Sea Control - Sea Control 445 - Drone Swarms in Amphibious Operations with Zachary Kallenborn

Episode Date: July 16, 2023

Links1. "Drone Swarms and Amphibious Operations," by Zachary Kallenborn, Small Wars Journal, May 30, 2023. 

Hey folks, it's Jared. Zachary Callenborden comes aboard today to discuss how drone swarms might be employed in amphibious operations. This episode was edited and produced by Brennan Costello. I'd like to pause here to highlight our local chapters, whether you're in South Korea, Egypt, Singapore, France, New York, India, or the Caribbean, chances are there's a SimSec local chapter near you. You can find a full listing of local chapters and contact information on our website at simsec.org, so if you're interested, please reach out and get involved. Finally, I want to take the opportunity to recommend our partners in the SimSec podcast network, the bilge pumps, you can find Alex, Jamie, Drack, and a pile of iron brew bottles wherever

you download your podcasts. And with that, Kim Erzman. You're listening to Sea Control, hosted by the Center for International and Maritime Security. Aloha, Shoemates, and welcome back aboard Sea Control. My guest today is Zach Kallenborn, and we'll be discussing his article for Small Wars Journal entitled Drone Swarms and Amphibious Operations. So, Zach, welcome on board. Could you start by telling the listeners a little bit about your background, please?

Yeah, sure thing. And actually, first off, I just want to say thanks for having me on the podcast. It's an interesting change. One of the first articles I actually wrote was for SimSec around drone swarms for the Future Capital Ship Week a few years ago. So interesting to come back now. So for my background, a little bit of excessive titles at this point. So I'm an adjunct fellow, not a resident, with the Center for Strategic International Studies. I'm a policy fellow at the Schar School of Policy and Government. I'm a fellow at the National Institute for Returns Studies. I'm a research affiliate at the University of Maryland. My personal favorite is the officially proclaimed U.S. Army med scientist.

And I also do national security consulting on the top. About 60 publications or so at this point covering autonomous weapons, drone swarms, weapons of mass destruction, apocalyptic terrorism, everywhere from parameters, terrorism, political violence, a piece coming out in Joint Force Quarterly soon, War on the Rocks, Defense One, Nonproliferation Review, et cetera, et cetera. That research has gotten quite extensive coverage everywhere from New York Times, NPR, BBC, AP, The Economist the other day, and about a dozen or so other outlets and a dozen or so other languages. Well, thank you for coming on. Typically, we link to some of our, I guess, past publications, but that list might be slightly too lengthy in this instance here,

but we appreciate you being here today, Zach. And as a reminder to listeners, all opinions are our own and not reflective of any of the institutions with which we might be otherwise associated. So you started the piece kind of referencing some recent war games that RAND had run. What did they conclude about the potential role of drone swarms in a U.S.-China conflict? Yeah, basically. So they were looking at, you know, the classic scenario of China potentially invading Taiwan. And they found that drone swarms would actually potentially be quite decisive for basically three reasons. They're imagining that you basically have a swarm of drones that are going in advance of a whole lot of manned craft.

And they basically have emitters that make them look like manned F-16s. But their actual role is primarily for sensing purposes, to look for air defenses. So, you know, the idea is basically if they can find the air defenses, they can send that information back to the manned aircraft and the manned aircraft don't have to take turn on their radars. And so they can stay pretty protected. So you have that advantage of sensing there. We also have a major dilemma for the adversary air defenses, Chinese, that is, you know, do I shoot these systems down? What wasting potentially expensive surface air missiles? But as far as they can tell, it's an F-16. So, you know, this seems like it might be potentially quite worth it. And given that you have so many of these drones, even if you lose a whole lot of them, but you can still spot and find a bunch of these systems that can be quite useful.

And of course, you know, anytime they're shooting, shooting down these drones, they're wasting potentially very expensive surface to air missiles and depleting those stocks that then, you know, won't be available to use against man craft. So essentially, it's paving the way for the more expensive, more valuable manned aircraft and other types of systems and basically making that air defensible bubble pop. Why did you decide to examine the role of drone swarms in amphibious operations specifically? Yeah, so I'd actually been interested in the topic well before this RAND piece came out. I've been playing around with this article and kind of got distracted with other things and came back for it. But what I think is interesting is that when we think about emerging technology, I think it's important to look at what are the relative advantages that the technology gives over other existing technologies. And then you can look at where the types of missions, where those particular properties are critical. So in the case of drone swarms, what you essentially have is distributed complexity.

That is, you have a whole bunch of relatively cheap drones that can spread over a broad area. And drones are just a platform, so they can be equipped with any possible payload you want. You could have decoys, you could have sensors, you could have different types of sensors, you could have bombs, missiles, you could have electronic warfare packages, you could have communication relays, you could have everything in between. And even within a single drone swarm, you potentially have all of that. So you look at where do you see, actually, I should note also that, you know, that comes with some trade-offs as well. With some of these drones, they may be relatively small, and so their payloads may be relatively narrow. So you look at where the types of missions and where the types of scenarios where you need a broad amount of units, cheap units that are disposable, spread across a broad area with limited types of payloads. And so amphibious operations strike me as one of the ones that very clearly fit that.

You know, you can imagine, say, a landing force that's coming in that, you know, is relatively lightly armored, relatively lightly defended in an area like that. you know, having a bunch of drones spread out, they don't necessarily need big payloads because if soldiers are packed butt to butt, small bombs, that's going to wipe out a large amount of activity there. So I think that's a big part of what got me thinking about amphibious operation. You know, imagine what would D-Day look like if you had a bunch of drone swarms coming at these landing craft, like it would be devastating. What do drone swarms offer for a force attempting to fend off an amphibious assault? Yeah, so multiple things. Essentially, as I mentioned, that you could imagine these drones sort of spread it all across sort of a straight or an ocean expanse or

sea or something like that searching out looking for these transport aircraft and either feeding that information back to other types of drones to come in and blow the crap out of them or say you know relaying similar to the taiwan snare relaying that information back to those more expensive man fighters um so once you know where the transport craft basically is you just follow up with some guided missiles bombs what have you whatever seems most appropriate for the time because these uh those landing craft are probably especially if we're talking about like a large scale invasion they're probably not going to have all that much defensive capability because you know you just got so many troops it's going to end up being costly to add on a bunch of defenses

but even without like sort of large scale you know if you only have a small number but you don't actually know where those troops are going to land, a swarm could be quite useful. You know, you disperse your swarm incredibly broadly across that area and essentially do coordinated searches, looking for indicators of where exactly these soldiers are coming from, where they're potentially landing, and then you can, you know, move your own soldiers and defenses, depending on how quickly they're coming, to get ready for them or carry out those types of attacks. Oh, and I would add also that, you know, even if those transport craft are defended, drone

storms can be useful there. One of the things that we've seen is that what makes drone storms particularly effective is that is that a treatable nature that you can essentially just mask them on a target and basically just overwhelm the defenses. You know, there's some research that was done by the Naval Postgraduate School. It was a few years ago, so it might be a little bit different now, but was looking at drone storm attacks on equipped destroyers and basically how well they would defend it against it. And I should say that technically these weren't even drone swarms it was just a large number of drones when i say drone swarms i mean drones that communicate and coordinate with one another anyway but so they had about eight drones and they modeled what

would happen if those eight drones hit the destroyers at the time they found that typically about four drones would actually make it through all of the defenses at that time that was about you know a few years ago so that might have changed but you know but that shows that like you just mask these drones against and you can potentially wipe things out so you know you have assault force coming in. And sure, there's some defenses against them, but you may be able to sort of use the swarm to, you know, just push through it. Like, okay, you'll get half your swarm shot down. But given that these connectors are really, really vulnerable, you still may be able to hit them and cause pretty massive and significant damage to that landing force.

So we talked about the defense and you mentioned a few of the quote-unquote offensive capabilities here. What else do they offer the force that's attempting to execute the landing? Yeah, absolutely. You know, swarms can absolutely go in the other direction as well. You know, you have that landing course, you know, it's particularly vulnerable from the air. So you launch up a bunch of your own swarms, perhaps through some dedicated unmanned surface vehicle, manned mothership, that type of thing, to basically help protect and guide the sky.

I'm against this. We've increasingly seen drones being used in essentially drone-on-drone warfare. The one, the cases we've seen in Ukraine, as far as I'm aware, have been more, I don't want to say accidents, but like more opportunistic rather than like, this is a comprehensive air defense strategy type thing. But certainly you can imagine things going that way. You know, some of the Lancet loitering munitions that Russia has developed, part of the idea there is that you potentially link them together into a swarm for essentially an aerial minefield

to fend off manned and drone aircraft that come into the region. So you can imagine, say, a mothership launching a bunch of loitering munitions into the air that provides sort of coverage over the landing craft as they make it in to fight off those adversary swarms as they're coming in, as well as other types of air capabilities. Similarly, you could also imagine sort of the swarm making the beachhead in advance of the landing craft. You know, you send up a bunch of loitering munitions that travel towards wherever the amphibious assault is planning on landing to look over the beach, look for targets, search around. If they don't see anything immediately, that's the value of loitering munition. You know, they kind of sit around and look around an area for targets. And then when they see them, they find them and blow the crap out of them. You can imagine, you know, doing much the same.

What type of supporting capabilities would a potential user need to cultivate to do this effectively? Yeah. And I think that's a big issue. It's something that is often underappreciated in defense discourse about swarms and around drones as well. So you imagine folks realize that, yes, a swarm would be really decisive, really important for this China-Taiwan conflict scenario. well you still got to get the things there like okay you have 10 000 drones how are you going to get them across that massive pacific ocean that stands in your way which means you need motherships you need systems that can deploy and carry them you need the maintenance you need the repairs you need potentially power stations you need a wide range of capabilities to potentially support that now i don't think those are particularly exotic you know as we see with aircraft carriers the notion that we can carry large numbers of aircraft large distances that's quite feasible but you

still need something like that to potentially manage and distribute these drones the challenge of course is that as you have these increasing motherships they also become vulnerabilities right because if you have say we have three or four of these motherships that collectively carry 10 000 drones that means if you wipe out just those three to four targets you've now wiped out 10,000 drones. So you may not necessarily need to hit every single drone. You just hit those motherships and you wipe out large amounts. And that's a larger question, I think, to think about how we use drones. But nonetheless, thinking carefully about those support capabilities, I think it's really critical. Yeah, you kind of stepped right into the last question there. So

what other drawbacks or challenges do you see using drones in these quantities? The first one to go back to your mothership is like, well, how are we talking to the drones? Because if all the communication is going to and from these mother ships now they're emitting the entire time how are you keeping those safe but what other sort of drawbacks and challenges do you see exactly i think that issue of communication is going to be a critical one i sort of alluded to it but what makes a drone storm really a drone storm is that communication and collaboration between them so how do you do that how do you prevent that from just showing like sending off a bunch of emissions that will make it really easy to track and destroy them,

as well as talking back to the mothership. I think there are some solutions in that, but that's still a challenge that needs to be worked through. I think one of the big ones is looking at increasing levels of autonomy. As you get more and more drones, it's going to be increasingly impossible for operators to really maintain really detailed command and control, you know, fly over here, shoot that, that type of thing.

So I think necessarily as you scale your swarm from hundreds to potentially thousands or even tens of thousands, you have to have large amounts of autonomy on those systems. They have to be able to make decisions by themselves. And we know that that's at least a capability that is out there. You know, basic DJI drones that you get on Amazon, they have pretty sophisticated autonomy where they can do, say, waypoint navigation, where you move from point A to point B to point C following a GPS route.

But of course, that's a dependency there. do we have gps in that type of conflict scenario how reliable is it that type of thing um the problem is it's also not very at least at the moment it often is very brittle and not very adaptable you know with current many of these current artificial intelligence they tend to be fooled fairly easily where a single pixel change can say to convince the system that a stealth bomber is actually a dog now that's a very narrow circumstance where that level of error types to happen. But battlefields are dynamic and adversaries are trying to fool us. So it's quite plausible that these systems make mistakes. They get fooled. So that's going to be a pretty big

challenge. Likewise, there's sort of the adaptability problem. Say, you know, with GPS-based autonomy, with that, you don't really need to worry about jamming emissions, that type of thing, because you're following that GPS route. However, that's also not adaptable at all. If you're following a GPS route and an enemy shows up unexpectedly, your drone is just going to keep following that GPS route. It's going to fly nicely overhead and just kind of potentially just sit there and wait to be shot by your enemy. So that's going to be a little bit of a challenge, but that potentially is a way to handle it. I think likewise, one of the bigger, longer term challenges that it's not obvious to me what the solutions are, are looking at sustainment, particularly in

that China, Taiwan scenario. So if you're using these 10,000 drones and you're accepting like, yes a big part of the value is that like you can shoot them down and it'll be fine because you can still accomplish objectives the problem is you're still getting shot down so you may have absolutely a tactical victory there where you lose half your drones you still accomplish your objective but how does that how do you maintain that over time not just like that specific tactical engagement but you know days weeks months potentially even years especially in a Taiwan China scenario where the United States has to go across the entire Pacific Ocean to get any resupply there. Where do those drones come from? Do you have the defense industrial base to support

that continued production? Do you have the transport logistics to keep moving them closer to the battlefield? If you can't have those two things, then you might get that value of the drone swarm and that initial engagement, but over time it's not going to be sustainable and you're not going to be able to continue that. We've seen, for example, in Ukraine, there's a recent RUSI report that estimated that Ukraine was losing about 10,000 drones per month. I'm not terribly concerned about that. Russia is using something like 20,000 artillery shells per day. But you can imagine in a Taiwan scenario where if the United States is losing 10,000 drones per month, do we have the sustainment capabilities to do that. Ukraine has all sorts of startups, all sorts of companies that

are building and developing these. What does that look like in the United States? How are we getting those systems across the ocean to get there? That's going to be a critical challenge, I think. So I'd like to pick at two of those points there and see if I could get some additional thoughts from you on the first one is autonomy is like you spoke specifically about navigation, but we haven't even touched on the autonomy piece of like sensor and weapon employment, which are far trickier issues than just the navigation piece, which you've already detailed is really brittle and subject to problems with adaptability. But do you have any thoughts about the sensor and the weapons employment part of that? Because eventually, I think a commander who's making

what amounts to a political decision to employ a weapon, depending on the scenario, is going to want a human being in the loop pressing a button or at least making a calculated decision to employ a weapon against the enemy? Yeah, I think that's going to be really critical. I will say first to deal with the easy question of autonomous sensor employment. I mean, I think that's one that is relatively low risk and I believe there has been work done on that already. Actually, I know that current drone swarms have been used for autonomous sensing capability. So the one case we have of a drone swarming actually used in combat was in Israel in May of 2021, I believe, where they basically had at least a few dozen or so drones that were spread across an area,

I believe it was the Gaza Strip, looking for Hezbollah Hamas terrorists. I can't remember which specifically at the moment. And they basically fed information back to Israeli defense forces to then, you know, do the actual weapons release. The drone swarm was not making decisions on who to kill and who not, but it was us doing autonomous searching. So for that type of application, I think that shouldn't be a problem. You still don't have that adaptability issue, but there may be ways to incorporate some level of AI into that to do more adaptable searches, to potentially even recognize that a target is there so that they can then either move away or do whatever or receive additional orders, whatever is most appropriate. The weapons

employment issue i think is a much larger issue and i think that's an issue that needs to get worked through so although there's been extensive international discussion and there's increasing movements that have been seeking like bans on autonomous weapons that is requiring humans to be responsible for selecting engaging targets that hasn't actually translated into actual law even though quite a folks would like that to happen so from a legal perspective at the moment that is fine, but there is still quite a lot of risk that comes with it. Because at the moment, no state has actually acknowledged using a weapon to autonomously select and engage targets. Several states do have that capability, but they've not said that they've actually used it

in that way. So, you know, that might be a deterrent from the United States actually using them. They don't, we may not want to deal with all of the massive international information warfare type things that may come with that, getting global attention, like, oh, hey, look, you're using that. But that said, I think in the scenarios that we see with China-Taiwan scenario, the risks and concerns that activists have raised around autonomous weapons, I think are actually not really all that significant. They're real, but they're less significant. So the issue primarily that you get is the that issue of brittleness that i mentioned before or that potential for error where it you know the the drone doesn't operate as intended it hits a target that is wrong either

like it might be a civilian target it might just be like the side of a mountain or a tree or of course your worst case scenario is it hits a friendly target something like that and that's problematic not only from a humanitarian perspective you know we don't want to kill civilians but it's also problematic from a military perspective because you know no soldier wants a weapon that isn't going to work. If it blows up the wrong target, you've both wasted your munition. You've also potentially given information to your adversary about where you are and what's happening. And of course, if you hit a friendly target, you know, you may have just lost your own capabilities. So I think there's an open question about how you get that

reliability. But I think at least in the case of China, Taiwan, where we're looking at primarily like relatively static surface to air missiles, things that like you really don't have much of equivalent type of thing, like a large truck. Yeah, that type of thing. Like there are plenty of civilians that have large trucks. Not many civilians and other types of things really look like a big surfaced air missile. So I think in those cases, it might actually be fine to potentially use some of this autonomy. But of course, that still depends on really significant, robust testing and evaluation to make sure, do these systems operate as intended? And I think even with swarms, that's going to be a bigger challenge because if you start getting that

communication and coordination, you also have communication and coordination of air either directly or implicitly, right? Because if the systems don't know what constitutes an air and they don't know that that system, that the target is not really a surface air missile, it's not really a tank, it's not really a landing craft, whatever, and it shares that information, you potentially have your thousands, ten thousands of drones making errors as well, which could be a pretty big, massive concern. You know, but that said, given potentially adversary electronic warfare capabilities, I think it is a genuine dilemma and concern about how you employ and use that autonomy and in what conditions, you know, is that risk worth it relative to that

conflict scenario? You know, hopefully the conflict doesn't break out and technology develops enough that we can work through some of these issues, figure out some of that brilliance, figure out ways that we can do this while reducing risk, incorporate, say, electronic warfare attack systems into the swarm. So, you know, you basically just concentrate on knocking out those electronic warfare capabilities and then you don't really need to deal with the autonomy issue. That might be a solution and there may be other ways as well. But I think absolutely the issue of autonomy and artificial intelligence is likely going to be a dilemma when it comes to how we use drone swarms, both currently and into the future. So then the last follow-up question I'm

going to have to you goes back to the sustainment pieces to talk more specifically about the operators, because this is specific to the sort of amphibious scenario. If you want to put those people on the drone motherships or whatever we're calling these things there, there's not a combat information center that exists in the Navy that can accommodate all of the operators that we're talking about to operate the number of swarms that we're talking about. There's just not enough empty seats for these people in addition to the people who are normally occupying seats in there so one there's that you have the problem of like actually physically seating these people a console where they can do the work that they need to do two you're talking about like

birthing space and like you can pack anybody onto a ship for a finite period of time here particularly when you're talking about conflict here but i think it requires some real rethinking about are you going to build a drone carrier ship from the ground up and like what is the interior that ship actually look like? Because you can't just configure it for normal fleet operations, like, okay, well, we have these two seats for the air warfare coordination cell and these two seats for the ASW, like, no, you're going to need like 50 seats in here. And they're just going to be for people dealing with drones. Yeah, I think that's a really good one, an interesting one on a couple levels. So first of all, you know, that checks into the autonomy issue there as well,

like, because the more autonomous these drones are, the fewer direct operators you need. I mean, ideally you have your operators are more providing like strategic level command and control like the stuff that like ai in a time you really can't do you know would you rather blow up a sam site versus electronic warfare thing or would you rather attack this bridge like that type of decision like that's not something that drones can do very well so you definitely need operators for that but the other interesting thing is like so much of this is based on remote communication so you know which makes me wonder well do you even need some of the operators on the actual care itself. Why not, say, station them in Hawaii or something like that and rely on satellite-based

communications or ad hoc communication networks to do that? Now, of course, you're then creating a greater dependency on those satellites, which may incentivize your adversary to shoot them down or blow them up, that type of thing. But nonetheless, I think it's at least worth considering what that looks like. And of course, it's not only the operators, but you also have to deal with the maintenance, the repairs, who's going to charge these things, what happens when they come back from the fight? Where do those drones go? Do they have landing spaces? How are those maintained? How do you deal with that? I think all of that is an interesting issue. And then, of course, where do you get those operators from? Where are they trained in?

We've seen some RAND reports, for example, that with the MQ-9 Predator program, I think it was in the Air Force, they've had significant staffing issues. Basically, their burnout rates are really, really high because they basically have these folks sitting in these councils for like 12 hours something like that and they basically it's really stressful because they'll see something happening on the battlefield and they can't do anything about it for the most part they're just watching and they feel powerless and so you're stressed you're staring at the screen you're feeling like like you can't have any control so you have these really high rates of burnout now that's a little bit i think unique to the predator program you know if we're talking

about a taiwan scenario we're probably not looking at a managing a drone that's like flying above for like 12 hours, something like that. But it might. But, you know, you still either way, you still need to recruit those people. You need to train them. You need to certify them, say like, hey, this guy actually knows what he's doing, that type of thing. You need all of that. And of course, that also creates a potential another vulnerability as well and sort of adds that mothership vulnerability. Well, you know, what if instead of killing off these 10,000 drones, you just kill the handful of trained operators that can actually do that, folks, and killing them. After all, you know, warfare is ultimately about human beings. Drones can be replaced. Humans are much

more difficult. Yeah. And we talk about a lot of this stuff through the view of this China-Taiwan potential conflict here, but 99.9% of the time, military is not at war. We're just kind of floating around out there doing the nation's business. And those drone carriers would be employed in much the same way carriers are employed now. It's like, you're going to go here and you're going to do the things that we're using drones for right now, but you're going to do from this carrier rather than operating over the horizon from wherever the nearest Air Force base is. So I think a lot of the human factors that you talk about that, yes, some of those would go away in a China-Taiwan conflict. A lot of those are still going to pertain in day-to-day operations

when there is no conflict. Exactly. And I think it's how it interacts with conflict is also a very important and interesting question, you know, because if we say start having drones be primarily over the horizon operation um i think an interesting important issue that hasn't gotten nearly enough thought is how does that affect escalation dynamics because if you basically move all your drone operators to a city in hawaii like near honolulu or something that creates pretty strong incentives to very quickly do focus on homeland style attacks because you know knocking down those drones isn't going to do that much damage like if i'm trying to prevent in the United States from doing this or like imposed costs, then drone, like knocking down

some drones, those are replaceable. Instead, I want to go hit those operators in Honolulu and Hawaii, wherever. And if those are near city centers, you're pretty quickly getting to direct attacks on homeland, heavily populated urban infrastructure, and that could pretty quickly escalate conflicts and create all sorts of problems. Could be good for our soft friends, you know, because they're like, okay, well, that's a good mission for the special operations forces instead. But nonetheless, I think it creates some scary dilemmas for a future of warfare. Well, I'm sorry to tell that we have time for today. I'd like to thank my guest, Zach Kellenborn. Zach, where can we find you online and what are you working on next?

Yeah. So first off, thanks for having me. So I'm all over the place. I'm on Twitter. Primarily LinkedIn is where I do most of the postings. You can also just follow me on my website. I have a newsletter where I basically just spit out wherever I write in terms of what i'm working on next so i was quite excited that yesterday i just got the final acceptance for a peer-reviewed piece looking at whether terrorists could literally destroy off all of humanity pretty excited to share that one closer to this space and a little bit more i don't want to say realistic but a little bit more immediate i have a piece finishing up peer review at a joint force quarterly looking at drone counter counter measures you know as we see drones become

increasingly important so too are the ability to shoot them down which also means your ability to secure your drones against those capabilities is also really critical too. So that's a particular one that I'm really excited to share. Well, we look forward to seeing both of those out in print and thank you again for joining us. To the listeners, thanks for tuning in. We'll see you next time. © transcript Emily Beynon We'll be right back.