing could soon deceive, refuel – and kill
You can already see in North America that this can work. “The USA are three years ahead of Germany,” says Auterion co-founder Meier. With the US manufacturer Skydio, his company is already implementing the order-based guidance of drones for the US Department of Defense. In the US, its software also runs on Quantum Systems drones.
Many experts see the main danger of drone swarms in the fact that they are relatively easy to replicate. If you have the necessary software, you could order micro-drones online and arm them. Another fear is that a swarm of drones could spiral out of control.
In the US there are calls for regulation of drone swarms
“The Americans see a need for regulation for swarm technology, and I think that’s necessary and sensible,” says Meier. “A human should be able to see and confirm the plan before it is executed.” In addition, drone swarms should not be given general assignments, but rather tasks that are clearly defined in terms of time, place and means of deployment – such as armed and unarmed drones.
Such scenarios are still a long way off in the Bundeswehr. The next phase of the study is intended to show that autonomous coordination is also possible in practice. The companies want to split the order as follows:
-
Airbus Defense and Space wants to include swarm algorithms, simulation programs and experimental hardware in the test.
-
Quantum Systems provides two different types of drones. The Vector is already in use in Ukraine and, according to the company, has a range of three hours and 30 kilometers. The multicopter Scorpion flies lower. It is designed, for example, to accompany military convoys and could also be supplied with power via a mobile cable.
-
Spleenlab is to contribute expertise in the field of edge computing: military systems must be able to process data locally with AI. Electronic interference can prevent opponents from sending large amounts of data back and forth.
The test is to build on the results of the first study phase. Since only Airbus Defense and Space knows these results, the company would probably have been awarded the contract on its own. Schöllhorn says about the cooperation with the start-ups: “Cooperations based on such studies provide important impetus to build an ecosystem in Germany with which we can strengthen our expertise in future technologies and contribute specifically to the defense capability of Europe.”
More: Avilus Rescue Drone: This drone is designed to rescue wounded soldiers
First publication: 07/03/2023, 10:00 a.m.
