Manned-Unmanned Teaming: Epic Fury Suggests a Clear Future Vector

Watch the Video

---

Read the Transcript

---

Heather Penney:

So now I’d like to welcome to the stage our next panel on collaborative combat aircraft and manned unmanned teaming. Brie General Tim Helfrich, please come up. Lieutenant Colonel Lyndon Otto Bartlett and Dave Alexander, Dr. Jason Levin, and Robert Otis Winkler. Over the past year, we’ve observed the latest evolutionary step in manned unmanned teaming from Operation Epic Fury with the latest employment of the MQ-9 to the testing grounds at Creech and Edwards Air Force Base where CCA development is well underway. The next step of autonomous aircraft is here. And this is an exciting time for the Air Force, but there’s still so much to do in order to make sure that the services next generation fleet of autonomous aircraft are an operational reality.

So to discuss this topic, as I mentioned, General Helfrich, Lieutenant Colonel Bartlett, Dave Alexander, Dr. Jason Levin, and Robert Otis Winkler. We’re going to dive into the questions because this is a really deep and rich conversation. And David, I’d like to start off with you. General Wilsbach referred to the MQ9 as the MVP of Operation Epic Fury, and the Reaper did a lot of good work and the Airmen employed it in ways that were very different from what they did during the global war on terror. But the Reaper is different now than how we envision employing CCA. And as a company that builds the MQ9 is also building the F-42. I’m curious what learning from the Reapers in this higher risk environment, how can we apply that to CCA Increment 1, which is a high quantity, lower cost counter air platform?

David Alexander:

Yeah, thanks Lucky. So just real quick, prior to actually Yemen and Iran, we probably averaged maybe one combat loss a year over many years. And then a recent, through all the news, we’ve been flying MQ9s in high threat areas deployed where manned aircraft, and I mean any manned aircraft, would not fly. So purposely put in harm’s way, but really for a very good reason. And its role was to persist and hunt down mobile targets. And to do that, you got to expose yourself through a lot of time exposure as well as different types of popup threats. And so since February, it’s been a bit of a roller coaster ride in and out and looking at ways to add self-protection and backfill and so on. But I think some of the lessons for CCA, persistence matters, gives you a lot of flexibility. You need to sustain a high rate of air power. And what that means now is that you have to have flexibility in your agile combat employment.

The days of a remote launch and recovery site is gone. Everything’s through overhead and over the horizon. So you’ve got to be able to update your agile combat missions real time. And dispersion is really important because there’s a lot of issues of getting attacked on the ground. You don’t want to be on the ground because you’re a target. And then the threat evolves. TTPs have to be, tactics, techniques and procedures have to be updated every day. So we need to be thinking about that with our autonomy and being able to update our autonomy real time through its own data links. And then probably the last, maybe most important is don’t wait till you’re in a war to start thinking about platform self-protection. And this self-protection I think needs to be tested ahead of time, but it also has to evolve during the war as well. So it has to be something that you can update real time. So those are the last, I would say seven and a half months of my life right there is dealing through that, but it’s been quite a rollercoaster. Thank you for the question.

Heather Penney:

So Otto, how have you been envisioning CCAs giving commanders more options? Because as the commander of the experimental operations unit, you are actually working through a lot of these operational and different employment concepts. How are you and your Airmen envisioning those operational advantages that CCA could deliver?

Lt. Col. Lyndon “OTTO” Bartlett:

I think we’re going to learn tactically, we’re going to learn a lot over the next couple years. What I can say we’ve done so far, and I completely agree with what Dave was saying a second ago on Tempo, the jets, what we’ve learned so far, most of our experimentation’s been done for generation, how easy they are to maintain, how small a footprint could be, and that we’ve been pleasantly surprised. The jets are extremely easy to maintain. Now, right now we averaged in our last three ACE events, ACE exercises, which is generally us testing how small the footprint and how fast we can turn the jets. We’re at a one-to-one maintainer to airplane ratio, which is crazy small. That was a very pleasant surprise from our side. Cross-functional AFSCs, huge applications here.

So CCA seems like because they’re a little bit simpler, because their role is so focused, what we found is that just about any, not just about, but we found that any AFSC can be trained to generate sorties, to load weapons, and we’ve done some of those things already and we’re going to do a lot more experimentation over the next couple of years. But it seems like if Dave and Jason, as they make these things, it’s their responsibilities. We see it to make jets that work, that the radios talk to each other, that they can shoot missiles correctly, et cetera. It’s on Brigadier General Helfrich and his team to make sure that those requirements are scoped appropriately. But from our side, from the war fighter side, we see it as it’s our job to ring them out and make sure that beyond the technical specifications of it, that we’re experimenting with the more complex things like logistics and C2, how do we battle manage these things? Those bigger topics that will actually make sure the force is ready for them when they field.

Heather Penney:

So you’re developing a lot of the DOTMLPFP pieces of this. It’s not just turning and burning that you’re doing out at the EOU that you’re stress testing. What are some of the challenges that you’ve uncovered as you’ve been going through this? So you’ve had some pleasant surprises, the one-to-one ratio is an example, but what are some of the challenges that you really need to get after?

Lt. Col. Lyndon “OTTO” Bartlett:

There are many, and I think that’s okay right now. The jets are pretty new. I would say mission autonomy is going to be the secret sauce that lets these things become decisive or not. And we’ll know that over the next couple years as that develops. From our end right now, if I were to identify our two biggest problems from the trenches of the EOU is that they is FAA integration or policy implications to an autonomous airplane. So these things don’t have pilots, they’re not RPAs, they’re not man fighters, they have no pilots. Our system is not set up for that.

And that is a challenge that we are finding extremely hard to navigate through and to not go around, but to make a policy that makes just like we need to make sure the Air Force is ready for these things, to set policies that make sure the enterprise and the country is ready to receive these and internalize an autonomous fighter that does not have a pilot that is running on a mission autonomy agent that doesn’t listen to ATC, doesn’t do those controls.

That’s one. I’d say the second one is, and I’ll just end on two here, is policy from the side of staff, from the Air Force implications of we’ve got decades of doctrine and whether it’s 202 standards or publications, but working the nuances of CCA into and fielding policies that are not just copy and pastes from old regulations that are actually something that enables us to move quickly with these things, to waive what we need and to experiment with novel types of air superiority is I think our big challenge right now. And we’ll be working with that over the entire team here over the next couple years.

Brig. Gen. Tim Helfrich:

Yeah, Otto, I love all the things that you’re saying because what that really is indicative of is that we’re doing what we said we’re going to. When we were together at the spring AFA, we said the most important thing we were going to do over the next six months was to put CCA in the hands of the EOU, in the hands of the operator. And that’s exactly where the learning that Otto was talking about came from. We’ve done multiple agile combat employment exercises. We now have a full-time CCA at the EOU where they are able multiple times a week they are doing experimentation. All of this, we sometimes whistle past it because we’re so used to it, but we are doing exactly what we said we’re going to. We have just awarded the EMD Light and Production contract just a few months ago and we already have operators that have spent a whole bunch of time with these jets. So just kudos to what the EOU is doing and the collective team to pull all that together.

Heather Penney:

No, and I think it’s so important because being able to deliver this capability in operationally relevant numbers with a level of maturity requires the kind of concurrency that y’all are doing, figuring out how do you get the nuts and bolts of just turning wheels and getting the aircraft airborne and doing the logistics in addition to the novel elements of the mission autonomy. Jason?

Dr. Jason Levin:

I just wanted to add just like General Helfrich was saying, last AFA was like, we need to get out to the EOU. And then since then, we’ve taken Fury to two ACE exercises. There was one in April and then one in July, and then we delivered an aircraft to the EOU out at Creech just last month. So we have a Fury out there. It’s been great seeing from the July exercise where we trained, sorry the April exercise, a handful of Airmen to basically turn the jet, but it was a couple people, they were flying from a NST, it’s a couple of pelican cases and a laptop. And then we went back in July and we just basically upped the ante, increased the stakes, and it was like, let’s get after more problems.

Now it’s down to a single operator flying a Fury from a laptop. They were doing disaggregated operations, flying from multiple locations. They were doing hot reloads on Fury, so getting it down from hours down to single digit minutes to turn the jet. And so they were really pushing the envelope, flying with mission autonomy. I think it was the first time Air Force operators are flying with mission autonomy on Fury multiple times a day in the airspace with crude fifth gen fighters, other CCAs, starting to build that trust. So I think the EOU is getting after exactly what they need and getting those learnings into the system.

Heather Penney:

So getting after what they need, we’ve had rapid iteration. So Jason with the YF44 and Dave with the YF42, what are the things that you’ve been learning in your companies as Anduril and General Atomics deliver these capabilities?

David Alexander:

Well, I think building the foundation of semi-autonomous and autonomous skills, I think once we got ourselves comfortable with that, then we can move out and like Jason was saying, really get after agile combat employment where it’s not experimented anymore, it’s an everyday event. And so I think that’s getting that understanding of the two types of autonomy and putting it to use and putting it into use in the most practical manner, I think has been the biggest thing to go after now. It’s really kind of laying the foundation so that we can go quick in the future because autonomy’s going to live on data and you’ve got to change it a lot to actually make it work.

Dr. Jason Levin:

Yeah, I think a lot of the learnings we’re seeing are across both the system, the hardware, and the software itself, but then just we now have actual Air Force operators using it. It’s not just developmental tests with engineers. And so they’re actually finding other problems, other issues. And then maybe the learning is how quickly can we fix those things because we need to build trust in the system. So when we go out to those ACE exercises and they don’t like something, if we don’t fix it by the next ACE, they’re not going to trust the system. They’re not going to trust that it does what it needs to do. So I think it’s continuing to just iterate and improve the system at each one of these events.

Heather Penney:

No, thank you. And speaking of iterating and developing capability, Anduril just shot recently a live weapon off of the Y-44, the Fury. And it’d be remiss if I didn’t mention that General Thomas, y’all will shoot your first weapon towards the end of this fall. Jason, would you please share the significance of this milestone because I think it was very significant.

Dr. Jason Levin:

Yeah, it was a big deal working hand in hand with the Air Force. I mean, we wouldn’t be able to do it without them. It was like one team to get there and it wasn’t just a safe separation event. And so this is where we like to put a lot of the hard things first. So we took off from Edwards, we flew into the range, Fury ingested a real track via Beyond Line of Sight. It was ingested into the aircraft, basically got on the right heading and then fired the AM120 and guided it to the simulated target. So it’s basically validating the entire Beyond Line of Sight kill chain here and pulling that thread all the way through, which I think is a key piece of what the CCA program is doing. Rather than focusing on each piece, it’s like, “Let’s make sure everything works and then expand from there.”

And then because Fury has external hard points, we’ve already begun some of the integration and fit check of some other weapons. We know a lot of partners and allies are interested in air to surface, air to ground munitions. So we’ve begun integrating some 500 pound guided bombs, small diameter bombs, rocket pods for APKWS, targeting pods, other air-to-air weapons. And then there was always an interest to start looking at long range strikes, so things like FAM or a Barracuda 500, and how do we get low cost cruise missiles onto something like Fury? So the air-to-air mission is extremely important. It’s a priority for the Air Force. It’s what we’re going to focus on. We started to put some calories against some of these other munitions integrations as well.

Heather Penney:

No, thank you. So General Helfrich, a major reason as we understand it, is that CCA are in the fighters and advanced aircraft portfolio to enable the greater integration across those platforms. So as Jason said, the air-to-air is the main focus of CCA. How do you take war fighter feedback from the EOU to ensure that we’re creating a converging alignment across all the aircraft?

Brig. Gen. Tim Helfrich:

Yeah. First thing, Lucky, is I’m going to push back on the entire premise of your question.

Heather Penney:

You got it.

Brig. Gen. Tim Helfrich:

Not just for sport, but really because CCA are in fighters advanced aircraft because CCA are autonomous fighters. The autonomy is an attribute of the CCA, but really it’s about what mission they perform. And that primary mission is counter air. And so they’re performing what would traditionally be considered a fighter mission. And so that’s why they’re in fighter advanced aircraft. It does most certainly help with integration that I also manage the acquisition sustainment modernization for all the fielded and next gen fighters except for F-35. And so it does make that integration much easier. But maybe back a little bit to your question of how are we integrating and leveraging what we do at the EOU? There’s three big things for me that the EOU can bring that will help us to continue to iterate. One is their location at Creech, which is determined or was chosen for this, but they’re located near Nellis Air Force Base where you have the weapons school.

You have many of, well, all of your fighter aircraft so that we can do integration work and get operators involved quickly. Next, I expect them to be what I would like in a software term, early adopters around mission autonomy. And early adopters are not just the people that go and try things early. They’re also the ones that provide rapid feedback so that that product becomes better and better. And so the expectation is that as we go and perform efforts both, well, with manned unmanned teaming and the EOU provides feedback on what needs to change with mission autonomy, that will then get incorporated very quickly by the mission autonomy vendor. And then I think the last one, which is an intangible, is the trust factor. The fact that our fighter pilots are going to get reps and sets working alongside the CCA through EOU’s experimentation will go a long way towards where we see CCA going for integration in the force.

Heather Penney:

Okay. So I have to dig a little bit because you said something that was really interesting to me that these aircraft are fighters. So do you see them. I’ve always understood CCA as augmenting a manned platform to provide enhanced effects for that mission package. Are you seeing and envisioning a demand for CCA to be independent aircraft executing air superiority operations?

Brig. Gen. Tim Helfrich:

Yeah. So CCA have been designed really around increasing both survivability and lethality of current force packages. So there are many manned platforms that you would describe in the exact same fashion.

Lt. Col. Lyndon “OTTO” Bartlett:

Can I just pile onto that and say the way we see that in a collaborative. I mean, it’s in the name that they’re collaborative. Taking them out of that could technically be effective, but would not enhance the joint force the way that we see a collaborative integrated force package to be. That I think is a critical thing for us all to remember as we move forward is the intent of this is to, this thing does not necessarily need to be the thing that is the hard counter to any counter air effort, but it is something that when force package integrated into the type of force pulse you should see when these things are fielded in that year group, then that’s where you start to see the advantages in the counter air role.

Heather Penney:

And how quickly do you see those iteration, that feedback loop Otto? When you’re doing flights of the CCA with manned platforms, how quickly are we able to get that kind of iteration back to the mission autonomy piece? And Jason and Dave, this is probably a good question for y’all as well.

Lt. Col. Lyndon “OTTO” Bartlett:

If it’s in a simulated environment, it’s immediate. I mean, at General Helfrich’s site, when the mission autonomy iterations are happening, our guys are literally there with the programmers and the vendors to do that iteration right then. They give the feedback and if it’s possible to do immediately, they take that feedback. If not, then they work on it on the next turn of the software. I would say just to caveat, keep it in the bounds of reality, we have not yet done a LFE with these for real. The first one’s going to be Emerald Flag, that’s in December. That’ll be our first tip dip of the toe in in actual force integration.

But that will, in Otto’s mind, be the first time that we will ever do some CCA end-to-end kill chain live. That’s the first time we’ll ever be integrated in a fourth and fifth gen combined OCA push. It’ll be small, but those type of building blocks are where I think you get the intangibles like trust in the CCA, the fact that it’s not going to be high aspect coaltitude with me when I commanded it to go set, that kind of thing. Those learnings are going to be just as valuable as the technical, does the radio work properly? Does the button that I push to say target work properly?

Heather Penney:

Yeah.

David Alexander:

Yeah. I just would like to maybe pile on a little bit about as we’re learning, you got to be careful that you don’t take an event that happened with a sample rate of one and then start changing things. So I think you want to take a problem sometimes and bring it back and bring it into the simulation so you can run Monte Carlos on it and really expand the fix, if you will. So we got to be a little bit careful of just I saw this happen and I’m going to make that one change and that one change could not be the right change. So learning in real flight versus learning in SIM, I think is something we’re going to have to think about in the future. I mean, if it’s just binary and focus broken, then you fix it obviously. But if it’s a skill that’s complex, you want to probably take it in and do a bunch of Monte Carlo runs and make sure your fix is really a fix.

Heather Penney:

And do you have the. I’m sorry, go ahead, Jason. PIO. Do you have the explainability to be able to dig down into that, especially with the government reference architecture of the flight autonomy, the mission autonomy and all those level two skills? Do we have the explainability to understand the interactions vertically across the skills through the various levels of autonomy?

Dr. Jason Levin:

Yeah, I think we’re getting there on that explainability. I think the rate to debug a problem in software today is extremely fast compared to what it was on previous fighter aircraft. So that’s already revolutionary. I think we’re getting to the point where when something happens on mission autonomy on one flight, probably by the end of that day they’ve debugged where the problem is, they have the explainability. And then I think we’re working as one team of how do you update that very quickly? That’s not something that is standard where the next day you’re going to change software on a fighter aircraft. That’s just not something that’s happened. So I think we’re working as one team with the Air Force to figure out, okay, how do we very quickly update mission autonomy? How do we very quickly get that back onto the jet? That is the entire point behind separating mission autonomy from the flight autonomy is to update the tactics, is to update very quickly. And so we’re just going to have to figure out as one team how to do that, and I think that’ll be happening pretty shortly here.

Heather Penney:

Excellent. So Otis, you’ve got fighter experience and from your KRADOS experience perspective as well from industry, I’m really curious about how that background might give you a different view on manned unmanned teaming, the iterations, the behaviors. What do you want the autonomy to do or not do, and how do we create baby steps to enhance that trust and what’s the right pace of that kind of feedback loop?

Robert “Otis” Winkler:

Yeah, so I think it’s a great question. Otto mentioned it. We have some kind of policy and process differences for what we’re learning now versus what had been traditionally really in manned flight. And one of those is when you have a product that makes it through the PAE, makes it out, gets fielded, goes through IOT&E and DT, and then gets put out into the war fighter for operations, they expect it to work perfectly. We’re not really good at min viable product ideas, but when you take a young wing man and you train him up and then you send him out into the field, you don’t expect him to operate as a fighter weapon school graduate, right? You expect him-

Heather Penney:

He is a min-viable product.

Robert “Otis” Winkler:

Exactly right. He is the min viable product, the minimum acceptable level of performance to be able to put into combat. And so when I think about it from an operational man fighter point of view, I think that’s what we really fundamentally want to get to with autonomy in the beginning. And I think that that’s what Otto’s trying to do, the whole entire program is trying to do is what is the requirement to put out into the field to get to the war fighter and to start operating this thing this real time.

And then you can decide the collaborative part of this makes it a much easier, I think, technical problem to be able to limit and grow the autonomy as you field this. And then eventually, and I’ll push back a little bit on Otto and the idea that I think eventually we may see these aircraft utilized as collaborative aircraft, but not in conjunction with your traditional what you see in a pulsed operation, but in support of that pulsed operation when the pulse isn’t there yet. But I do think that that is a bit of a stretch goal that’s going to be down the road that I think we’ll eventually get to.

Brig. Gen. Tim Helfrich:

Yeah, Otis, I just want to piggyback on something you said. I love the new graduate from pilot training. Because I think it goes to one of the common questions that we have is people are like, “Is mission autonomy going to be ready?” And I think that there’s a false question there because the mission autonomy that you have on day one is going to be wildly different than the mission autonomy you have on day 101, and it’s going to be gradient through that whole time.

Now, we haven’t exercised those muscles through experimentation yet, but we have set up our relationship with the EOU and our contract mechanisms so that we can get very rapid feedback from the EOU and other operators and use a digital infrastructure, push that data into the mission autonomy vendors and to update that software and whether it’s the next day or two days later, be able to go test to see whether or not that fix has been made. And sometimes it’s not even that it’s broken, it’s that we’d rather it do something different. And so there’s a huge room for growth of mission autonomy just as you would expect a pilot to mature over the years of sorties.

Lt. Col. Lyndon “OTTO” Bartlett:

Otis I-

Robert “Otis” Winkler:

Go Ahead.

Lt. Col. Lyndon “OTTO” Bartlett:

Go ahead.

Robert “Otis” Winkler:

I was just going to say the trust part of this I think is fascinating, but when you have a young wing man on your wing, there’s not a lot of trust. You pay attention to what they’re doing. And I think we’re going to find that same thing as we develop the autonomy. We expect that it’s going to reliably execute the tasks that we give the aircraft, but I don’t think that from the operational side, you’re going to get as much pushback on the trust factor. So I’m interested to see if that pans out.

Lt. Col. Lyndon “OTTO” Bartlett:

Yeah, I am too. I think back to the, is it collaborative, is it integrated or is it not? I think that those are the exact type of experiments that we need to do. And those tactics or the perfect use of them is likely not going to be found by somebody in the EOU. I’ve got some hitters like all my guys are handpicked, we love them. Chelsea Bartels is one of the smartest crew chiefs I know. She, Danny Schreiber and a couple of the cats are the reason that we are at one to one human to machine ratio on these. What I would say is that the EOU’s job for the tactical side, the how do we solve capability gaps in PACOM, those are things that we need to, that whether EOU does it with our own humans or whether we lean on the 433rd or the 6th, the weapon schools, the 422s, those are the guys who are going to actually figure it out. And it’s our job to just write it down and make sure that that gets back to the policymakers or the force employers that actually need it.

Heather Penney:

So I’m really curious, we talked about trust, minimum viable product. If you get that young wing man, you don’t trust them so much, you’re going to keep a tighter leash. You might not want them to have complex behaviors. You might want to have a little bit more tight control and dial down the autonomy, if you will be more directive. But there are other scenarios in wartime where things go crazy, you get the exploding cantaloupe scenario and you might want to be more directive. So I’m curious, Otis, from your perspective, are there scenarios in which the war fighter would need the ability to dynamically dial up or dial down or do you see a pathway of how you enhance the complexity of meat machine behaviors to ensure mission success?

Robert “Otis” Winkler:

Yeah, I think that absolutely would I want the last starfighter mode at some point here to be able to shoot all the weapons everybody has autonomously and to kill-

Heather Penney:

R2-D2 yeah.

Robert “Otis” Winkler:

The hoard that’s coming at you. I don’t think that should be a gate for fielding though. We’ll get to that at some point. Policy will come along as we end up demonstrating that we have the capability to do this correctly. But I don’t want to withhold until we have the perfect solution in that. And then an AMRAM right now, when you shoot an AMRAM, you’re sure it’s going to go out and it potentially might hit something, but there’s no guarantee that the thing that you have locked on your radar is the thing that’s going to get hit by that AMRAM. So it is autonomous. It’s an autonomous weapon that is going out there and trying to hit the thing closest to cue.

In the same way that we’re going to have autonomous weapons that are going to behave fully autonomously at some point, which is going to be an outstanding capability for the men and women that are sitting in the man fighter that are assisting it or the person on the ground that’s waiting for somebody to come up from the air to apply support for them. We’re already there when it comes to weapons today, but we didn’t get there. We didn’t wait to field the AMRAM until we were perfect. I don’t think we should wait to field this capability to get it in the hands of the war fighter, whether it’s to support folks on the ground doing agile combat employment, folks at the 422, the weapons school or wherever else to get it out there to learn more and more about things.

Heather Penney:

Thank you. So we’ve been obliquely talking about software development as we talk about the iterations of mission autonomy. And so gentlemen, General Helfrich and Otto, would you mind describing how you’re approaching building the flight autonomy, the mission autonomy and those level two skillsets for CCA, how you’re envisioning that?

Brig. Gen. Tim Helfrich:

Yeah, I’ll talk some of the acquisition mechanics, then I’ll let you jump on top with that, Otto. I’ve talked about it in many forums and I probably sound like a broken record, but we have fundamentally taken a different approach to acquiring mission autonomy than how software is typically acquired. We are using what we call software sold separately. We are able to do that technically because of the autonomy government reference architecture, which splits out that flight autonomy, which is heavily coupled to whatever platform you’re using and the mission autonomy, which really gets at the higher level tactics and the control behaviors. That has allowed us to get into what we’re currently doing now, which is we’ve awarded, we call it a production mission autonomy contract, which has six different companies on there.

And allows us to continue doing what we’ve been doing the whole time, competition, continually competing to get the best of breed, to get the best product for the war fighter. We now have three vendors that are going through a couple six month competition phases where at the backside we will have one vendor that will field our initial mission autonomy and they are going to be required to have engineers, software developers ready to go to take in the input from the field and to make those mission autonomy updates. So there’s a lot of things we have to do, but we do have the mechanism through the contract and whoever wins that wins to be our initial mission autonomy provider to be able to iterate and make the changes that Otto and his team needs.

Lt. Col. Lyndon “OTTO” Bartlett:

The benefits that we get as the war fighters from that competition is immense. I mean, we see it all the time at our level where the vendors are trying to outperform the other, they’re trying to figure out who’s doing better and then perform faster. It’s beautiful. Our pitch back in from the EOU side or from just the EOU’s particular squadron and the guest help we use from the rest of the test community is we get the mission autonomy in a perfect world ships directly to us. We iterate on it, we provide those feedback results back to the vendors and to General Helfrich’s team and then they take that feedback immediately, which they definitely do.

Now oftentimes our feedback is, I think there’s some friction there, which is okay as we give in sometimes the acquisition community and the war fighters, we don’t understand what we’re asking for. But that level of learning, it’s all about the learning for us and getting that back to the program office so they can make decisions on reprioritizing budget funds and then the effects we need. What we need autonomy to do is written down. I won’t discuss the particulars in this forum, but we know what we need it to do or we know what we think it needs to do. And then we’ve structured our experimentation game plan to analyze can it do those things? And if not, then how does it need to be different? And then as they iterate on it, which we’ve already seen a couple times, we get that at the EOU, we get the new version down, we fly it again. Is it fixed? Yes or no? And then we move on from there.

Heather Penney:

Gentlemen, is there anything you’d like to add to this?

David Alexander:

Yeah, I would say that we have established through, I don’t know, maybe the last five, six years, doing a lot of surrogate testing with our MQ20 program and taken commands from F-35 and F-22. And like I was mentioned before is to establish this foundation of the autonomy core mission validator, which is on the airworthiness side and then the airworthiness software that flies the airplane so that when you do dump in a new skill and the skill says, “Go do stupid,” the validator grabs it and says, “No, I’m not going to do that.” And I think that’s an essential part of this program because if you have to go through a full airworthiness every time you want to add a new skill, so you got to make sure you can’t break the airplane every time you load something new. And especially if you’re loading it over the air and go, “Yeah, try this, try this, try this.” So that’s probably on the platform side, one of the biggest things that we really needed to put together is that foundation so that you can grow the autonomy as you go.

Dr. Jason Levin:

I would say everything Otto has said is working with the mission autonomy out at the EAU. I mean, I think it’s been critical. And I think we’ve talked about a lot about trust, but it’s just building trust even in simple things. So even when they’re just flying patterns or racetracks or using the mission autonomy for different components that are just normal parts of flight, it needs to do that really well and then we can build up from there. And then I think that gives a baseline. And also the mission autonomy doesn’t always do what we ask it to do just in general, but then we get feedback and maybe it did what we though it was supposed to do, what was programmed, and then we can update that, but just getting them to use it and repeat it is really goodness. And then like General Helfrich said, I mean, the Air Force has done a huge amount of lifting with the AGRA.

So if you make your aircraft AGRA compliant and you make your software AGRA compliant, you can just slap the two together. I mean, in February of last year, we took Fury off. We turned on Shield AI’s Hive Mind Mission Autonomy. We turned that off, we spun back up Anduril mission autonomy, we turned that off and then we landed. And so that was praise is a huge success. We switched out the AI brain twice on the same flight, went back, told our software developers in Anduril, and they always look at me shocked, “Why is this a big deal in modern containerized software? Like your phone, you turn on many apps a day, you don’t high five each other.” So the fact that we’re doing this on a fighter jet is good. I mean, the Air Force has embraced modern software development and it really shows with the AGRA. So just give them a lot of credit for doing the heavy lifting there.

Robert “Otis” Winkler:

Yeah. I was just going to say, this is something remarkable. I don’t know if everybody else is caught in the audience, but GA is doing their own work on their own dime in support of this program. Anduril’s doing their own work on their own ranges in support of this program on their own dime. KRADOS for the Marine Corps program is doing the same thing. I think fundamentally, the program itself and the acquisition process in the last year and a half in the department writ large has fundamentally changed to allow for, in the first aspect of this, to allow for this thing to be able to move faster. And then the second, the PAEs are now demanding that to be able to move forward. And not necessarily CCA specific, but in general, the acquisition process is really moving things forward so much faster and such a faster rate, whether it’s software defined, autonomy, or just the hardware integration with weapons and things. It’s really remarkable.

David Alexander:

And I would fine tune the acquisition to the ADO, doing an A+ job on bringing the GRA architecture. So definitely the vendor lock is gone and this is the first program that’s doing it.

Heather Penney:

Yeah, the constitution works.

David Alexander:

So hat’s off to the ADO and making this happen, making it happen fast.

Heather Penney:

Absolutely. So last question, General Helfrich. We heard yesterday morning about cost-efficient warfare, that the department is very concerned about the cost of all of its platforms. And so it begs the question of the cost of autonomy because there’s an impression that autonomous systems will lead to cost savings. So I’m really curious, what business case analysis have you done? What cost analysis have you done beyond the development to the business model for the CCA, for mission autonomy, sustainment, modernization, skill advancement? Have you been able to do this kind of analysis on the life cycle cost of autonomy?

Brig. Gen. Tim Helfrich:

Yeah. Lucky, we’ve done a substantial amount of analysis. As you might imagine, when you go and you do a completely different paradigm, you start to implement some of the key building blocks of the Sec Wars acquisition transformation strategy, you have to think about how to do things differently. And so we know that software sold separately is going to likely drive a different model, especially when you think about how you buy Microsoft Office today through a license and that Microsoft Office continues to get updated versus Microsoft Office back when I was a second lieutenant.

We don’t need to talk about when that was, but back then where you just got a disc and you put that disc on and that software is what it is and maybe you had a phone number to call. So as we’ve done a huge amount of market research analysis, looking at what is going to not only be cost-effective, but also get to mission outcomes because it’s about mission outcomes, we found that a licensing model is the right approach, at least to get started. So that’s how we’re getting started with a licensing approach because we see that that continues to drive competition. And then we also see that that will drive the responsiveness by the mission autonomy vendor to incorporate that feedback rapidly from the EOU and other operators.

Heather Penney:

Okay. Very quick lightning round. Otis, let’s start with you so that General Helfrich can have the last word.

Robert “Otis” Winkler:

Yeah. All I say is that we got one buydown risk and lower cost of the mission systems, which is what ADO is doing. Get things out to the war fighter, get it fielded at a min viable product, and let’s start utilizing these things to the maximum capability.

Dr. Jason Levin:

I’d say the Air Force CCA increment one is the fastest moving and most mature CCA anywhere in the world. It’ll be the first to field of any CCA in the world.

David Alexander:

Just a note to the EOU, don’t wait until you’re in a war to worry about self-protection. You might want to include that in some of your testing. And secondly, General Atomics will lean forward and go at risk, but we can only lean forward so far. So we got to remember that we got to get budgets on time and it’s really tough to run a business.

Heather Penney:

With the CR, I understand.

David Alexander:

With everything that’s going on these days, it’s really tough. I think it’s hurting our national security. So just a little dig there.

Lt. Col. Lyndon “OTTO” Bartlett:

Nice. My hit-and-run would be, I think that the ADO and the vendors are going to produce jets that work. I think that we are a fifty fifty shot of being left with jets that work and regulatory policies that are not ready for them. And I think that level of work is not attractive, but it is equally as important as making sure the radios talk to each other and that it’s easy to miss that amongst the sprint, but we need to make sure we keep focus on that.

Brig. Gen. Tim Helfrich:

All right. And I don’t have much time. I got to get back to work because we have to deliver 500 vengeance and fury by 2032 and we need them to be effective and integrated. Thanks.

Heather Penney:

Gentlemen, thank you so much for your good work and thank you for being here for the Manned-Unmanned Teaming Panel. Have a great airspace power kind of day.