General Atomics' Master Software ‘Brain' That Will Operate Its Drone Fighters

Yahoo

15-05-2025

Controlling swarms of drones in high-end combat, soaking up and parsing gobs of surveillance data, unleashing advanced weaponry, and constantly evolving onboard systems to meet the rigours of contested environments all via one integrated software suite is becoming an achievable goal. Incorporating machine learning and artificial intelligence to augment human involvement and oversight is key to making this happen. The War Zone spoke with a subject matter expert from General Atomics Aeronautical Systems, Inc. (GA-ASI) to get an inside look at what the company is doing to usher in this potentially transformational facet of modern combat air operations.
Quadratix is General Atomics' answer to integrating various software systems and development teams from across the company into a more cohesive enterprise and for controlling unmanned systems on complex missions. It's borne from over three decades of corporate mission experience with its uncrewed aircraft, such as the MQ-9, and supporting many different types of operations around the globe.
General Atomics likens Quadratix to Adobe's Creative Cloud or Microsoft 365 in that it groups together products and capabilities that enable operators to undertake many different tasks in a single software environment.
For example, Quadratix includes the company's command and control (C2) and tactical situational awareness software known as TacSit-C2, which does its work by concurrently viewing sensor data to cross-cue airborne payloads. TacSit-C2 alone integrates multiple intelligence signal processing (Multi-INT) C2 into a single application that fuses radar, electronic intelligence (ELINT), communications intelligence (COMINT), as well as electro-optical/infrared (EO/IR) data, all with the aim of significantly reducing manpower inputs. TacSit-C2 also features GA-ASI's Multi-Mission Controller (MMC), which enables one user to control multiple unmanned vehicles at the same time, even if they are different types.
Leveraging this capability, a single controller might be responsible for a flight of multiple unmanned aircraft, of two distinct types, all searching overlapping areas for objects or events of interest. That user might tell the system to search for a green car and issue an alert when it detects a green car. The software therefore expedites the processing, exploitation, and dissemination of intelligence by automatically sending updates and data to relevant workgroups.
Quadratix is also designed to act as the brains of the new class of semi-autonomous Collaborative Combat Aircraft (CCA), including General Atomics' YFQ-42A. These aircraft will require a significant level of onboard autonomy to complete their complex missions in coordination with crewed platforms or independently, if needed.
Traditionally, unmanned aircraft have relied on a datalink back to a crew at a remote ground control station (GCS) for flight control, sensor manipulation, and data dissemination. Operating at extremely long ranges from the operator and possibly in dense electronic warfare combat environments means that these links may not always be assured. The autonomy enabled within Quadratix ensures that the uncrewed aircraft will be able to carry out their missions even when isolated from the human operator.
The War Zone spoke to Darren Moe, GA-ASI senior director for Automation, Autonomy and Artificial Intelligence, about how Quadratix has developed and how it actually works.
TWZ: Can you explain the genesis of Quadratix and how it has evolved from the company's experience in the uncrewed aircraft world?
Moe: We've been in the technology business all along, but what people mostly saw were our aircraft and not the supporting systems that made them work. Yet that's always been a major part of our effort to support the U.S. and allied nations. When you spend more than 30 years inventing, trialing, operating, evaluating and repeating – as we have done with unmanned systems – you build up a great wealth of insights, not only about the aircraft hardware but also about the enabling systems.
There's also a big aspect to our work that takes place downstream of the aircraft. What we're doing is flying around for a really long time, soaking up huge quantities of data and intelligence, and then we and the user need to be able to exploit it. So again, although what everyone knows is that an unmanned aircraft took off and flew a mission and then came home, don't forget that there are military and intelligence and other users that are then getting insights out of the mission data the same day, or later in the week, or months or more – and that processing, exploitation, and dissemination can be as critical as the actual flying missions.
One thing that's difficult to understand at first but important in all this is that these aircraft collect much more data than any human could process. Imagine flying in an airplane for 24 hours or more – just think about all that you could see with your eyes – and these aircraft are taking in much more than just information in the visual spectrum. So processing and distilling are critical to getting just the information that matters to the operators and others depending on this.
TWZ: Can you detail the major individual capability areas that have been fused together in Quadratix and how you have been able to combine them to create this software? For example, how does TacSit-C2 fit into it?
Moe: Think of it in terms of maximizing operations, sensing, sense-making, and processing, exploitation, and dissemination. Classically, a large unmanned aircraft was flown live via a remote link by a human crew at a ground control station somewhere. So you'd taken the crew out of danger but you still had a crew of several people focused on operating the system live. The first task for the software suite is to reduce the number of people needed and the effort required by those still engaged.
One way we do that is by automating a great deal of the mission events. So instead of needing a pilot and a sensor operator looking down, fully focused, thinking 'where is the vehicle we're supposed to be looking for,' we have the system look for it as it's flying the surveillance mission. You don't need a set of eyeballs glued to a screen – where if they happen to miss it, the mission may not succeed. Instead, the aircraft and the supporting software fly a route, look down, and when they see a vehicle that might be the one, the system pings you to say, 'hey, here's a truck that matched your description; is this the one?'
That's what applications like TacSit-C2 do, or they let one human operator supervise multiple flights all at the same time. Say you want to operate a group of aircraft over a big area to have excellent ISR coverage. You don't need the same number of human crews in individual GCSs all with their hands on a throttle and stick and so forth. You could have one person with a display looking at those aircraft all doing overlapping loops so they, collectively, can see what's below.
One of the aircraft might pulse you and say, 'I just spotted a vehicle that looks like what we're supposed to be looking for; here it is.' And the operator can click to see its video feed and think, 'yeah, that's the truck we want.' Maybe that's all you need to do, or maybe the operator does decide to put his or her hands on the controls and take the aircraft off its automated orbit and get closer or do something else.
I could go on and on, but the other aspect to Quadratix is that it's not just about the mission. It's about the supervisors and the intelligence officers and supported units and others that need to know about what the mission uncovers — and if that aircraft's discovery of that truck is a triggering event for something else. Let's say we have another operation that's contingent on the discovery of that vehicle. Quadratix can push that insight downstream to customers automatically. My operator doesn't need to scribble it down and have somebody pick up a phone and call somebody. For those inside the enterprise that need to know when things happen, live, it's just like getting a push notification on your phone that your team just scored.
TWZ: Does Quadratix offer a menu of options which you can pick and choose from for the most appropriate application for your mission? I'm thinking of the Adobe Creative Cloud analogy here where I might select Photoshop for image manipulation, and Premiere to edit video.
Moe: You've hit the nail on the head. Not every mission will require every application and not every user will touch every application. In your metaphor, the creatives that are working in Premiere might be on a different team from the ones that are using InDesign because you have a video production workflow that belongs to the same department as the graphic designers but isn't exactly the same and doesn't have the same individual users.
TWZ: Quadratix enables one user to take control of multiple different kinds of unmanned aircraft. How is this achieved? Is it via a ground control terminal or a laptop, for example?
Moe: We've been incorporating high degrees of automation into our systems for many years. Our newest types of aircraft can take off and land automatically or over satellite, as mission needs require. Another reason to add this capability was to enable the multi-aircraft control you've mentioned. Much of the time when medium-altitude, long-endurance aircraft go on patrol they are making very long transits to areas of interest or they are spending a lot of time over areas in which perhaps not very much is going on – empty desert, for example, or open ocean.
This is part of the value of using unmanned systems. So it's natural to reduce the number of human personnel as much as practical while also maximizing what these aircraft and supporting systems can do, and this can all be done from an office desk or an expeditionary laptop.
TWZ: Can different roles be achieved by one operator – i.e., could Quadratix control an MQ-9 gathering data and a CCA prosecuting the target that's being sensed by the MQ-9?
Moe: Broadly, yes, but it's important to make two critical points here. First, the MQ-9 capabilities and mission are very different from CCA and so although they're all encompassed within Quadratix, it's helpful to go back to your Creative Cloud example: the users doing MQ-9 ISR and the users doing air superiority with CCA likely wouldn't be the same people or necessarily organized together in the same squadron.
Second, CCA always would prosecute a target under human supervision, or under conditions in which an air force pilot or other combat aircrew were directly involved. But Quadratix is ingesting the data from both platforms and correlating with other data sources so that disparate squadrons and other stakeholders have a common targeting display and shared situational awareness.
TWZ: Quadratix uses machine learning and AI to control CCAs. Can you explain why this kind of software is critical for CCA in particular and how machine learning and AI are such important factors? Is this about operations in degraded communications where a CCA would need to be semi-autonomous and operating at the edge?
Moe: In this context, what machine learning enables is for software running on an aircraft to take in as much as possible about real-world conditions and then, when it's confronted with those or similar conditions again, act in the way you want. We say to the aircraft, 'what you did was right, do that again;' or 'what you did was wrong, don't do that next time.'
This is important for a number of reasons. One is – and you alluded to this in your question – let's go back to the classical ways that people operated unmanned systems: a crew is sitting in a GCS somewhere connected via satellite, live, to a real aircraft booming around in the air someplace. But if that aircraft has to stand in against an adversary that can degrade the satellite connection, you're in trouble.
Making YFQ-42 semi-autonomous takes that particular play for an adversary off the table. Now it can press in, see what it sees, warn other aircraft and, if necessary, take action ahead of the friendly fighters and let those human pilots stay back out of harm's way for the initial action if that's what happens.
TWZ: Could you give us a combat scenario and explain how Quadratix would be used? Perhaps a complex Pacific scenario that includes CCAs.
Moe: It's possible to imagine a situation in which large numbers of aircraft, communications networks and Quadratix combined in order to defeat a revanchist adversary in a big, mostly air and maritime environment.
You could fly large numbers of MQ-9B SeaGuardians from allied bases in the region and get them into a consistent battle rhythm watching key waterways or other areas of interest. This is one of the great things about unmanned aircraft – when you have sufficient numbers of them, you can patrol nonstop. Even with the exceptional endurance of an MQ-9B, it will eventually run low on fuel and when it does, another one can take its place and so on.
This makes it very difficult for an adversary to move around in the air or on the surface without you seeing exactly what they're doing. MQ-9B also has an anti-submarine warfare capability so that even brings undersea contacts within reach as well. Quadratix lets human crews quarterback it from anywhere. They don't need to be in the area of operations necessarily, but they can be extremely relevant. These aircraft, these systems and these crews can be the first ones that see changes in conditions or new patterns of activity that suggest something untoward might be happening, and that really reduces an adversary's ability to use the element of surprise. We call it deterrence by detection.
For this example, let's imagine an adversary has made the decision to act. Its vessels and ships might take certain actions, and American and allied commanders then, in turn, might make the decision to respond with actions of their own. You can imagine that FQ-42A would be in the initial waves of aircraft that would launch alongside human-crewed fighters, ISR, tankers, and others.
And you can also imagine that events might hit a crisis phase but short of full-on hostilities, and the responsible powers, the U.S. and its allies, say to a notional adversary, 'Hey, we really discourage you from committing your forces east of a certain point. If you do, that will be taken unfavorably.'
FQ-42A could fly that barrier combat air patrol, right on the line, and the U.S. and allied air component commanders could know exactly when an adversary stepped across and also have the first aircraft at risk be these CCAs. Quadratix is what enables the FQ-42A to sense what aircraft are over there, where they're moving, whether they're armed and so on. What we'd hope is that the presence of these and other assets prompts the fictional adversary in this scenario to decide not to become an aggressor.
That's what we're trying to do here – we want them to look at us and say 'I don't know if I can succeed today and I doubt it so much I'm not going to try.' But clearly part of this has to contemplate the idea that they might commit. If they shoot, an FQ-42A might shoot back. In so doing, the adversary is probably going to energize their radar and maneuver and potentially do other things that let other FQ-42As or other allied aircraft in the area see what is happening and respond appropriately.
Quadratix, sharing data and sensing and so forth, will be what helps enable a lot of that capability in the future. We don't want FQ-42A to have to engage with any enemy aircraft in the first place, but it can and will if it must. We certainly don't want it to be shot down, but as you appreciate, we'd all much rather that it be what takes these hits and lets the human pilots come back, because we can build many more FQ-42As to go back and take up the patrol or do the other missions, but we can't pump new conventional fighter aircraft or most importantly, new pilots, out of a factory.
Contact the author: [e-mail]