Recent government and industry announcements signal clear intent by both the US and the UK to press ahead with the development of a new generation of AI attack drones despite serious concerns about the development of autonomous weapons. While most details are being kept secret, it is clear from official statements, industry manoeuvring and budget commitments that these new drones are expected to be operational by the end of the decade.
The current focus is the development of drones that were previously labelled as ‘loyal wingman’ but are now being described either as ‘Collaborative Combat Aircraft (CCA) or ‘Autonomous Collaborative Platforms (ACP). As always, the nomenclature around ‘drones’ is a battlefield itself. The concept for this type of drone is for one or more to fly alongside, or in the vicinity of, a piloted military aircraft with the drones carrying out specifically designated tasks such as surveillance, electronic warfare, guiding weapons onto targets, or carrying out air-to-air or air-to-ground strikes. Rather than being directly controlled by individual on the ground such as current armed drones like the Reaper or Bayraktar, these drones will fly autonomously. According to DARPA officials (using the beloved sports metaphor) these drone will allow pilots to direct squads of unmanned aircraft “like a football coach who chooses team members and then positions them on the field to run plays.”
In May, the US Air Force issued a formal request for US defence companies to bid to build a new piloted aircraft to replace the F-22. However, equally important for the ‘Next Generation Air Dominance (NGAD)’ program is the development of new autonomous drones and a ‘combat cloud’ communication network. While the development of the drones is a covert programme, US Air Force Secretary Frank Kendall said they will be built “in parallel” to the piloted aircraft. Kendall publicly stated that the competition to develop CCA was expected to begin in Fiscal Year 2024 (note this runs from Oct 2023 to Sept 2024).
While it is planning to build around 200 of the new crewed aircraft, Kendall told reporters that the USAF is expecting to build around 1,000 of the drones. “This figure was derived from an assumed two CCAs per 200 NGAD platforms and an additional two for each of 300 F-35s for a total of 1,000,” Kendall explained. Others expect even more of these drones to be built. While the NGAD fighter aircraft itself is not expected to be operational until the 2030s, CCA’s are expected to be deployed by the end of the 2020’s.
It’s important to be aware that there will not be one type of drone built under this programme, but a range with different capabilities able to carry out different tasks. Some of them will be ‘expendable’ – that is designed for just one mission – something like the ‘one-way attack’ drones that we have seen increasing used in Ukraine and elsewhere; some will be ‘attritable’, that is designed that if they are lost in combat it would not be severely damaging, while others, described as ‘exquisite’ will be more capable and specifically designed not to be lost during combat. A number of companies have set out their proposals, with some even building prototypes and undertaking test flights.
CCA concepts/under development
|Kratos’ XQ-Valkyrie||Lockheed Martin’s Speed Racer|
|General Atomics’ Gambit||BAE Systems’ Concept 2|
Asked what the CCAs will be counted on most to do, Lt. Gen. Richard G. Moore of the USAF laid out three basic mission sets. First and foremost is “the ability to augment the combat force as shooters.” Second is “the ability to conduct electronic warfare” and the third is “the ability to be sensors in the battlespace.” According to Aviation Weekly analysis of the recent US Defense budget, the US has allocated $6.375 billion to develop CCAs up until 2028.
Separate to this development, last month US Deputy Defense Secretary Kathleen Hicks announced funding for a new initiative named Replicator which will “field attritable autonomous systems at scale of multiple thousands, in multiple domains, within the next 18-to-24 months.” Specific details of what drones will be developed under the rubric of ‘Replicator’ have yet to be revealed other than Hicks indication that Replicator will develop “platforms that are small, smart, cheap, and many.” In other words drones that are at the smaller and so-called ‘expendable’ end of the military drone spectrum.
A UK Strategy
The UK too are working behind the scenes to develop these type of drones. The MoD Minister James Cartlidge told the Defence Select Committee in June:
“Whilst the operational requirement for Reaper (current) and Protector (planned) UAS capability remains essential, the MOD recognises we must advance UAS concepts and we are developing an Autonomous Collaborative Platform (ACP) strategy, which we aim to complete by this summer.
Early operational analysis shows that ACP will increase the mass and survivability of our forces, and offer opportunities to saturate and overwhelm adversaries’ situational awareness. In the short term, our focus is on low-cost expendable ACP. In the medium term, we are looking at higher end attritable and survivable ACP. The strategy also recognises the importance of remaining in lockstep with our allies. The RAF is now a signatory to a 5-Eyes Human-Machine-Teaming agreement, to ensure that we develop a common language, and that we begin with an interoperable and integrated framework with our closest allies.”
Regular readers will remember that the MoD abruptly cancelled its Mosquito programme to develop a loyal wingman drone arguing that other systems with more capability were already being developed. Soon after, the then Chief of the Air Staff Air Chief Marshal Sir Mike Wigston, announced there would be a new process to acquire drones in this area. Little detail of what is happening has been made public except for the UK Defence Journal noticing a £42m budget line for Autonomous Collaborative Platforms in the ‘MOD Acquisition Pipeline’,
The Chief of the RAF told journalists in July that it was developing an Autonomous Collaborative Platforms Strategy in conjunction with the Royal Navy, while at DSEI in September, MoD minster James Cartlidge spoke of an “Uncrewed Systems Strategy” that would be forthcoming by the end of the year. These may or may not be the same thing.
Meanwhile, BAE Systems and QinetiQ quietly announced a “framework agreement which will see both parties collaborate in the area of autonomous uncrewed air systems.” Their press release explained: “The framework agreement will see BAE Systems and QinetiQ continue to work together with the relevant regulatory bodies to help develop the appropriate clearance processes for autonomous systems.”
AUKUS and the Ghost Bat
While the UK, as it has stated is working to develop its own autonomous “low-cost expendable” drones, it is likely that the UK will work with the US and other partners to develop the more ‘exquisite’ (i.e. expensive) type.
An important element here is the AUKUS military agreement between Australia, the UK and the US. While the main part of the agreement is helping Australia acquire a nuclear-powered submarine fleet, a second part – known as Pillar 2 – focuses on both developing and co-operating on advanced military technology including artificial intelligence and autonomy. A trial of small autonomous drones under AUKUS Pillar 2 has taken place.
Separately however, there is a new large autonomous drone, developed by Boeing Australia, called Ghost Bat. Originally given the cumbersome title ‘Airpower Teaming System’ (ATS), the drone was re-named in early 2022 after a species of predatory bat found only in Australia. The drone has been rapidly developed after its first flight in Feb 2021, with the Australian Air Force having already ordered 13 with an in-service date of 2024/2025. According to US media reports, the drone is now under active consideration by the US Air Force for its CCA programme. The drone, which is just short of 12 metres long and has a range of around 2,000 nautical mile, has an eight-and-a-half foot long, ‘snap-on, snap-off’ nose that has around 750 cubic feet of space to carry a variety of payloads which can be swapped between missions to carry out different tasks.
While it is perhaps too early to say if the drone will be acquired by the UK, it would be surprising if it is not being considered.
Controlling autonomous weapons
As industry and governments push ahead with developing autonomous drones, campaigners and experts are rushing to ensure that there are real and meaningful international controls over autonomous weapons. While we here primarily focus on the development of drones , it is crucial to be aware that the drone itself is often only the visible part of a ‘data acquisition / data processing / targeting / strike’ infrastructure in which Artificial Intelligence (AI) /Machine learning (ML) is increasingly being employed.
This autumn there will be a resolution at the UN urging the Secretary General to seek a way forward with states to prevent the profound legal, ethical, humanitarian and security problems raised by autonomy in weapon systems. Now is the time for all concerned about the development of autonomous weapons systems to push our governments to join this international control effort. More on this soon.