Collaborative Mission Autonomy: The Key to Next Gen Airpower — Ep. 273

Welcome to the Airspace Advantage podcast brought to you by Penn Fet. I'm your host, Heather Lucky Penny. Here on the Airspace Advantage, we speak with leaders in the DOD industry and other subject matter experts to explore the intersection of strategy, operational concepts, technology, and policy when it comes to air and space power. So this week we're going to explore one of the more cutting-edge topics in combat air power, collaborative mission autonomy. Alright, so what does that mean? What I'm talking about are the technologies that team varies uncrewed aircraft, all collaborating with each other with a crude aircraft component to go accomplish a combat mission. It's what links the autonomous collaborative platform or ACP, which many of us know as collaborative combat aircraft CCA, with a more traditional aircraft like an F-35 or a B-21. It's really complex technology, seriously, varsity level innovation and integration. But the reason the Air Force and Industry are pursuing it is the net effect, the teaming between a crude and uncrewed aircraft, promises to unlock greater mission effects than what either plane could realize on its own. And given the future operational challenges will face, that's vital. So with that, I'm honored to introduce two experts on the topic. Ryan Bungie. Ryan, glad you're here. Yeah, thank you. Happy to be here. So Ryan currently leads the Strategic Defense Solutions team at Collins Aerospace. And we've also got Steve Munchfino. Munch. Great to have you. Yeah, great to see you again, Lucky. Appreciate the time. Absolutely. So Steve began his career as an Air Force fighter pilot and is a technical fellow at Collins. And if you all haven't gotten some of Munch's books, I mean, he's written several books. Tiger Check is my favorite. Go order them. You will be so glad you did. All right, so let's dive into this topic. Big picture. Gentlemen, can you help our audience understand the term collaborative mission autonomy? What is it? And perhaps more importantly, what isn't it? Yeah, so hey, Lucky, maybe I'll actually start with answering the second part of your question. What is what it's not? We as an industry have been flying remotely piloted things for decades. So this is not that. This is not somebody sitting in a basement in Nevada. This is not a drone that you programmed to go fly from X to Y. Well, we're really talking about here with collaborative mission autonomy is the software solution. The AI that makes these CCAs truly autonomous trusted we man that allow you to go partner them with manned pilots and achieve new tactics, new outcomes that you otherwise wouldn't have been able to. And again, if you think about, you know, we we talk a lot about what that future fight is going to look like. It's going to look different in that air domain. It's going to require some level of affordable mass some level of these autonomous things to go be effective there. We've really tried to kind of tailor how we're thinking about this around those three descriptive words around collaborative mission and autonomy. And really simply that collaborative piece is that how you create that crude uncrewed teaming how you provide that man pilot with an autonomous capability that makes them more effective that creates greater outcomes for them. That is one that they can trust that collaboration provides more than than you would otherwise be able to achieve when we think about the mission aspect of that. And that the fact that you can do different things with these uncrewed we can they can accept a different risk profile you can put them into a different sensor or weapon range threat profiles. And really that they can perform a mission as you would expect them to that we've trained that autonomy with the same tactics the same procedures the same focus that we would a real pilot so again you continue to kind of build that trust. That is really that autonomy. So relying on these self organizing unmanned aircraft equipped with these a hydro then capabilities in a way that does truly seamlessly integrate with those man platforms. If you think about kind of what these platforms have to deal with in this future fight beyond just the speed and the physics of flight that contested airspace the threat of a determined adversary understanding the rules of engagement like these aircraft have to be able to keep the people of doing all of that. You know, I want to miss me with autonomous league without creating more burden on the pilot. They also have to think like pilots and work together like pilots in short really you know they have to be an effective teammate in what's a high speed no fail no room for air environment. Yeah so okay I'm going to geek out hard right away right we're not because when you start talking about like for example the mission autonomy earlier in your discussion talks but we're going to need to be able to real time swap between missions and I struggle with that because what I understand autonomy in order to be able to have the most effective autonomy is probably going to be far more specialized and not necessarily think differently like for example as a pilot as I'm thinking from a multi role perspective. In one mission I might go from air to air to ground and then back to air to air are we expecting our autonomy to be able to do that because from a software standpoint those algorithms are vastly different they're going to probably need different types of training data and so to have both of those programs if you will together might not be as effective as maybe specializing one especially when we're thinking about putting that into an aircraft and how we optimize that aircraft for cost. And how we weaponize that aircraft as well am I going down the right track. I think how do you're your spot on right part of this is we have clearly defined tactics for different roles and responsibilities and we expect our autonomy in CMA to be a little respond to that the advantage we have is that unlike the human brain or you know the young wingmen you're not trying to the autonomy in its mission processor is not trying to remember all these things like in that in that brain it's it's there. It's there it's embedded in the it's embedded in the computer and so we will probably build up and train differently for those different use cases different missions but I can cram all that together on to the vehicle at the same time and so then it can choose what mission it's in and then pick and execute the most optimized tactics because it's already stored in its brain so we have to train separately. Just like you do today with training different tactics but now it has instant familiarity like there's no training curve it hasn't forgotten how to fly air has gotten rusty doesn't need a knock off the rest and go back to basics and do the building blocks because yeah once it once it has that training it's as good as it ever was. The way I'd like to think of it right imagine that if we could take weapon school graduates all their knowledge all their expertise and package them up as wingman and then make them actually go fly according to the brief so they don't go off and do the reading things like they always classically do right but now I get to take all their expertise and all the recommendations on how to go fly and they fly it perfectly their execution is spot on there is no like I had a bad day the sun was in my eye I had a rock in my. In my shoe or anything right they are executing perfectly again we have to spend the time to make sure they learn those tactics properly the in the beginning but then once they've learned that they can continue to execute and now we can continue to develop those additional tactics and improve upon them as we go forward that's really the growth potential that for the autonomy yeah I was just going to add I think that you start some locks some of the power of what we're talking about here with collaborative mission is on these we think about future heterogeneous collections of man and unmanned things. If they're all operating the same collect collaborative mission autonomy software baseline as much said that's been trained to these different mission sets. They think about the power that starts unlock from different tactics temporally at the same time right or reassigning tasks between wingman and partners. You have it all of the ability at your fingertips there and as much said it's executed perfectly to be honest I've been thinking about it very differently from you because I would think that I would want to have a very highly specialized ACP that was very focused on one particular mission area so it was absolutely the best like we're using the weapon school example what I'm hearing from you is we're basically taking all of the different squadrons and all of the different platforms all of those different patches and amalgamating their various expertise into one particular agent. So we've got whether or not that's eagles or f-22s or b-2s or kc-135s like it's all together and then we just well and then much you said it gets to select it gets to choose what it's where it's going. Yeah so I want to say that I'm necessarily amalgamating all of them but I am let imagine if I could put an f-22 weapons officer and an f-15 weapons officer and a team weapons officer in the same airplane. And at the time based off the mission they swap seats right now ideally my CCA is probably optimized for one type of thing because again to drive down lower cost I probably can't make one thing that does it all. And so I probably generally have one pilot that's in charge the whole time. But I want to have the ability if I have to take a CCA and it's got to go do something that it's not necessarily optimized for I want to fly the best tactic though that it has available. Which I totally agree with by the way right so so that's that's where I see it's slightly different right and so now it's not that I have one agent I have seven distinct agents or seven distinct pilots in the vehicle and I will pick the right one for the right mission at the right time based on what the circumstances are. And the human the human is picking that right agent or is the is the autonomy selecting what it wants to be who is putting in the seat. You can go either way right like you can go it can provide a recommendation going hey I really think you wanted me to do an error mission but I really think I need to go do this thing over here. Are you willing to let me go swap my mission and and as I swap mission I'm going to swap pilot. Or are you go or the pilot the crew pilot goes. I want you to go do this right they they have that level of flexibility and then that's where it gets into really interesting things too of what is that command and control coordination look like. How does the human pilot direct with the CCA or the ACP should go do and what can the CCA or ACP recommend back to that pilot and what does that exchange look like. Yeah that that user interface I think is going to be something that's going to be crucially important in order to minimize friction create clarity and maximize that collaboration. Before we go there like to ask you both where do you where is this right now in terms of technical maturity what lessons have you learned as you've been building these algorithms training them and envisioning what that future looks like. I'll start there lucky so we've actually been working on this at Collins our scientists and engineers for over a decade. So it's a dynamic that we certainly saw coming and kudos to our DOD partners they also saw coming so we're not often able to advertise a lot of the work that we've done in this space but we've been working with various government customers here across those last 10 years to tackle some of these complex tech challenges. And we've really taken kind of all of that early start things like DARPA did with code as an example and continue to iterate and improve on that over the last decade to get to what today are offering in the collaborative mission autonomy spaces, which is a product that we call sidekick sidekick is a standalone total software solution that can support multiple different aircraft types multiple different hardware configurations. We've had the opportunity to fly sidekick in a few different aircraft both surrogate and fully autonomous aircrafts and we continue to partner with a number of different customers to mature that even further. Yeah, I'll just piggyback on to what Ryan said right and kind of getting to some of those lessons learned that part of it is we as we kind of went down this model. Hey, I want to build out individual pilots that are specifically tailored right these key kind of capabilities. We call those DRMs or design reference missions is really just an organizing principle which allows us to attack kind of bite size components of autonomy, but drive them to that weapon school level of expertise where you get. Key capabilities key synergies and then go pivot to the next one and then like I said eventually you can package all those up, but as we did all that what we also learned is that there's this knob that sits over the autonomy. That allows the human operator to go I want you to have lots of autonomy right now little buddies or I don't want you to have lots of autonomy this notion of flexibility in how you use the autonomy like you could have very capable autonomy on the vehicle. But if it decides that it's just wants to push early ahead of the entire package that is not useful autonomy to the larger team and so we will naturally constrain that autonomy to fit within this human machine team in construct. There's this dial over it and now we're back to that user interface to talk about how does that human operator that quarterback use that dial and because I can't manually push buttons right there's too busy doing their own thing already. What are they interact with the autonomy and communicate this level of autonomy that they want the little buddies the CCA's ACPs to have. And what happens if they don't provide that like how do they how do what's the what are the little buddies think their autonomy level is and how do they request changes or and so working through kind of that notion of flexible use of autonomy regardless of the level of autonomy that's actually on the vehicle and what is that interaction. That's all key lessons learned out of this decade long experience of kind of building out these DRM's with operational pilots to figure out what we do with this stuff. I really like that notion of flexible autonomy because it's a lot like when you go out there as a flightly do you have a brand new wingman or do you have a very experienced wingman is the person who's flying that wingman position. Are they also an instructor pilot are they also mission commander so you can give them more autonomy because you trust them inherently more or do you want to scale back and more tightly control and therefore have a lower level of autonomy because you're like I got to watch this guy not really sure what he's going to do. Yeah, when we see it even right again using your example like a mission that we would fly today with human wingman that changes on the phase of the mission too right if you're in the cap waiting to push that wingman better be information if they decide to be out of formation guess what they're probably going home. If that wingman gets lit up by a Sam you're like I good luck to you like i'll do what I can but you better do everything you can to try and survive. And so I'm going to give you lots of autonomy but if you survive that then I need you right back with the flow and on the weapons targeting plan that we briefed because I need you to execute your roles and responsibilities and if you're unable to do that I need to communicate that to me. So we often hear from people that think involving more uncrewed aircraft will somehow cheap and accelerate the innovation would say cheap like it's going to cost less and less and less. But I investigated this in a past research effort and my main takeaway was that the potential operational positives of autonomy in ACP's that could yield significant operational benefits but it's not necessarily going to be an easy journey you guys have been at this for 10 years what's been your experience because we want to see this I mean I'm fully convinced we have to remember. I'm fully convinced we have to remain committed even through some of the bumps in the road to be able to feel that and I do think it's going to be an operational imperative. Yeah so I think along the lines of kind of where we've been on this journey over the last 10 years and what we've learned we really embarked on this kind of starting from a perspective of what we call kind of cognitive engineering. We wouldn't got a bunch of really smart people that know how to fly these missions right people like lunch that have had the time in the seat and starting with kind of their expertise and then leveraging the broad expertise across the corporation like RTX understand the sensors and the effectors and starting to pull all that together to to inform like how do we go. How do we go build the autonomy that does these missions and how do we continuously iterate on it to make it even better. And I think some of the learning along the way that's been really interesting is you just think about like one collaboration is historically just hard to achieve even we're talking about humans talking to humans particularly in the stress of the combat environment. And we made great strides in the last decades to enhance collaboration through more book more robust information sharing think like 16 in the cockpit and things like that but it still requires a human on the other cockpit receiving that information understanding that information making a decision based on it and accordance with whatever your pre range and agreed upon plan was to maneuver the aircraft or get the weapon to the right place the right time that that's hard. Our pilots are accepted line that they practice that but getting that into the autonomy and getting that right and deterministic is very difficult from a technical perspective and then getting it to the point where the pilot can trust that behavior that's going to come out of it is maybe even a little bit tougher. So we've had great experiences where you know as we've built more AI and more autonomy into this we've started to see it actually unlock new tactics and new con ops that maybe starts to move the needle and some of those simulations and war games so again getting us to the place I'd say both industry government. Society as a whole that we trust that AI that that autonomy can generate that tactic and we can trust that is going to do deterministically what we think it's going to do in that use case is maybe as big of a challenge as the technology itself. Oh my goodness so we have we've talked about trust a lot especially with respect to collaborative combat aircraft and other elements of autonomy so much because you really just do have to get the experience the reps and sets and it's important piece that that you mentioned that you trust the autonomy will do something that's operationally beneficial whether or not that's predictable is in deterministic or you trust that the innovation that will be done on the fly will actually achieve the desired effect that you want. Manage your money on your time with 24 seven digital banking on the pen fed mobile app easily make payments transfer funds and deposit jacks through mobile deposit. That's music to my ears learn more at penfed.org federally insured by NCOA. That leads me to think like when we talk about collaborative mission autonomy it's got to deliver something that we wouldn't see with our standard con ops and in war games that we've done we've seen that that the presence of autonomous platforms. Decrease risk for the commander but they also as you mentioned unlock unusual things like we might not think about doing before and I call my D's right you know and they've grown over time detect so we can send autonomous platforms out there to do detection at ranges and on axes that we wouldn't necessarily be coming from with a standard push we do want to detonate force adversary to respond maybe perhaps sub before the main push gets there degrade their situation awareness three W or confusion or other chaos. We want to deny them through jamming and other things wanted to complete their weapon stores so we want them to use up their weapons destroy their stuff so we're going to blow them up make them honor the threat. And importantly I think this is kind of fun desynchronize their operations get them out of sync so that they're on the ground out of gas and out of weapons when we come in but that's just kind of very very big picture and I know that you guys are getting into the more detailed elements of these innovative operational con ops. What do you think we could accomplish through this teaming paradigm that wouldn't be possible solely relying on career craft. Yeah, I mean, first of all, I think that's a great list lucky and and I see guys that spend a lot of time. TM TM lucky TM starting through right. I mean part of it and what we're hitting to is this notion that that context matters right like there is not going to be one autonomy solution right one like going back to our multi agent there's not one amalgamated agent that can do all this optimally especially when you start putting it on very specialized aircraft that are that way because they need to be low cost and so. I always like to say right all autonomy whether it's in fire plans or on your computer or whatever else right all autonomy is contextual it's all based on that environment. The problem is that human warfare is the most contextual activity in the world right like so our autonomy for human warfare that enables or supplants or augments human warfare by nature is going to be hyper contextualized. And now we get into okay what does that mean in practice of building out these technology building out these algorithms when do I decide when I'm going to detect versus degree versus deny or detonate based off the equipment that's on my aircraft based on what I see from the thread what's the commander's objective what's the weather today all of those things that you got into. It's all that judgment contextual yeah and you have to sort through that and you have to have autonomy that can sort through that but now to get to that other point it's got to do that in a way that the human pilot that it's helping to augment or the human warfare the human commander can help understand and they can rely on it to understand the context and operate in the context in a way that moves the mission forward. That's the real challenge that we're all trying to sort through and again our way to kind of work through that is go I can't bite all that off at the beginning. Let's go tackle pieces and parts build out that context make sure the autonomy understands that get the operators comfortable with the autonomy operating in that context and then I'll go build out the next piece of context and then I have the ability to put it all together. And because those bite size pieces are built with trust and explainability and resilience kind of all baked in as I start to pull it all together I still have that element there. One way to get after this that starts attacking your 5Ds and that's then what we found is how you then start unlocking true new condoms. If I can get the little buddies to understand that context of what are we trying to do when a fighter pilot talks about the what ifs or it's the all the it depends the contingency. We can start to unlock those it depends and we can start to come up with things we don't just rely on little buddies to be geographically deconflicted or deconflicted in time right we can actually put together these synergistic cohesive packages where they're operating like a true team integrated and providing value that you can trade off responsibilities on the in on the moment or in the moment on the fly to adapt to this changing context. So Ryan you had mentioned that bringing combat pilots and your algorithms together that you were seeing new and innovative con op so what was surprising to you what when when fighter pilots went wow okay I didn't know I could do that what did you find. Yeah, I think it was really enlightening because you know I think we kind of came at it from an expectation of like hey like this is the manual says right like we get to this point we do a grinder we do settings right and three dash one yeah you know just just program it in there. Tommy would start to inform different things or I think in some examples we would even think like hey logically we talk about the risk profile these things can go case we're going to put them out front they're going to shoot their shots first they're going to bug out. And actually what we found in a lot of scenarios is not always the right way to go do it that maybe they serve as kind of extending that sensor range you allow the man platforms to fire their shots help guide them in get the mandates that's out a little sooner so maybe they come in and delos maybe it's yeah yeah so a whole different way of maybe starting to think about the how we get after that. And I think the other couple things that that are really important to kind of keep in mind here from what we've kind of find is just the ability to iterate at speed right so sure you go try something out right whether it's a virtual environment or real environment kind of figure out from a mission debrief perspective man that didn't quite work right. It's a little bit maybe more straightforward than how you would do it in a real world man piece of going and thinking working with your partners think about how we go approve that you just change a few lines of the software and you go run it again in the environment and you start to see. But new tactic refined so I think you know as we think about keeping pace with the ever evolving threat there's some really powerful iteration ability kind of built into how we go approach this no I think it's going to be absolutely crucial especially as if we go into. A hot conflict and we see adversary war reserve modes and behaviors unleashed we have to be able to respond rapidly and adapt rapidly and whoever does that adaptation cycle faster. We'll likely achieve the initiative in the operational advantage yeah absolutely and I think it starts to at least in my mind so much is our fighter pilot expert my background I was a contracting officer so I'm the business guy so so I also think about it from an affordability perspective right like you ask the question a little bit earlier about or made the statement. You know a lot of people think this autonomy is easy it's cheap we're going to get to these low cost drones well I think what you're hearing us describe is there's a significant upfront investment right you got to go create and train these six or seven agents. That much is talking about here and it takes quite a bit of training to kind of get to that point but there's a tremendous life cycle cost benefit of it once you've got that train if you get if you think about. Affordability from how many different CCAs am I going to put out there tailored for different missions if I've got a number of. Various trucks and I can now task them different missions give them different capabilities give them different sensors and I'm leveraging that same core collaborative mission autonomy software backbone I start to drive also a different level of affordability. That's also where credit to our DOD partners they've you know they've really stuck to their guns from an open systems perspective about how we think about this right so. We're working closely with our partners to create the right reference architectures so that you have a government owned in a transportable autonomy infrastructure that you can take from one CCA to different CCA to things of all various different sizes to think like air launched effects or. Surface vehicles and you have that common language that common collaboration that you can extend that mission set and you can extend frankly that affordability across a number of different platforms. But you know it's not one and done I mean once you feel that you'll need to continue to improve it whether or not that's in response to adversary behaviors and new capabilities or the addition of new capabilities ourselves or as you continue to grow the sophistication of the model so there always be an element of upgrading modernizing new drops and debugging frankly that will need to occur as well as a demand for additional training data and training cycles for the reps and sets to be able to get that autonomy to that that whatever that new drop is. It is trusted and capable so we can feel to go to combat with it so it's it's not necessarily a one and done will need to be remain committed to the sustainment and modernization of autonomy over the course of time. Yeah absolutely we really do it as continuous development there's constant new features new behaviors new tactics that will evolve as we actually start to feel these these things you know in different different use cases different mission sets get the sets and reps start to to build that build that backlog of knowledge and understand that yeah there's there's absolutely continuous development continuous support it like any weapon system is going to have to be sustained and maintained and improved upon. We don't expect that our tactics stop on that 16 or F 22 right this is again these are we're using them in this evolving battle space and those tactics will continue evolve it's just now instead of putting pilots in the skies north of Nellis or near Falon to go learn these new tactics they're learning them in in virtual environments with lines of software code which there's advantage that it's instantly incredible I don't need to get squadrons worth the people across the Nellis ranges I get there's no F.T.U. for an F.N.G. for autonomy right you just scored it into every drop yeah right and now instantly everyone is as smart as what we just last flew at red flag and we now have new tactics right and so this is the key power that Ryan was talking about that again huge upfront investment to dissect and get people to understand what autonomy is and to learn how to trust it and build out those core tactics build out that baseline capability. But then you rapidly see this explosion and capability and a significant decrease in time to develop new capabilities because you built up this these kind of foundation foundational building blocks all of that is what we've seen over the last decade and it that's the part that's really exciting. So how do we how do we introduce collaborative mission autonomy how do we how do we get this to the field begin developing the trust so that it can change the way that we fight. Yeah this is I mean this is the crux of the problem that I think everyone in D.O.D. is facing right now is okay great idea is great S.N.T. development over the last 10 years how do we get into the field right and part of that is we got to get pilots comfortable with this idea of autonomous wingman and I think there's classically been well I hate that's going to be really hard because a silk scarf mentality or you know resistance to technology we haven't really seen that too much with the pilots we've been talking talking to and working with right. They know that they need help they know that this is a viable way to go help them right what we need to help them understand is is how to why can they trust it and what we see there is we got to get it in the hands of the operators to go start practicing and playing with. Yeah and it doesn't have to be in the real vehicle right again the power of autonomous line lines of software code so I can I can put into simulators but the more I can get people kind of exposed to what the capability is and how it functions. It doesn't require us making them all PhDs and AI we never when knows you can't do that but again it's no different than learning how to trust your RWR or your sensor fusion algorithms or something else right it's getting them exposure to this idea building out these capabilities. We see that the services are starting to go down this path they're starting to stand up the DT and OT infrastructure that supports these kind of activities and supports kind of the. The distribution of knowledge that comes with that kind of that that tactics role that comes with it so I think that's a key part. So you know it's interesting we can fly with these and get them get pilots habituated and learn how to use them and learn how to trust them in the simulator but obviously the real world is a is an important element of proving out that technology because there are elements of the physical world that you just simply cannot model and so you've got to be able to move that into the physical world. The challenge that I see is really has to do more with national airspace and then how we operationalize these autonomous collaborative platforms are they the squadrons are they at their own individual squadrons where do we meet up with them do we take off with them information are you having any issues or is the airspace you're flying in right now more specialized and so you're not run up into issues with ATC. We have not encountered those direct issues yet I can safely say that we know that they're on the horizon and so this for operational squadrons for sure yeah right like they will have to and again it doesn't even have to be an airspace right if the if we're going to expect these little buddies CCAs to fly in a cap in the weather with a whole bunch of aircraft while we're waiting them before we push they need to make sure that they're on altitude and everything else just. It doesn't need to be at the airspace to make sure that you're on altitude and the right heading and everything else and this gets to a core capability that has to be in the autonomy which it has to be extremely predictable and explainable when it needs to be right doesn't always need to be that when I start trying to be decoys and deception and and all your your deeds that you describe right there I may not want that level predictability explainability. But I need to be able to stop it's not you have a need to be able to dial that explainability predictability for when I do need it and that will be when you're flying through FAA airspace and this is one of the places where Collins and RTX really has a strength. Yeah I would just pile on that I think that those are challenges maybe still ahead of us and we don't see those as certainly unsolvable challenges right you know we as an industry community have been flying remotely piloted things for decades right. And at Collins our pedigree you really is commercial aviation right we produce a number of FAA certified capabilities across the across the company and in fact we've in the past certified remotely piloted FAA certified flight management systems for users across the globe. Yeah I mean in ultimately the National Airspace element isn't a challenge for industry to solve it's something though that senior leaders within the Air Force when the department and certainly at Congress and the administration need to be aware of if we're going to be able to fully develop and exploit and train to using these platforms of full up for combat readiness. Ryan you had mentioned remotely piloted aircraft and months you grew up as a fighter pilot but you also have time in the RQ4 Global Hawk which is one of the Air Force's first operational UAVs or remotely piloted aircraft. That said it's pretty autonomous so can you talk to us about what that experience taught you and how you see that application of autonomy evolving through the modern notion of crude and uncrewed teaming. Yeah happy to I was fortunate to have kind of an eclectic Air Force career right with a little bit of background and fighters a little bit of background in RQ4 kind of an educational thing to go top that off and whatnot. My time in RQ4 for Chappin after I was did F15 for a while was interesting about it was this notion of the autonomy being what I described as elegantly simple like really powerful autonomy but it needed to be simple enough that you're sitting in the mission control element the MCE you could predict what the vehicle was going to go to you have very little control over what's going to do but your job is the pilot was to think ahead of the robot and know what the robot might do. So that you were prepared to go kind of let people know or counter act with the robot want to go to you as you had to you're monitoring predicting and monitoring. Yeah right you had to understand really this this okay I think it's going to go to this and that means that autonomy has to be simple enough that the human can follow along in the playbook but elegant enough to go ahead and respond to all the kind of different contingencies. That notion of like elegantly simple autonomy this notion of hate piloting pilotage pilot skills situational awareness it all still applies in these autonomous vehicles. It will be more challenging as we try to put the as we put the pilots that are controlling these things and move them off the ground at 1G and we put them into fighter aircraft the quarterback is now trying to fly their own vehicle while they're also supervising and now it's not just one RQ4. It's eight or more little buddies all trying to go accomplish this really complex mission. So that's that's part one is this notion of elegantly simple trustworthy predictable autonomy when I needed to be. The other part that was really interesting to me is the as a deep as a director of ops and as a squadron commander in RQ4 unit was the sociotechnical element of autonomy as well right this relationship between the technology of autonomy and then the social elements. That's the importance that go in and are influenced by that technology and that that reciprocal relationship that occurs when you say social elements can you talk a little bit more about that so our audience can really kind of get what you mean. Yeah exactly right so so in this case the social elements are like squadron structures it they're also tactics it's how the people use the technology. And so what we see is when we in you act more technology it changes the social elements of what the structures are and how people operate. And their tactics is an example of social constructs and whatnot and then we also see that reciprocal relationship is those change that they put new demands on the technology. And so what we saw was you know if you just inject new technology but you don't change how the pilots are trained or you don't change. What their work flow structure is they have columns adapting to that technology once we make some start making some of those changes. Then they start requiring new technologies or new capabilities in that technology and so this becomes this this feedback loop. And really what it gets to is you can't focus on one or the other right and this is the challenge that we'll have come going for it. If we just focus on fielding autonomy technologies and we don't figure out how the pilots are going to train to use it or what's their feedback loop to build out new tactics. Or how do we get the experience with it or how do we explain how do we debrief or do we have engineers that are embedded within the squadron that can reprogram or help explain the autonomy what in it happening. All of those elements that support the way the humans use and interact and understand so that they're not frustrated. They're not doing things the old way with just additional stuff but they can actually really maximize the potential of the autonomy. That's exactly right right and then we'll see is as we do all that stuff guess what will happen. I'll start having new demands of the technology right will start coming up with new tactics. I really want the thing to do this or I really needed to help me do this kind of debrief or what if we you know. Yes right and so this is the cool part right then you get in this really powerful feedback loop of industry and the user community working together in this really rapidly accelerating innovation cycle. That I would say classically hasn't been there for some of the other kind of technologies that defense usually builds like if you build a radar you're probably don't need kind of this level of quick innovation sociotechnical kind of working together. But when we're talking about autonomy what's unique about is again goes back to that context it goes back to that unique relationship with the social parts that are using that technology where you have to get this kind of really tight coupling. Of industry and the warfighter to go really develop and unlock the potential that we have in front of us. Yeah as you said much it's that interaction that feedback cycle that I think is going to be really exciting and really interesting and I think increasingly important especially as we face advanced capabilities pure threats war reserve modes. I think the collection capability of the autonomy and of our future systems when combined together can provide us the opportunity to develop more rapidly our adaptations whether or not that's how we program the autonomous platforms. Whether or not that's human formations and the way that we engage our own weapons engagement or tactics and procedures or even if we discover something new and in one theater and we're able to hyper specialize to defeat that threat in one theater and then we've got early warnings and signals that we can share that kind of intelligence with other theaters it might help us think more differently. And so that I think that debriefing capability not only will be important for the actual physical platforms and how the humans interact but then how we even just think about warfare as we move forward. Okay so we got to go back let's go to come back to present day though help us understand the variables it'll shape how the collaborative mission autonomy is developed and optimize because there's not any pixie dust solution and this is a frustration that I have is it unless we take the time to really fully understand what autonomy is how we make those algorithms how we have to train them what's the kind of data that we need. We say sprinkle some AI on it it'll all be okay that pixie dust isn't going to fix stuff we need to optimize the autonomy algorithms based on the mission assets involved share with us how you're approaching that. Definitely no pixie dust as we've as we try to do it it's kind of it's analogous to me like what much was just describing from those aha moments of can I get the little buddy to do this and that it's kind of what we call those one plus one equal three moments. But to get to that it's really analogous to kind of the valley of death right if i've lived in the system T world and now I need to cross this valley of death and get an operational world except maybe the valley of death here is that true adoption of what the pilot is going to do you know we we know you know from your experience we know from all of our work. With our defense partners all over flight deck work across the industry how to overcast over burden our pilots are so if we just give them this thing there's another thing that they have to do if it's not actually intuitive and helpful right out the bat it's not going to be adopted we're not going to see those one plus one equal three moments so. There's certainly an element of you got to refine the algorithms and we got to make the autonomy itself effective and I think that's going to require prioritizing which of those first six agents we want to focus on what's that first mission said we go get that absolutely right. But I think i'm also advocating we can't lose side of the human side of that in the adoption from the pilots perspective so getting both of those right and then I think it's just a matter of then the blocking tackling like getting the set the sets and reps getting it out there getting it trained getting people comfortable with it. So how does the collaborative mission autonomy respond in a comms graded environment we spending a lot of time about how we transition and field these platforms so that we can get the most out of them with their human collaborators but this kind of depends on having communication having that data link having that connection and in our analysis you know a major driver for why crude aircraft need to be in the forward edge of the battle space is to retain connectivity with a lot of these autonomous teams. So i'm curious how you see this problem of data links communication in an M so contested environment. Yeah, one of the unique things about the sidekick solution like by itself it isn't dependent on any particular comms it does absolutely need some well connectivity is designed to play through periods of degraded denied it has the ability to survive through that carry on the mission but it's also extremely flexible to whatever comms or connectivity is available to it. And it's relatively agnostic to what the platform has whether it's a line of sight or beyond line of sight. Like 16 one day metal the next day laser you know laser comms the next okay. Yeah, I had laser comms now I don't like what's what's available to me out there. It's actually one of the areas that I think our pedigree from Collins and RTX actually really helps us. We specialize in developing and delivering really connectivity in the most contest environments we actually spend a lot of time working to also ensure that in those degraded environments that we've got the algorithms that can prioritize what the message sets are what's the pack with the critical data that I need to get through you know if you're. Extremely denied degraded maybe I don't need the full band with video of whatever's going on but but I need to ensure that i'm statusing with my wingman and my crew they're happy to where I am. I think that point is so important we have to footstomp that for folks the ability to prioritize message sets in that contested environment something that we as humans do as pilots in the airspace do. But I don't know that we really take that in consideration so thank you Ryan for bringing that up. Yeah no absolutely the data pipe gets constricted it's kind of the analogous to the helmet fire right of all the messages on the audio so so how do you prioritize and pick out which one. Is the most important and and again I think the thing that's maybe Collins is even unique in is the degree to which we model that having that level of columns in our DNA. When we go to organs and we model that we actually model the link in a pure physics kind of perspective so a lot of the war games to see it'll be model as link is on link is off. And we actually put a lot of shades of gray into that to maybe denied in this part of the spectrum I have availability here I'm degraded here I can get this kind of bandwidth we have the algorithms behind that to go do that prioritization we can show that in the model and we can train that in the AI. Ryan I'm really glad that you were talking a lot about throttling that comes environment and being able to continue to operate and prioritize in that I'm so contested battle space. And we've also a major theme going through this entire conversation has been the importance of getting the pilots the operators involved early on the autonomy development process and I'd like to talk about that a little bit more because. It seems like the Air Force is moving in that directions that stands up its experimental operational units that you use and I'm really glad that they're doing that because it seems to me that this interaction is really going to be key to cultivating a robust autonomy industrial base. Yeah, I think your spot on it's I mean we see it as a critical component of how we do our autonomy development where you the way to think of it is you do not want a bunch of autonomy engineers coming up with tactics on how you're going to fly with these things with that 22 or 16s and at this point you may want to computer kind of coming up with some ideas but it's probably not going to be very good in the beginning like you need somebody to go validate what's coming out of the computer to go hey this is actually reasonable feasible we could actually probably go execute this. And that's the key part of kind of building out this early DTOT community I think the challenge will be figuring out how do they then engage with the broader parts within the Air Force right you can have a DTOT community that's focused on ACPs or CCAs but guess what that DTOT community they're probably not the ones actually using the CCAs in combat right it's going to be the F22 pilots or the F16 pilots or you know or they may not be again directly supporting them but they'll be in the fight with them. And so how do we then get the DTOT community of the CCA talking to the DTOT community of the fighter jets to go okay now we need to come up with a synergistic tactics of what this looks like and I think that's that's future growth that needs to happen that will happen as the technology gets more prevalent as we start to get people more and more excited about it that comes with getting it in their hands helping them kind of participate in the development understand that it can be a trusted component of their fight even when there are degraded. And when there are degraded comes right it's not going to just turn around and go home it's going to stay in the fight just like you'd expect your human wingman to do and it will prioritize the things that it does to try and help you with your particular type of mission. And it seems to me that that kind of interaction is going to be good for the broader autonomy industrial base writ large because it then that experimentation opens up the opportunity for different teams to be able to come together and interact with the operator so it's not just those teams that win a contract. We have the opportunity for the community to develop prototypes to come in and test them and to have that engagement interaction and that feedback loop that we've continually talked about. I think you're spot on right that's the other part the whole idea here is you don't want one particular company to have a lock on that. Well expertise right and so is these units stand up just like they do currently they will have connections to all of industry and now all of industry can go off and figure out what are the key components that they're specialized to go do. And then we have this open architecture that the government's provided that allows us to plug it all in and really provide this best of breed capability and we're no longer nobody's locked into a single solution because guess what this particular manufacturer made that kind of aircraft and therefore you get their stuff right. The whole idea is how do I get the best of everything industry can offer to go solve our warfighters most challenging problems that's I mean all of this it then creates this giant ecosystem which is really what we want to unlock. Yeah, because that creates dilemmas for the adversary because it's unpredictable they don't know what autonomy and what agents are going to be in what type of aircraft and they don't know how we're continuing to adapt and modify in advance those algorithms. Even the important role of the experimental ops units or the EU's and their tactician so the pilots the engineers and the process that you've described to build and develop the autonomy. How do you envision this flowing into the line is we've talked about it before but let's get some nets and bolts that people can begin to actually think about if they're at a unit what does that begin to look like as we operationalize this capability. Yeah, I mean again I think this from my perspective and this is coming from a guy that used to fly OT so I still see a strong role for the OT community whether it's at Nellis or Fallon or wherever right but this notion of I need some way to validate tactics. Understand those tactics work and then push them to the field early on in our journey we there are a lot of people going hey it's going to land it's going to self learn or the young wingman's going to be like I got a cool idea I'm going to plug into the eye and the next day we're going to go fly and we're like we don't do that today I don't I'm sorry the AI at this point is also not capable of doing that that maybe something we can do in the future. But that's not how we operate today and we shouldn't expect to put that level of tactics burden on the frontline pilots they don't have the ability they don't have the resource now the time to go to that and so how do we continue to leverage that OT community. How do we super charge it with this very tight coupling with industry and how do we accelerate kind of the normal tactics dissemination process it can't be tied to a three dash one TTP timeline but that's an artificiality that we can work through but so to me it's all about again kind of this. What's left pilots at the frontline focus on their daily task which is getting to be better and learning how to use this and going and preparing to fight the wars and let's use the tacticians to take their good ideas to take all the data coming out of this to interface with industry. Do that validation and quickly get disseminated to the field. Yeah, I think it's also important months just to focus on I think the EU is going to bring also more than just the tactician piece of it mentioned a little bit earlier the full set of it is that you also think maybe have to develop to support this thing differently like if I want to deploy this from some far from dirt airfield. How do I provide for the mission planning the debrief the connectivity all of the things that go around it we talked to some a little bit earlier how do I how do I create this DevSecOps pipeline of how I continuously do these updates what's that infrastructure look like I don't go train and continuously keep fresh the not only for the pilots and use the autonomy but perhaps the organic software factories that are going to help maintain it or the industry partners that are going to be there along the way to help maintain it so I think there's kind of a full. Yeah, to include logistics and maintenance and sustainment and how do we refuel these things and who launches them how do they launch do they come back who catches them. What do we have to do with spare tires do we even worry about spare tires. How do we mission plan how do they get incorporated into the ATO so as you said there's a thousand things that have to happen that we need to think about that will only occur as we begin to move beyond. And it's important is essential that we look at the tactics development and that collaborative teaming and we also need to think about the daily nuts and bolts. Well gentlemen, thank you so much. This has been a really great conversation before I let you go we've got a ton of different companies small and large that are all chasing autonomy and AI and specifically those agents that will be going into ton of collaborative platforms. How should users how should the government how should requires how should we measure maturity success and value when it comes to the ability to deliver on these promises then to field and support warfighter outcomes. Yeah, I think lucky to me it starts with does your product provide operational utility right does it does it help to achieve those one plus one equals three kind of elements. So I think there's going to be a measure of you know the maturity of the collaborative mission autonomy software what it brings what it offers. And then I think there's certainly aspect of you know as we think about speed to rank right speed to fielding I think there's a maturity aspect there that measures goodness. And then does it provide all the benefits that we've talked about from an open systems perspective does it create kind of that. That open marketplace to take best of breed new tactics is it flexible and extensible to not only keep pace with the threat but to go live heterogeneously on on all the things that we can envision that is going to take to be successful in these missions not only Air Force things maybe other service things maybe partner nation things. These are all things that were focused on at Collins like we've said a couple times we're 10 years into this we feel pretty good about where we are in the technology. Where we're now really working to kind of leverage the full breadth and depth of being part of the RTX family to bring all of that expertise of how do I make the most out of the tactics I'm developing here understanding the most exquisite sensor models on the most exquisite factor models all the expertise that we can put into it. I think it's one of the things that really makes us unique I think we've got the we've got the core technology in the autonomy we got a lot of experts work in that we're pulling we're pulling from a lot of expertise across across the company across industry and continuing to iterate and make our product better. Yeah, I would just I would pay you back and just really hammer on that point again if it's that delivering useful and impactful autonomy that helps the warfighter right and that requires that holistic approach to understand how that autonomy fits amongst all of those different components that'll be in the future battle space both on that individual CCA but also all the other vehicles and all the other com links and all the other sensors and all the other weapons flying through the battle space. If you're not doing that then you're just kind of building autonomy for autonomy's sake which may seem really interesting but we've all seen that in current fighter jets to where you get a kind of hit that thing will probably look really cool in that lab that technology that you know particular pop up window or that feature on the radar I never use it because it doesn't actually do what I needed to do in the fight and therefore you don't you don't use it and you don't need it and we need to make sure that we're focused on the things that provide that useful and impactful capability to make. Our warfighters better and I'll just add on top of what you both said that demonstrating it the show me element of it is going to be crucial because power points are nice and there's a huge element of the autonomy and the algorithms that may be inside black boxes but if you can't connect if you can't interoperate with standards if you can't plug into the open mission system and you can't actually fly and deliver what you promise it doesn't matter so the EU use the experimental operational units I think we're going to be core to weeding out those who can deliver showing those who've got the potential what they need to do and then seeing those that just need to go back and start back from scratch. Gentlemen thank you so much has been a great conversation it's been a lot of fun to and we look forward to having you back on the podcast. And with that I'd like to extend a big thank you to our guests for joining in today's conversation I'd also like to extend a big thank you to you our listeners for your continued support and for tuning into today's show. If you like what you heard today don't forget to hit that like button or follow or subscribe to the airspace advantage you can also leave a comment to let us know what you think about our show or areas that you would like us to explore further. As always you can join in on the conversation by following the Mitchell Institute on X Instagram Facebook or LinkedIn and you can always find us at Mitchell Aerospace Power dot org. Thanks again for joining us and have a great aerospace power kind of day. See you next time.