Skyborg: the US Air Force’s Future AI Fleet
The US Air Force is flying at supersonic speeds towards an AI-enabled fleet. Under project Skyborg’s direction, future fighter jets won’t be supported by a wingman, but by an unmanned combat aerial vehicle.
Finger four has been the dominant fighter aircraft formation since the 1930s. The world’s most advanced fighter jet, the Lockheed Martin F-35, costs around $100m per jet. Four of these in formation means almost half a billion dollars of hardware in the air (not including the per hour cost of flying them). Losing just one fighter would be catastrophic for the US Air Force’s budget.
The US Air Force’s (USAF) project Skyborg aims to address this cost risk by replacing some of these expensive fighter jets with more affordable unmanned combat aerial vehicles (UCAVs) acting as unmanned wingmen.
Teaming up with drones
Under the project, the Kratos-built XQ-58 Valkyrie drone will team up with the F-35 and F-15EX, cutting the number of highly valuable fighters in the air, as well as cutting costs and risk to human life. At a cost of a few million dollars per unit, the autonomous Valkyrie drones are more easily replaceable, and could form a central role in the USAF’s air power. The F-35 is billed as a force-multiplier; when partnered with a Valkyrie it could get a new capability boost.
Skyborg program manager Ben Tran explained the significance of the program:
“There is heavy investment by our near-peer adversaries in artificial intelligence (AI) and autonomy in general. We know that when you couple autonomy and AI with systems like low-cost attritables, that can increase capability significantly and be a force multiplier for our air force. The 2023 goal line is our attempt at bringing something to bear in a relatively quick time frame to show that we can bring that kind of capability to the fight.”
With Skyborg the manned aircraft is the centre of the network, with the drones augmenting around it. Think of the fighter as Skynet and the Valkyrie UCAV as the T-100, only with added wings and less Arnold Schwarzenegger. AI will govern the autonomous wingman, reading telemetry, flight plans and weather, all the while acquiring targets and supporting the manned aircraft.
Cultural questions facing the air force
If an autonomous combat drone is to act as a wingman who pulls the trigger? The US, UK, and Russia have pushed against the UN trying to ban autonomous killing machines, which gives a clue to where the Pentagon is currently leaning.
With the adoption of autonomous systems becoming imminent, armed forces will need to confront serious ethical issues. On the one hand, it makes sense to give the drone trigger control. A pilot in an F-35 performing counter-manoeuvres to avoid an enemy fighter may not have time to pull the trigger. On the other, if an AI system accidentally fires at a civilian site who is held accountable?
The pilot, of the networked fighter, or someone higher up the staffing chain. The US Air Force has not said it plans to give the Skyborg drones control of any weapons systems, but this co uld be regarded as the natural evolution of the system in the future.
Human pilots know how to fly with other humans, operating with fundamentally similar instincts. An autonomous vehicle does not have these same instincts. This means the US Air Force will need to figure out how to train pilots to work with AI.
As Assistant Secretary of the Air Force for Acquisition, Technology and Logistics, Dr Will Roper, told Defense News:
“We need to understand when the machine will be at its best and when the human will be at its best. We will need to get the person trained to have an instinct for AI just like they have an instinct for stealth”.
Automation tends to shake up every industry it touches, and the military is no different. To deal with this inevitable disruption, the air force will need to assess how far autonomous capabilities can go in complementing the existing and proven use of manned fighters. As the technology evolves the question may evolve from how autonomy can complement fighter jets to whether it should replace them.
Breaking through technological barriers
The F-35 is still in production but due for upgrades in 2020, which could open doors for the integration of AI integrating systems such as those developed under Skyborg. Both Lockheed Martin and Boeing are already working with the military to bring AI into their fighter platforms.
Air Force Research Lab aerospace systems directorate engineer Matt Duquette explained the scope of the system: “Skyborg is a vessel for AI technologies that could range from rather simple algorithms to fly the aircraft and control them in airspace to the introduction of more complicated levels of AI to accomplish certain tasks or subtasks of the mission.”
A major issue to solve is getting the AI up to speed. Skyborg requires more advanced systems than are currently available on the market. The system not only has to be more sophisticated than any AI available now, but also has be developed for a world that understands how it works. If the AI is too predictable, it is easily to beat, rendering it useless in a combat setting.
Other investments in AI and unmanned aerial systems may help pave the way for this technology. Recently, the US Air Force tested a drone swarm system called PERDIX, which involves fighter aircraft dropping arm-length drones that then network together and form patterns using AI to disrupt enemy systems. The UK’s Dstl has also begun development of an unmanned system designed to complement its airpower in a similar style to how Skyborg will work. Dstl’s plans call for a lightweight affordable novel combat aircraft to be paired with the F-35.
While current AI cannot match the instincts of seasoned pilots, current research and investment trends across the military sector show a clear trend towards automation. With projects such as Skyborg pushing the limits of capability, it won’t be long before we see autonomous aircraft take to the skies, be it in a defensive or offensive capacity.
Ingen kommentarer:
Legg inn en kommentar
Merk: Bare medlemmer av denne bloggen kan legge inn en kommentar.