ENG

23.10.25 22:28 24 0

DEVIRO technical director Denys Chumachenko: Drones will gain artificial intelligence in two years. But not all of them

Drone Industry

Denys Chumachenko, Chief Technical Officer at the drone manufacturing company DEVIRO, began working with UAVs long before they became mainstream. He gained his experience abroad in the early years of the Russian-Ukrainian war. Denys decided to apply it in practice back in Ukraine, at a time when the number of drone manufacturers could still be counted on one hand.

Back then, the topic of unmanned systems sounded like science fiction; no one even knew whether it would catch on. Today, it’s hard to imagine our resistance to the aggressor without drones. UAVs, and the countermeasures against them, are becoming increasingly complex and "smarter," and with the advance of artificial intelligence, they will gain new avenues for deployment.

As part of the Drone Industry project, Censor.NET spoke with Denys Chumachenko about where the sector is headed, whether there is any ceiling to it, and which promising directions will develop in the near term.

– Tell us more about your background: how did you end up in the drone industry, and why did you choose this field rather than, say, classical aviation?

– I graduated from the Kharkiv Aviation Institute with a major in "design engineer for aircraft engines and power plants." After graduation, the job placement process began and everyone was being offered to go to Russia. I refused and stayed on for postgraduate studies. That was in 2014.

One day, my academic adviser told me, "Your PhD isn’t going to work out, these are hard times and they won’t let you defend. I have friends working with unmanned systems; go to them." I had to go abroad.

Back then, UAVs were a whole lot of nothing; I understood nothing. My adviser gave me two books to read. That was it. I learned operations, technology transfer, maintenance, and flight work hands-on.

Eventually, I decided to return to Ukraine, realizing I wanted to work only in this field. I started looking for jobs at Ukrainian companies. This was late 2017, and there were very few firms at the time — five or six.

I sent my CV to all of them. One well-known Kyiv company responded. I accepted the offer and started in 2018. By 2023, I had worked my way up at that firm from technical writer and technical documentation engineer to the company’s technical director. After that, I changed jobs, ending up where I am now — at DEVIRO.

Broadly speaking, I’ve been in unmanned systems since before it became mainstream.

– Today, no one is surprised by drones. But back at the start of your career, did it feel like UAVs were something fantastical, obscure, and of doubtful utility? Especially given how expensive they were?

– They still don’t come cheap. We’ve simply come to understand that UAVs save lives, and that’s not where you cut costs.

Otherwise, back then, almost no one had heard about them. When meeting people, if they asked what I did, they were baffled, "what are these unmanned systems?" There was a sense of being in something unique. At the same time, it felt like a strange technology, no guarantee it would stick around for years, but, as we can see, it took root and works successfully.

– We’ve touched on classical aviation, which is also your university specialty. Can UAVs replace classical aviation? Are these sectors in competition, or is there room for synergy?

– If you look closely at modern civil aviation — the latest-generation Boeings, Airbuses, and other airliners — they have autopilots capable of doing almost everything with virtually no human input. And if you look at many young pilots just starting on the flight deck, a lot of them don’t even hand-fly.

So, from the standpoint of the flight itself, this is already effectively a large unmanned system. Granted, there’s still a pilot on board who sometimes does things manually. For example, the pre-flight check. And of course, the pilot can always take over control and make the necessary decisions.

But in practice, by system design and by how it all works today, it’s already very close to unmanned technology.

There’s bureaucracy and certification that won’t allow aviation to hand everything over to automation just yet, but in the future, why not? People will simply board an aircraft the way they would an unmanned taxi somewhere in the U.S. or China, disembark at their destination, and never even know whether there was a pilot on that ship.

– And what about military aviation?

– There’s a current trend to make large aircraft unmanned so the human is not on board but on the ground, safe and able to preserve their life. For example, the American "Valkyrie," which has been performing quite well.

In general, there are many doctrines stating that there must be a high level of integration of UAVs into combat aviation, where they operate alongside pilots. It could be a flight of aircraft with one of them unmanned, prosecuting the same mission. From the ground, no one would even be able to tell which one had a pilot and which did not.

– Drones began to be used actively at the outset of the Russo-Ukrainian war in 2014. Among the industry’s pioneers was DEVIRO with its now-legendary "Leleka." Yet UAVs became a crucial, if not the central, component of the war only in 2022. In your view, why did that happen: did the technologies become more accessible, or did the right people enter the field?

– No. We just got a wake-up call, pardon the bluntness.

Everyone understood this was necessary; everyone understood the world wasn’t investing money and resources in the sector for nothing. But before the full-scale war, we had this phenomenon of a "Ukrainian paper army." People were afraid to sign off on things or take responsibility, what if they got locked up later, and they only had seven years to go until retirement; just let them finish out their service in peace.

There were plenty of such cases. And the firms that existed then were nowhere near the size they are now. They weren’t sustained by Ukrainian defense procurement at all; they didn’t earn from it. Money came only from fulfilling certain foreign contracts. Everything done in Ukraine with UAVs was driven by patriotism and the desire to help the army.

In essence, they became indispensable when there was no turning back, when "tomorrow" might not come at all.

– Today, Ukraine is already an undisputed leader in UAV production and employment, a leadership won in blood and sweat. But how long can we hold that position? Russia has the upper hand in scaling projects; Europe and the United States are talking about mass-producing drones while looking to our technology; China is showcasing swarms. So, where will this arms race take us? Where is the industry headed right now?

– The sector is shifting to more employment options and cross-domain integration across UAV types.

When the full-scale war began, there were some reconnaissance drones but no strike means. Everyone rushed to devise strike kamikaze drones and develop something to deliver strikes. But everyone lagged behind on scaling production volumes.

Russia, by contrast, invested heavily in this direction. Its engineering potential and the quality of its engineering work should not be underestimated. These are highly trained specialists who know their trade. And their UAVs are at a very high level, in execution, technological depth, and in designing with a clear view to scaling into tens of thousands of units.

Many Ukrainian companies can build a superb UAV, but how to turn it into 20,000 units a year, no one knows. The component base, manufacturing technology, process engineering, and operations were selected incorrectly. What to do with that next, hell knows.

The adversary’s design standards are high. And Russia doesn’t have a so-called "menagerie" of drones. They have standardized models that scale and are produced in very large batches.

And us? Take deep-strike drones, for example. You couldn’t count them all on the fingers of a whole family — you’d have to start bending the dog’s paws too. And there’s plenty that’s unpopular, not public, or only just taking off that nobody even knows about.

Russia has only two deep-strike types. True, they didn’t design them, but they scaled and upgraded them to the maximum. All their effort went into scaling.

Same with fixed-wing reconnaissance drones. We’ve got an incredible "menagerie". Russia has far fewer types, but far higher repeatability.

I don’t agree that Ukraine has technologies no one else has. We use what the whole world knows — vacuum forming, oven curing, 3D printing, everyone knows these. We’re not using any space-age tech. What the world wants when it funds Ukraine is our experience. Operational deployment, rapid development, scaling, and modernization.

Take one very well-known American strike drone. At the start of the full-scale war, we were tossed a few as a token gesture and they proved absolutely useless for our war. The UAV was developed for years on government money, and then saw zero modernization. They simply don’t know how to do rapid upgrades.

I know of a case where, when these drones were handed over to Ukraine, there was a delay in providing the accounts needed for authorization and to start operations. Our research institute, together with two units, began flying without any accounts. The Americans arrived and asked, "How are you doing this?" They were shocked.

We have a mission and the motivation: if we don’t do it, there might be no tomorrow. That’s what pushes us forward. Everyone is interested in the experience, specifically, how to build fast and integrate fast. A normal development cycle for a medium-class UAV is 2–2.5 years. In Ukraine, the norm is 8–12 months. That’s exactly what our partners are looking for. And we already have models whose performance outstrips many global

So where is it all headed? In my view, the future lies in everything that flies being integrated with other types of unmanned systems, for example, with everything that moves on the ground. In other words, ground complexes will start operating in tandem with ISR drones, various strike assets, close- and medium-range multicopters, and long-range loitering munitions. Ground robots, for their part, will deliver payloads, conduct demolitions, and fire weapons.

Building that kind of communication is the future. It will improve combat effectiveness and save lives. The war has shown there’s nothing more valuable than human capital. Destroyed equipment is just metal; what matters is keeping people alive, because people can run out.

People must be preserved to the maximum. That’s why scaling unmanned technologies and building ecosystems for mission execution is where the industry needs to move in the near term. We’re already seeing it: brigades are developing ground robotic systems, for example, K-2. This brigade strongly advocates for ground robots, for evacuation and logistics. But it all works well when there’s an aerial relay overhead so that the ground robot can range far. In short, the future is integration.

– So we can say that in the near future, or in the medium term, there will be fewer people on the battlefield and more machines?

– Yes. And it’s already happening. For example, there’s a 15-kilometer kill zone that people try not to enter at all. Only assault elements operate there; all other units stay roughly 15 km back from the zero line.

– Haven’t drones already reached the limit of their development? Does such a limit even exist?

– There is one — physics. Many people still treat the word "drone" as something magical. But it’s a technology, and like any technology, it has limits.

If you look at reconnaissance drones at the start of the full-scale war, their median endurance was two hours. Then there was a breakthrough — four hours. Following a shift in composite materials, which enabled the use of larger batteries.

Now they’re flying five to seven hours thanks to an energy-cell technology introduced by a U.S. company. In other words, when everyone was at four hours, it was 250 W per kilogram of battery; now it’s 350 W per kilogram, and technologies at 500–550 W per kilogram are already emerging, with further growth to come.

So we keep hitting a given technology ceiling. Once a new solution appears on the market, everyone pushes past that limit and grows in performance and capability.

– How quickly do these new technologies get integrated into drones? Say a new battery comes out — how long before it’s on our UAVs?

– That’s exactly what our partners are after. Every Ukrainian company working with UAVs has its own R&D unit, with dedicated people constantly scanning for new products.

When such a person spots a new battery at a trade show, initial talks have to happen. That’s a couple of weeks. Then it’s three to four weeks after purchase for it to arrive in Ukraine. Building some prototypes takes up to another month, as does initial testing.

If initial testing is passed and everything checks out, bulk procurement lots arrive in about three to four months, and production can start.

So from the moment the technology is identified, it can be in serial production in roughly five to six months.

– Is that just how we do it here, or is this standard worldwide practice?

– Here. The army often gets criticized over testing and scientific trials as being overly complicated. But no — in recent years things have improved on that front.

Before the full-scale war, Ukrainian practice was very similar to the global one: if you wanted to change the design, you did it no more than once a year, an incredibly tedious process. Now the state has handed the baseline design and its vision over to the manufacturer. If the manufacturer is motivated to be the best on the market and deliver the best product to the user, the whole thing takes up to six months without any issues.

Those standard trials can be run at any time, formalized, incorporated into the technical specifications, and production continues under the new standards.

Nowhere else in the world is the procedure this fast. In Ukraine, in my view, it happens almost instantly. And that’s good. Not perfect, but good. Manufacturers are given latitude.

In the past, for example, there were costing-and-calculation materials (CCM) requiring the manufacturer to show the full price build-up with all process documentation. I’ll give you a real case from my practice with those CCMs: in one military department, I had to convince a person that "Obukhiv" toilet paper was used to wipe resin off composite parts after bonding, not stolen to take home so the whole family could use it.

There’s more freedom now. Yes, pricing has to be validated by an independent institute, but that’s not difficult. It’s a routine procedure now.

The ideal, of course, would be an open market like the automotive industry. If Mercedes changes a knob on the dashboard, it doesn’t go anywhere for recertification, it runs internal tests and keeps selling.

– After the surge of Russian drones targeting NATO countries, interceptor drones shot to the top of Western defense agendas. How rapidly is the interceptor segment developing now, and what does that mean for the drone industry?

– This area is developing very actively. In unmanned systems, there’s always been a "vogue" for building something. Either longer endurance, longer range with strike payloads, X-tail configurations, and so on. Development in unmanned systems tends to follow trends. Right now, it’s interceptors. Many teams in Ukraine are developing them, but there’s a price cap — $5–6 thousand, max.

Which leads to a problem: for that money you’ll get no more than six minutes of flight time, using thermal imagers of dubious quality and an equally questionable comms link. As for automation or an onboard computer that auto-acquires and homes onto the target, forget it.

So everyone rushed into this craze. A lot of prototypes were built that sort of, more or less, work. But truly effective systems are far fewer, just a handful. You can see them deployed in the defense of Kyiv and a few other major regional centers.

And it’s already clear that flying five kilometers from launch is interesting, but you need more. High speeds demand automation and greater time on station. That’s when the realization hits: you won’t meet the initial $5–6 thousand price cap.

– So does that mean even more capital investment and higher costs for those interceptors?

– Unfortunately, yes. But Russia is also changing its tactics. When the first interceptors appeared, the Shahed flight paths were fairly predictable, there weren’t large speed changes or altitude shifts. That’s why interceptions were initially quite successful.

Now the enemy constantly varies Shahed altitudes, speeds, and routes — flying over air-defense zones as high and as fast as possible. All of this is intended to render first-generation interceptors ineffective.

Interceptor upgrades are underway, but they will cost money. Yes, interceptors will become pricier, but they will retain operational effectiveness.

– But it will still be far cheaper than, say, shooting down targets with a Patriot.

– Far cheaper. And it will be cheaper than the Shahed itself or the damage it can inflict if it reaches the target zone.

– DEVIRO is now working on an anti-anti-drone technology called "Snitch." Could you tell us more about it — how the idea came about, how it works, and whether this means drones will soon become even harder to shoot down?

– Drones are already difficult to shoot down. The "Snitch" technology isn’t new — it’s a system of optical sensors that use artificial intelligence to process the visual field around the drone and make a decision to perform an evasive maneuver when a threat is detected.

In other words, the sensors "see" something that grows in size over a certain period of time. It "realizes" that the object is heading toward it, so an evasive maneuver is triggered. The maneuver is completely random. And the maneuver was programmed based on the experience of soldiers who intercept Russian drones.

If the maneuver fails, the drone returns, as it can no longer proceed with the attack.

– Since we’ve already mentioned artificial intelligence, what’s your take on it, and what are the prospects? Will drones become fully autonomous, and how long might that take?

– These are fundamentals, basic algorithms, not self-learning neural networks. It’s a pre-trained algorithm that knows what to do. Computer vision and several other technologies are involved.

Overall, AI is an exciting field, and many types of drones will move in that direction. But war has its own math. And there’s a price-to-capability balance you simply can’t exceed.

With the emergence of interceptors, all drones have seen their service life shortened considerably. And AI requires fairly expensive computational hardware. Therefore, it’s not practical to equip all standard drone designs with it.

In my view, AI will begin entering certain classes of drones within the next year and a half to two years and will scale up actively. But the segment of the market that remains manual, piloted by operators, isn’t going anywhere.

– There’s probably no point in putting AI into FPV drones, right?

– That's exactly what I mean. No one’s going to cram AI into FPV drones. But it’s a great topic for deep-strike platforms — those that fly far and integrate extensive functionality like automatic object recognition and terrain mapping.

For example, if a drone enters an airfield area and "realizes" it’s being targeted or jammed, or if a smoke screen is deployed for visual masking, AI would decide on its own which target, given the current situation, offers the highest efficiency and hit probability. If the hit probability, say, drops below 50%, it diverts to a secondary target — turns around and sets its own course. That’s real artificial intelligence.

But that takes money. Both for the hardware and for the developers who’ll spend a year training that AI.

– Please, tell us more about your products. Recently, they’ve received several upgrades, for instance, the "Bulava" has become even more lethal. What are your future plans, and how important is feedback from the military for your development?

– The "Leleka" received a major upgrade. The "Leleka-100M2" evolved into the "Leleka-100M2R," increasing flight time from 3.5 to over 5 hours. The communication range has grown significantly, and there’s been major modernization in the gimbals and payload systems. Image quality has greatly improved, making the overall system far more user-friendly.

The "Bulava" has also undergone numerous upgrades. This loitering munition started out with a 55 km range, a limited attack angle, and restricted functionality. It now reliably operates with a 3.6 kg warhead at a guaranteed 100 km range.

After range upgrades, the "Bulava" was also improved in terms of warhead mass, it can now deliver a 5 kg warhead out to 100 km as well.

There’s a night-capable "Bulava" variant. With shorter daylight hours and enemy activity higher at night in both winter and summer, this upgrade is timely.

The night "Bulava" has given users many additional employment options, increasing its tactical utility.

– What promising directions for UAV development do you see?

– I think ground platforms are about to scale up massively. They’ve been underrated, yet they solved logistics problems — delivering ammunition and food to the zero line and they’ve proven excellent for remote mine emplacement and casualty evacuation.

There are also systems that mount machine guns on a chassis, or kamikaze — they fly into a dugout at high speed and do the job.

They’ll be a trend next year. But ground platforms have a problem, communications. Beyond 3–4 km, keeping a stable terrestrial link is very hard. They’ll need relay nodes, i.e., cooperation between ground platforms and short- and medium-range reconnaissance assets.

We might see a comeback of ultra-small glide bombs. At the start of the full-scale war everyone was hyped by Bayraktar footage showing them designating and striking targets. Then air-defense measures appeared and the battlefield changed radically. Recon assets still fly and will keep flying and that’s why systems like the "Bulava" became popular: you can find a target with an ISR drone and strike it yourself.

But at long, and even medium, ranges there’s a big time lag: the recon drone locates the target, then a second drone has to fly to it, and you’re hoping the target doesn’t move. Future development could solve that if ultra-small glide bombs prove effective.