The Sea Changer | Ep 96

Intro Voices
Coming to DARPA is like grabbing the nose cone of a rocket and holding on for dear life.

DARPA is a place where if you don't invent the internet, you only get a “B.”

A DARPA program manager quite literally invents tomorrow.

Coming to work every day and being humbled by that.

DARPA is not one person or one place. It's a collection of people that are excited about moving technology forward.

Tom Shortridge
Hello and welcome to voices from DARPA. I'm your host, Tom Shortridge.

On this episode, we're talking with Doctor Christopher Kent, program manager since 2022 in DARPA's Tactical Technology Office. As he prepares to wrap up his time at the agency later this year, he's building a legacy of transforming how we identify maritime threats, track potential targets, and safeguard global shipping.

Christopher Kent
Went to school for Naval Architecture. I did my PhD in Naval Architecture, University of Michigan.

That led to a fun career designing ships and the large platforms for the US Navy. I was a professor for a couple of years before that, and then I came to DARPA. Immediately prior to DARPA, I was the deputy chief engineer for the Department of Navy.

I like to joke, I became useless when I became the chief engineer of about 800 people. And then it was quite the step up to writing policy for 85,000 people. Sometimes you get to engineer real, complex systems that are made of real stuff, and sometimes those things are humans. And it was back to engineering complex systems of humans and trying to remove perverse incentives to make sure that we remember and structure our systems so that we make the most expedient choice.

One of my jobs prior to that was a director of science and technology for the 800 person department out at Naval Surface Warfare Center Carderock. And during that time, there have been a number of DARPA programs at Carderock, so I supported programs there - actually was involved with some of the ideation for a program I ended up running here, the Sea Train program.

I was aware of the opportunity that DARPA represented to really take high risk choices. One of the things that was still surprising to me, even after I got here, was understanding the level of risk that DARPA is willing to take on, and the level of bet that we're willing to take on. It's a little foreign to most of the wider DOW, because, you know, back when I managed a portfolio, I managed 80 programs.

I didn't run them. I was like overseeing a portfolio of about 80 programs in my department. It's a very different world here. And I had the opportunity to come to DARPA to work on some real tech problems, and I spoke to my leadership, and they're like, you got to go try. So I really came to DARPA with a passion to solve real world problems with high tech bent, obviously, and try and push the ball down the field for improving our warfighter capabilities in American football.

Tom Shortridge
Pushing the ball down the field requires a team effort. It's not just the skill players with the fancy highlight jukes, spins and stiff arms. It all really hinges on the blockers ability to protect the ball carrier and clear the way for them.

Christopher Kent
There was a previous Assistant Secretary of the Navy for Research, Development and Acquisition who used to sign off his emails to the entire acquisition workforce, which I was a member of at the time, with, “I'm your pulling guard” because he would say that he was blocking senseless requirements, etc., and making us more effective as a fleet of as an acquisition workforce. And that stuck in my mind. It's not a play on that per se, but that terminology - my son is plays flag - and I thought it was an interesting one.

Tom Shortridge
That idea was on Chris's mind when he named one of his current programs, Pulling Guard, which is focused on creating a new layer of protection for maritime logistics. Pulling guard is creating a semi-autonomous unmanned system that can be towed behind the vessel it protects. This guardian on a tow line, acts as a dedicated mobile sentry capable of detecting and neutralizing threats before they reach the host ship. Being able to attach pulling guard to a ship after it leaves port and detach it before it reaches its destination, simplifies a host of challenges related to insurance coverage and international regulations.

Christopher Kent
Pulling Guard concept is to try and provide the ability for large logistic vessels to not defend themselves, but to provide a military escort that is scalable across an entire fleet. Back to World War two, we started doing commercial operations just as we had been doing them before the war. That very quickly became apparent that the Germans would sink everything that crossed the Atlantic. We changed to doing escort, and we escorted groups of about two large destroyers and a bunch of airplanes, escorted these ships for the transit, and we drastically increased our ability.

And we also were making liberty ships like mad. We still lost a lot, but we basically were able to bend that curve. We're not in that situation anymore. Are basically our destroyers have gotten a lot more complex, a lot more expensive. And if you just look at the scaling of the amount of logistics it needs to move, you can't supervise groups. So you can't do the classic convoy escort anymore. And it also doesn't scale to a worldwide problem.

I saw a real gap in what we were doing, and a gap in the people that were looking for solutions of that problem, and also a real opportunity. And this was one of the fun things in the Pulling Guard program, is that we're solving both a commercially needed solution, you know, anti-piracy, anti-terror, keeping insurance rates down, but also solving a key military need, the ability to protect our combat logistics force and the larger shipping that we need during wartime.

Sadly, as cheap as spaceflight is becoming, I don't think we're getting away from ships. When you're moving to move big, heavy stuff, ships are going to be it for quite some time. Pesky physics gets a vote.

So commercial vessels have very limited ability to defend themselves. And so you have a ship that's worth hundreds of millions of dollars, plus its cargo is probably worth many more times that. And they have no real defense other than being a big giant ship in the middle of the ocean. And part of that is, is that it's hard to get to the middle of the ocean. So that's been a suitable defense over the years. But that landscape is changing with the proliferation of USVs. And so Pulling Guard was really to address that sort of - I like puns – “sea change” in the situation and go after that.

It's also very hard to perform that defense like from a purely practical perspective, because if you're a commercial vessel, you don't want to become an armed military vessel for many reasons, but not just, hey, you may not want to, but I mean, from an insurance perspective, the insurance carriers don't want you to be a military vessel. Generally, if the wider world doesn't know governments self-insured themselves.

Basically, there's no insurance on a military vessel. If it breaks, we pay for a new one. You and I pay for a new one, at least in the case of the US. So there's functionally no insurance, and insurers aren't set up to take liability for the effect of military actions, which makes a whole lot of sense. There's also all sorts of rules about like military vessels entering ports, and you need a whole different set of clearances. And so how you get into port actually makes a huge difference.

Tom Shortridge
Putting weapons directly on a commercial cargo ship creates a cascade of complications, from insurance liabilities to port access restrictions. To sidestep those operational hurdles, the defense mechanism needs to remain legally and physically distinct from the vessel it's protecting.

Christopher Kent
So what Pulling Guard does is deliver that in a way where the military action is still owned by the military, that is running Pulling Guard and not the vessel that is doing the towing. So it keeps them clean and separate. Instead of a cruiser destroyer with 500 sailors aboard and risking 500 sailors lives, we have one or two sailors supervising a single unmanned vessel that is attached to the large commercial vessel, providing not equivalent, but close to equivalent levels of protection for, you know, pennies on the dollar.

Tom Shortridge
Detecting threats and neutralizing threats are two related but distinct technical challenges. Operating an uncrewed system in the open ocean means dealing with constant motion, glare, waves, and other visual clutter. A towed platform has to process that chaotic data in real time to figure out what belongs in the water and what may pose an actual threat.

Christopher Kent
We have five teams going on Pulling Guard. We have three teams working on the sense and sense making portion of it, which is replacing in somewhat of an automated fashion seamanship, for lack of a better word, scanning the horizon, tracking what's there, looking around, making sure that you understand your current operational picture and understand who is near and who could be a threat to you, and who isn't a threat to you, and doing that in an automated manner.

And that would be what we would usually term the more DARPA hard part of this, doing that part. There's also building a kill chain for the engagement half. In addition to that, nobody tows stuff at 16 miles an hour, which is what large ships, usually large bulk carriers, travel at that kind of speed. Container ships go in the 24-ish knot range.

So we have a second set of teams, two of them working on the being towed part of that. So the sense and sense-making teams own the drone up in the sky, and a box on the deck that contains their compute and some, let's say missiles or other effector. But all of the ones we're currently looking at are missiles or something like a missile on the deck, and then the boat that's on a line. That boat on the line is the TA2 teams.

Tom Shortridge
In addition to developing the Guardian on a tow line, Chris's team is rethinking modern object recognition by drawing inspiration from human vision. Current methods try to identify everything in a frame, which creates a computational burden that can put warfighters lives at risk when every second matters in battle. His team has found a way to speed up the targeting process through a DARPA program called Heimdall, named after the all seeing Norse god always watching for invaders, Heimdall focuses on tracking objects in dynamic environments, such as a small boat or a torpedo that may appear and disappear with the rising and falling of ocean waves.

Christopher Kent
So the way that Heimdall differs from many of the other trackers that are out there, open source ones you can go get were developed with a very specific goal in mind, which was object recognition, and they sort of flow from that idea. So they want to look at an object and identify what it is, and then track it in a sea that's fundamentally a little different than how our vision system works.

Our vision system really has two parts. It has a part that's near your eyes, your frontal cortex, that actually tracks objects and keeps your fovea focused on an object. And then there's a whole other part that looks up all the history and figures out what that is, looks at you and says, oh, that's Bob, or whatever the case may be.

But what's interesting is, is that part that says that's Bob, then turns off and doesn't think about it anymore. So the idea with Heimdall was to not do the object identification part, but more separate, and just do the object tracking part that our eyes are really, really good at. And anybody who's walked around with a kid that gets distracted by anything off in the scene, the point of this was to leverage that behavior that comes from that, what we like to call a low level object tracker.

That behavior has some real strengths. That makes human vision really good at, you know, understanding occlusion, something passing through the scene somebody passes in front of Bob. You don't actually see that person pass in front of Bob. Your mind automatically functionally ignores it, like you're walking in a crowd and you're trying to go meet with Bob and a thousand people passing in front of you. You really only see Bob. The rest of it gets ignored.

Heimdall basically is designed from the ground up to ignore all those distractions. It basically replaces the human inputs with that automatic joystick control. So it is a lot like in a video game, aim assistance that allows us to not avoid, but implicitly does not wake up some of the safety related slowdowns and roadblocks for implementation that are perfectly valid.

And in no way - we're not going around or making anything less safe. It actually just means that we're just helping the human do the human part that we already trust the human to do better. It actually helps the warfighter make a better decision as far as the engagement, because instead of the object moving around in the scene all over the place, and they're trying to decide what it is when it's dead centered in the scene, they can see clearly what that object is much more clearly with a Heimdall enhanced system.

So the human is in charge of all firing decisions and actually in identifying the target that it wants it to track. That's how the program was designed from, from day one. From our perspective, it was very important, what this was, is I see it as an enhancement to the warfighter, not as a replacement to the warfighter in any way, shape or form.

What this allows them to do, though, is engage targets, arranges that they could not do before. You can track those targets much, even much further away than you can engage them. So you still have lots of time to make up your mind.

Tom Shortridge
Because the Heimdall tracking software operates independently of the weapon system itself, it doesn't have to stay on a boat. What started as a maritime effort quickly found a use case on land.

The goal when we designed Heimdall from the program level was to assist a gunner doing the things with the same amount of information made available to the gunner, right?

So when a person is asked to engage a target with a gun, not a missile or anything else, fundamentally all of our systems help them aim that reticule. But it's a human using visual systems to engage that target. So Heimdall was to make that targeting process as good as possible. So we do not try to modify anything in the system other than we ask for a video feed in and the ability to control the Pan Tilden zoom or where the radical goes.

So we had originally targeted the system towards the Navy, and we're still in conversations with the Navy, but the Army opportunity came up. I had never talked to the Army with regards to CROWS. We talked to them in middle of December, and less than five months later, we were successfully engaging targets on a range with the Army. In early April, we started a round of demonstrations with the Army.

We started with a turret mounted crow system and some target drones in Aberdeen Proving Ground. We had successful engagements of small 12 inch drones in shots that the operators gave us feedback that they were unable to do manually.

The Army implementation that we're doing is a, I would say, a temporary installation type thing where we're putting an extra small computer box in the in the cab with the warfighters. Now, the long term goal is to then move the software, which is already from ground up, was designed with cybersecurity in mind. It is what's called containerized, which then allows us to put it in a container, be that virtual or in our case, right now we have a little box. But in the virtual world, they'll put it in a virtual container and load it into the main stack.

And then it'll just be a software upgrade that's pushed out to all the warfighters. If we are successful in fully transitioning to the Army, which is looking pretty positive currently.

Tom Shortridge
Getting an experimental capability into the hands of a warfighter is the ultimate metric of success for a DARPA program. Transitioning these systems out of the lab ensures that the engineering work done today translates into practical security tomorrow.

Christopher Kent
We couldn't do it without industry, and it's edifying to see what people and the ideas that come into this building are. It's a I probably one of my favorite parts of it is that there is always people, smart people, coming to you with their best idea. Sometimes it's not formulated well, sometimes it's not. Sometimes it's related amazingly. But it's really great to see the passion.

This is as close to the feeling of being at a university that I've ever been. I love every time I get to go back to a campus, it feels like I text my wife that it feels like I'm in a place of, you know, where dreams come true. And I feel like DARPA is as close as you can get to that in the DOW.

So I plan to stay deeply involved in the DOW after my time at DARPA. How exactly that works out, I plan to - I had always planned to remain a Navy civilian. We'll see how that works out. But definitely, definitely still in this space, be that even in academia.

But what do I want people to think of - or at least the SETAs, which are the memory in this place -think of, was, he was, you know, demanding, technically rigorous and driven to make warfighters more effective. I think that is I try and live that every day. Hopefully I achieve it. I know Heimdall is going to -it's looking really good that Heimdall is going to grow some serious legs and make the world a less safe place for our enemies, and a safer place for our warfighters.

When Pulling Guard lives up to its promise, if it doesn't run into some policy issues, but on a technical level, it should execute, that will fundamentally change how shipping and insurance is worldwide. So DARPA is a place where you can come and be here for 4 or 5 years and have a worldwide impact. That’s a pretty special thing that not very many people get to say, at least not in the engineering field. I would argue that in some other medical world, maybe so, but not not normal for us engineers. So pretty excited about it.

Tom Shortridge
That's all for this episode of voices from DARPA. For more information on Doctor Kent's portfolio will have links in the show notes. As always, thanks for listening.