Drones are super-cool and fun. They could also be a terrorist’s favorite weapon, allowing them to strike from a distance in safety.
But how do you defend against tiny, almost invisible flying machines that fly fast, elevate over fences, and can carry explosives or toxins into dangerous proximity to infrastructure and people?
Support TechFirst: Become a $SMRT stakeholder
In this episode of the TechFirst podcast, I chat with the CEO of Dedrone, which “dedrones” sensitive airspaces using sensors, AI, machine learning, and orchestration of defense and interdiction activity via lasers, jamming, and yes … “putting lead on it.”
Check out the story on Forbes …
Subscribe to the TechFirst podcast: drone defense
Transcript: how do you defend against drones in the hands of terrorists?
(This transcript has been lightly edited for length and clarity.)
John Koetsier: How do you protect against drones in the hands of a terrorist or criminal?
Drones are great. They’re fun, they’re capable of very rapid flight, and they’re hard to detect. They’re also almost impossible to shoot down. They can carry dangerous, explosive, even radioactive substances. It’s kind of a nightmare for large open air stadiums, airports, and other potentially sensitive locations.
So how do you protect against drone threats? Joining us today on the TechFirst podcast is AD, he’s the CEO of Dedrone. Welcome, AD!
Aaditya Devarakonda: Hey, John, thank you for having me here and, you know, love to speak about the questions you just raised. It’s something we think about every day.
John Koetsier: Excellent. Let’s start here: How big of a threat are drones?
Aaditya Devarakonda: So, firstly, let me start by saying this: Drones are … drones do a lot of good, right? Drones, the primary use of drones is like transportation and compressing the last mile from logistics, etc. They do a lot of good.
But what they can also do is get in places that you don’t want to get in, right? ‘Cause if you’re a bad guy, you don’t want to be inside a prison and blow up something. It’s easy for you to send it through a drone.
So, it is a big problem. It is a huge problem in the industries we work in. It’s our customers, our security folks and defense national security folks. And, you know, sometimes we talk about it, sometimes we don’t talk about it, but it’s an imminent threat.
It’s the most asymmetric threat out there, for $2,000 you can breach any fence in the world … and it worries me. It’s the north star for us as a company, and we’re always looking to find ways to kind of get ahead of the bad guys in this.
John Koetsier: So, yeah. I mean, there’s a lot of things you can do with a drone, right? You can fly into an airplane or a helicopter; it gets sucked into a jet engine. You could also carry a payload that could be explosive. It could be something else, perhaps something infectious as well, or poisonous.
But let’s assume that you’re running some big space, maybe an airport or whatever, how do you detect drones that might be entering that airspace?
Aaditya Devarakonda: Right. So, John, I mean, I’m a technologist at heart, and there’s a lot of problems in this world that you can solve with technology. This is actually one of the tougher problems to solve. J
ust think about like the vastness of the 3D space on top of a large airport, and think about the size of a commercially available drone the size of a shoe box. So finding that accurately and fast enough for you to do something about the situation awareness is actually a very tough computational problem to solve.
So we use multiple sensors, in terms of capabilities, so you can do it primarily by three ways. One is you can use a radar. The other we can use a radio frequency, so you kind of listen to the RF signal between the drone and the controller. And the third one is you slew to cue a camera and use computer vision algorithms, AI/ML.
So this is actually like a perfect area for application of machine learning and enhanced sensor fusion technology, if you will, to kind of put three, four inputs together and come up with the best output and keep iterating that, or millions of iterations and keep improving. So it’s a very good use case for AI and that’s what we’re doing.
John Koetsier: Are there differences between guided drones versus autonomous drones in how to detect them?
Aaditya Devarakonda: Yeah, of course. I mean, our drone technology is running way faster than most people know. Like you know, yesterday you could go two miles, tomorrow you could probably go six miles without an RF signal.
So, yes, most people ‘careless and clueless’ — that’s what we call them in the industry, harmless people just wanting to take pictures in locations that they don’t want to be — they’re flying drones that have an RF signature, most likely, right? So they have an RF connection with their controller and they fly to that location, take a picture, and come back.
But what is available to bad guys these days is also autonomous drones. You can just type it into a GPS location, pre-program that and let it go.
So we have, by doing a multi-sensor, multi-modal sensing of the airspace, we first look for RF. If the RF signal is not there, we look through the radar and then we slew to cue our camera to kind of confirm that it’s actually a drone. And all this has to happen real time for it to be effective for the security personnel who are our customers. So, yeah. It’s a great question. Both these drones, both these types of drones can be detected based on the same system.
John Koetsier: So drones are pretty fast. I mean, you can get ones going 30 clicks an hour, 60 kilometers an hour, you know, 40 miles an hour, maybe even 70 — some are really crazy fast, they’re the kind that people actually race with. Once you detect them, how much time do you have to interdict them, intercept them or down them … and how do you do that?
Aaditya Devarakonda: Right. So there’s two parts to this question: How do you detect them, in time … and what do you do, once you detect it? So I’ll answer the first question first. So, drone speeds can be anywhere from 10 miles an hour to 200 miles an hour. It is totally…
John Koetsier: Wow.
Aaditya Devarakonda: … you know, especially if you’re looking at like larger, we call it Group 3 drones, but they’re drones that are used in the military space. I think those are pretty fast moving drones.
So sometimes you have as little as 180 seconds for something that’s coming in from 10 kilometers.
So, we’ve done the math on a lot of these models, but what is important is, again, I think, leveraging … leveraging software, leveraging the AI/ML. If you can understand the flight pattern of something that’s coming towards your target or the area you want to protect, and figure out that a bird is not going to fly like this; this is not a normal airplane. So you need to kind of be fast and kind of figure out, prioritize that up the sensing chain and come up with a solution quickly.
And once you know, you have whatever … one minute or a half a second, or whatever that time is depending on the situation … you can do one of two things. One, is you can do some of what we call in our industry EW, electronic warfare, which means you just jam the drone. Like it’s got an electric signal, electronic signal, just jam it.
But if it doesn’t have that, the second level is primarily available to the DOD these days, I would say, where you could throw a laser towards that to kind of blow it up in the sky, or simple, put lead on it.
Right? That’s another thing that people do. Or there are some new technologies coming up, which, it’s called high power microwave, where you basically fry the drone in the sky. So, it’s pretty cool to look at it at a test range, but it’s actually really scary when it really happens.
John Koetsier: [Laughing] Yes, exactly. So I kind of heard three things there. I heard laser. I think I heard you say, ‘put lead on it,’ which is basically shoot it…
Aaditya Devarakonda: Yep, that’s right.
John Koetsier: And there’s probably automated solutions for that. We’ve seen some of those anti-missile type things and a scaled down solution for that. And then, of course, microwave. I think I’ve also seen these handheld systems where you shoot like a net or something like that. I don’t know how effective that is depending on the speed of the drone.
Aaditya Devarakonda: No, absolutely. I mean, so the first bucket is electronic warfare, so you jam, so there are RF jammers. That’s the most prevalent type, I would say, jammer/takeover technology.
The net guns are great obviously, I mean, I’m friends with companies that are building these and promoting these so … the challenge with the net gun is the drone has to fly very predictably for you to hit it. And then if you’re that bad of a pilot then you’re not a real threat, to be honest.
So, you know, I think net guns are great. I think it’s eventually a layered solution, right? So you’ve got to have all these things, depending on how important the location you’re securing, you’ve got to have multiple different ways you can capture the drone.
And you try all of them based on a rules engine, right? What is your escalation mechanism? You kind of align that upfront with the security personnel and say, ‘We’ll try this first. If this doesn’t work, we go to this. If this doesn’t work, we go to this.’ Because it’s easy to say, ‘Just shoot the laser,’ but like lasers can go all the way up to satellites and take out satellites, if not airplanes. So there are problems with every technology.
This is a very, very hard problem to solve, but I think a combination of multiple sensors is usually how you kind of get to like 99% solution in some ways.
John Koetsier: And I’m guessing that you want the solution, the interdiction solution to be as automated as possible. You might be having a 100 mile an hour, 150 mile an hour drone coming in, operating unpredictably, because it knows it’s there for a negative purpose; it knows the target might be defended; and it knows, or the operator knows, I’m going to try and get around that. So you want those systems to be linked, right? This is the detection software, targeting software, and actual interdiction hardware, correct?
Aaditya Devarakonda: Yep. Yep. So I think you’re absolutely right. I think there’s a high level of autonomy required for these independent systems to operate autonomously. But given the collateral damage that these things can cause from a mitigation perspective, it’s always good to have a man in the loop.
Like, I’ve seen many deployments around the world, and you always have a man in the loop … because you don’t want to by mistake shoot something that will upset people. Or, you know, if you get it wrong, the cost is too high. So there is always a man in the loop.
But you need to, like, again, AI/ML is critical, that’s why we are very proud of what we do, because I think we need to reduce operator workload. If there’s one person or two people sitting in front of the computer looking at different screens, you need to be able to tell them that, hey, there’s a 99% probability to look at this versus all the other junk that you get when you kind of survey the airspace for that kind of a, you know, that large an airspace.
So it’s a classic problem to solve with machine learning, and I think, as a company, we have 20 million images of drones in different locations and with different backgrounds, which is always kind of learning and training our database we have. We have 300 different SKUs of how radio frequency sounds of different drone SKUs are, so, we’ll keep building on that.
That’s what the money is for — I’m probably preempting that; that’s what the money is for, because this is … this is a great problem to solve, and we know that it’s a very asymmetric thread that is of a big concern for public safety and national security.
John Koetsier: Yeah. So, so AD’s letting the cat out of the bag. Today actually, when we’re recording this, which is December 17th, Dedrone announced a funding round — and we’ll get to that — and that is a big funding round and it’s a cool funding round.
Before we get to that, what are the components of your solution? Are you more the software side? Do you have hardware sides as well? What do you actually provide?
Aaditya Devarakonda: Yeah, that’s a great question, John. So we actually started our journey as a company in this space, building sensors to detect radio frequency. So, we started as a hardware-ish company, but we quickly realized the solution is in software.
So we pivoted the company and we built the entire stack, which can take any input — from a sensor perspective, from a detection, tracking, and identification — and any output from a mitigation perspective where we can kind of cue the right mitigation, but what we want to be is the platform that brings everything together. Because I think that’s where the value is, that’s where we can actually bring the AI/ML capability to life.
And we know that this is where the market should go in the next, you know, it’s already there, everyone who’s worried about this problem knows that they need a multi-layered solution. They need an orchestration software platform that can do it better, and we’re very proud that that’s what we spend all our time doing, basically.
John Koetsier: And you have about a hundred customers right now. Talk to us about what kind of customers they are, what kind of facilities those might be.
Aaditya Devarakonda: Yeah, absolutely.
So we have four primary sectors, segments, if you will. The first one is obviously our defense. So we work with a lot of U.S. and global G4 actually, four out of the G7 countries’ militaries and federal governments. In fact, within continental U.S. we work with — not just continent, just federal government in U.S., we work with nine agencies across different, across the U.S. federal government. So that’s a big sector for us, that’s almost half our revenue, or a little less.
And on the commercial side, we have airports. We have 20 airports around the world … very obvious use case at the airport. We are market leader in the U.K.; we have like five or six airports in the U.K. right now, out of the 10 that matter. And the second vertical, the third vertical we have is prisons. We secure about 60 prisons across the world, at this point. And that is a market we think could easily double/triple for us in the next year or two.
But last but not the least, is critical infrastructure. So if you follow what happened with, you know, FBI put out some information earlier last month which said there was a drone-based attack on a utility infrastructure facility in July, 2020. So they just released that information, but they know that that was one, you know, the intention was nefarious and they wanted to disrupt the entire northeast in terms of power supply. So…
John Koetsier: Wow.
Aaditya Devarakonda: …critical infrastructure is picking up for us, big time. We, right now, have about 60, 65 locations where we’re securing critical infrastructure facilities. Think stadiums, think oil and gas, think utilities, power grid, etc. And these are primarily depending on us to make sure that we can, you know, these assets are worth billions in terms of liability and value.
John Koetsier: And not only worth billions, but I mean, hundreds of millions of people depend on them…
Aaditya Devarakonda: Absolutely.
John Koetsier: …you know, think of something like a pipeline, how defenseless that really is if somebody wants to get to it. Or an oil refinery, right? Or a generation facility, a dam. You mentioned stadiums as well, right? I mean, 50, 60, 70,000 people in a place. It’s an increasingly dangerous world that we live in. It’s an increasingly difficult world to keep people safe … really challenging. Can you also do like pop-up events? I mean, imagine a festival, a music festival or something like that?
Aaditya Devarakonda: Yeah, John, we absolutely do that. I mean, we’ve — I can’t name the locations and the customers, but we have secured a lot of one-off events, and this is where we work with law enforcement folks … be it, you know, music festival or horse racing event and things of such nature where there’s — or even like car racing for that matter, we’ve secured that as well, or a golf event, high profile golf event.
We secure all these things, because people want to just bring a drone and try to take pictures where they can’t be.
And, you know, imagine if you’re in an F1 car going at 200 kilometers an hour and suddenly a drone hits your visor … you’re dead. That’s a pile of 20 cars going into the corner of the curve and hopefully people don’t die, but it’s highly likely that they can die.
So, that’s what people want to avoid. They want to kind of, you know, what we do is we set up a couple of days before and make sure that the security personnel are able to survey what’s going on. And a lot of times people who actually want to do it during the event come and try it, test out and do reconnaissances before, and we’ve avoided. We’ve avoided a lot of these things, of not happening. Anything not happening on the day of the event, and then that’s the best outcome for us. If there is no event, that’s the best outcome for us. So…
John Koetsier: Yes, exactly. One of those challenging places where you work and you want nothing to happen [laughing].
Aaditya Devarakonda: Yup. You just want nothing to happen.
John Koetsier: You just raised a significant chunk of capital. Tell us about that.
Aaditya Devarakonda: Now, we are very proud and humbled to — we were thinking about a strategic partner to kind of think about big picture public safety in, not just U.S., global.
And we started talking to Axon Enterprise, which is the company that makes the tasers and the body cams for law enforcement. And we absolutely loved their vision, in the team. I mean, it’s run by a visionary co-founder, Rick Smith. So he’s seen markets before people have seen markets in law enforcement, so, he sees this as a huge opportunity. His team is super revved up and very smart about the industry, and they’ve done a very good thorough analysis, and we are very proud that they picked us as the best team and the best product in the market.
And we see ourselves being one key element of their broader public safety connected — connected safety mission. So we raised about $30.5 million, led by Axon. Every major investor we have has participated in the round. So it’s a ringing endorsement for our product and our team. The intent to use this capital for primarily R&D work, so that said, you know, this is a fast-evolving threat. Every day there are probably 500 new drones coming out, so we’ve got to be ahead of that … I exaggerate, it’s not 500, maybe it’s five.
John Koetsier: Yes, [laughter] I think everybody realized.
Aaditya Devarakonda: Yeah. So, so, you know, we’ve got to keep ahead of that and we’ve got to keep putting stuff into our AI classifiers so that it keeps getting smarter and smarter by the day.
And we can get then, with our footprint which we, you know, we have a hundred plus customers, as we keep growing our footprint in terms of deployments, we start getting more data and our engines become smarter and smarter, they will. So that’s hopefully the wish, no, virtuous cycle we want to get into, and it’s primarily in R&D investment in the company.
John Koetsier: So let’s end here, let’s take out our crystal balls and look three to five years in the future, or so … what do you think the threat looks like at that point? And what do you think the solution that you will need to have in place by then resembles?
Aaditya Devarakonda: Right, John. I think in three to five years, if you ask me, I think some parts of the developed world, or even some parts of even Asia or undeveloped world will start having drones for last mile delivery. That’s definitely happening, especially not-so-crowded areas.
You can’t have Amazon go everywhere, right?
So they’ll probably use drones, so with that, you know, it’s beyond visual line of sight. So if that comes through, you will start to see drone highways of such nature where people have to have specific ways in which drones fly in the air.
What that would also mean is every critical infrastructure facility … every airport, every prison, you know, these large groups of assets need to know if there’s something flying above them, and they need to know if it’s a friend or a foe, right? So for the friend, you need to be able to register that in a website, so that you can actually fly and tell everybody that ‘I’m a good person, I’m flying.’ And if you’re not registered, you need to have my software to know that you’re not the right person and then you’re an integrator. People will stop you and say, ‘Hey, do you have the right to be here? Yes/No. If it’s a no, why are you here? Please leave.’
So that is the future, I think, in three, four years. I feel like every critical infrastructure facility, if you will, in the world will need this. Because you cannot have a rogue drone hit your facility and you lose millions of dollars and public and PR problems for you.
John Koetsier: Super interesting. It makes me also wonder if part of the Dedrone or the anti-drone functionality includes drones, which you can incorporate into your sensor suite to incorporate into enabling greater visual and other means of surveilling the territory that your asset is over … perhaps even interdicting as well, but then using that in other ways for security for your facility. Does that make sense?
Aaditya Devarakonda: Yeah, absolutely. I mean, that’s one of the things we kind of keep tinkering with, right? What is the right format and it should deliver this capability. Could you put it on a drone? Could you put it like on a car? Could you put it on a fixed site? So these are different things that we kind of think about as well, the deployment configurations.
But you’re absolutely right, I think this problem is only going to get worse. This is a much larger problem and, you know, bad guys have always been there … so now they have an easy avenue to do this. I don’t, I hate to sound alarmist, but I think for every paradigm shift that happens in society, you need to make the security layer robust before it actually takes off. So we want to be that company that can assure our customers that, hey, even if there’s a drone above you, we’ll tell you, and you can do whatever you want with it.
John Koetsier: Well, congratulations on the raise and thank you for your time.
Aaditya Devarakonda: Absolutely. John, thank you so much.
TechFirst is about smart matter … drones, AI, robots, and other cutting-edge tech
Made it all the way down here? Wow!
The TechFirst with John Koetsier podcast is about tech that is changing the world, including wearable tech, and innovators who are shaping the future. Guests include former Apple CEO John Scully. The head of Facebook gaming. Amazon’s head of robotics. GitHub’s CTO. Twitter’s chief information security officer, and much more. Scientists inventing smart contact lenses. Startup entrepreneurs. Google executives. Former Microsoft CTO Nathan Myhrvold. And much, much more.
Consider supporting TechFirst by becoming a $SMRT stakeholder, and subscribe on your podcast platform of choice: