Starting a new company blog feels weird — like the first day at a new school. However, the story behind Obriy Sim (short for ‘simulator’) is one that truly stands out — and it’s worth diving into. Since Spring 2022, the Obriy Sim team has worked in “stealth” mode, exercising extreme caution in disclosing anything that might benefit an enemy or put team members and their families in danger. The software distribution was (and still is) limited to active service members of Ukraine’s defense forces.
These precautions worked well, not only for the Obriy Sim team, but also for other early adopters. The nascent Ukrainian strike drone community had time to establish itself and grow. The enemy started using combat simulators only in Summer 2023 — almost a year after the Ukrainians. This time gap enabled Ukraine to train a sufficient number of FPV drone pilots by Summer 2023 and achieve a strategic advantage in the affordable, below-$300 strike drone category for almost a year.
The “stealth” mode of the Obriy Sim story also gave our team a legendary status among actual combat pilots. Today, we are opening a new chapter in the public story of Obriy Sim and the wonderful people who created it. We will still use only first names (and sometimes change them for security). And sorry, but no early 2018 team photos with that late-night pizza (the top cliche of any foundation story in a corporate blog). “The true heroes must live in oblivion,” as one good old video game tells us.
Building the (Not Really So) “Secret Government Drone AI Lab” — Why We Decided to Tell Our Story
“For some reason, people were sure we were some ‘secret drone lab,’ owned by the government,” smilingly but confidently recalls Viktor, a co-founder of Ukrainian AI lab named Twist Robotics and its wartime subsidiary Saker Aerospace (dedicated to manufacturing combat drones). In reality, three programmers and a few early AI codewriters ran Obriy Sim for most of 2022 and the first half of 2023, when the team started expanding.
Wait, but what is Obriy Sim? It is a virtual combat training environment designed to prepare drone pilots for real battlefield conditions without the risk or cost of using physical drones (and losing human lives). Obriy Sim replicates real-world terrains of the Ukrainian battlefield, combat scenarios, and electronic warfare (EW) conditions, enabling pilots to sharpen their tactical skills, test new flight paths, and train with minimal delay. The information about Obriy Sim and the people behind it was minimal… until now.
The “Stealth Mode” of Obriy Sim Backfires
As it often happens in the information vacuum, the secrecy of Obriy’s operations led to some surprising outcomes. “At some point, one of our pilot-users pointed out a fraudster offering fake ‘Obriy licenses’,” comments Rostyslav, the company’s co-founder. “Because nobody knew who we were, a dude seized the opportunity to call himself ‘the developer of Obriy.’ And the story repeated itself several times. It’s Eastern Europe, baby.” This fraudster was “selling” an ancient version of Obriy Sim, from early 2023, before the team introduced the software licensing mechanism. So how does it work now?
Obriy Sim Licensing Mechanism and How it Works
The Obriy team developed dynamic software licensing to prevent situations in which a legitimate copy of Obriy Sim falls into the wrong hands. This licensing mechanism plays a crucial role in preventing unauthorized use of the software, for example, when the enemy takes pilots and their equipment POW, or when hackers manage to access the contents of a pilot’s computer or the equipment at a military training center.
“We made our best efforts to make the reverse engineering of Obriy’s source code a complicated task,” explains Serhii, who has been a team lead of Obriy Sim since 2022. “The licensing prevents unauthorized use of the software itself. Even if hackers somehow bypass the monthly licensing mechanism, they would lose access to the updates, which are released on a monthly basis. And with our current pace of improvements and adding the new features, that will be detrimental for anyone using Obriy to plan actual combat missions.”
Looking Forward to the International Sales of Obriy Sim
Rostyslav explains another reason for Obriy Sim’s exit from the “stealth mode.” “The licensing exists only for security reasons. We don’t charge the end users any fee for the Obriy Sim license. The free access for the Ukrainian military means we had to subsidize Obriy Sim’s development from other income sources throughout those three years.” Saker Aerospace sells its drones to the Ukrainian Army via the Ministry of Defence and direct purchases by the military units. That’s where the money to sustain and develop Obriy is coming from.
However, deep R&D in the drone industry is costly. That’s why Saker Aerospace should ideally use the income from drone sales to sustain further technology improvements, not to develop Obriy Sim. Additionally, “our current product backlog for Obriy Sim is a dozen times the capacity of our development team; we would love to have more people coding, testing, and managing the process,” remarks Serhii. Even “business subsidies” that come from the sale of drones cannot cover the cost of Obriy development. Selling the export version of Obriy Sim to international partners from NATO countries might be a solution. However, how would those partners discover us if we work in stealth mode?
Seeing the Bigger Picture of the Obriy Sim Story
Established companies, such as Raytheon, Rheinmetall, Thales, Leonardo, Saab, Airbus, BAE Systems, and Lockheed Martin, as well as software integrators like Booz Allen, SAIC, General Dynamics, L3Harris, Palantir, Hensoldt, and Indra, currently dominate the market for tactical simulators. Many of these entities have a decades-long history of doing business. Mostly, these vendors create personnel learning software and digital combat practice environments for large professional military equipment, such as jet fighters, cruise missiles, bombers, and tanks. Equipment of this type costs millions of dollars. Now compare this to the $300 FPV drones for pilots, who were anything from wedding photographers to PhDs in botany and arboriculture just a few years ago.
Ecosystem Differences and the Realities of the Obriy Sim Market: Compare NATO Tank Simulator to Ukrainian Strike Drone Simulator
It’s relatively easy to incorporate the costs of developing decent simulator software into the price of a multi-million-dollar tank. It’s a much more difficult task for a $300 drone that can destroy the tank in minutes. Moreover, the tank manufacturer has enough resources, people, and time to see the bigger picture. Such a manufacturer wants to groom the ecosystem around its equipment. A typical Ukrainian FPV drone maker faces harsh competition for each dollar — the basic drone technology is not that complicated. Therefore, the manufacturers of such drones usually care for the (relatively) reliable components and their product’s sales volume, hoping that the pilots will take care of themselves.
We want to tell the story of combat drone technology and the digital learning environment that emerged and grew in Ukraine during tough and turbulent times. That is why we decided to make the development process for Obriy Sim more open, to the extent that the ongoing war in our country and constant enemy surveillance allow. Therefore, let’s get back to the origins.
The Eureka Moment: “I Need that Footage!”
It was early autumn 2015, and Viktor was upset, to put it mildly. “Do we need to always have the computer vision team at the very end of the queue?” A robotics startup he had co-founded less than a year before had a long drone queue for engineers who needed to fly the minimal fleet of their new venture for various debugging and feature testing.
The drone Viktor and his young partners-in-venture were developing was a huge heavy lifter with quite costly components. Therefore, the team had only a few actual models built and ready to fly. The molding team needed their flight time, and the engine team needed theirs. “Well, why don’t you use your computer vision to come up with a solution?” one of the colleagues jokingly offered to Viktor. And then it clicked!
Test flights on October 11, 2015. Open timecode 1:21 for Viktor’s work. The basics of computer vision remain the same, almost a decade later: analyze the pixels from the drone footage and compare them to the pixels from the reference image, assign one of “the 256 shades of grey” to it, and use reference points to speed up the process. Meanwhile, “speeding up the process” is where most innovations happened throughout those years, including the Obriy Sim story.
Luckily, in 2009, Jordi Muñoz, a robotics enthusiast who grew up in Tijuana, Mexico, and Chris Anderson, an English-American author, entrepreneur, and former editor-in-chief of Wired magazine, released a robotics controller software called ArduPilot. Initially focused on a fixed wing, the team launched ArduCopter (focused, well, on copters) in 2011, and in 2014, it introduced the SITL simulation.
The SITL: Obriy Sim Takes Its First Shape
If you’ve made it this far, chances are you already know what SITL is. But just in case someone accidentally wandered in from Google, let’s break it down real quick. Software-in-the-Loop (SITL) is an environment that helps developers simulate flying a real drone using a computer game (nerds would call it a “simulator”). Still, the drone’s “brain” (the autopilot software that runs the drone’s flight controller) is also running on your computer, not on a real drone. This approach enables the developer to emulate the behavior of a drone using SITL. What SITL still leaves out is the physics behind the environment (the “virtual airspace”) in which this drone would operate and the graphics engine for visualization.
This video shows the Boeing 747 flying with the ArduPilot autopilot using SITL in X-Plane — the legendary flight simulator for commercial and recreational pilots. The authors make a funny but essential disclaimer: “The real 747s do not fly on ArduPilot” (yet). Replacing 747 with a more miniature combat FPV drone will give you a glimpse of how Obriy Sim operates.
“ArduPilot had no Gazebo or AirSim integration back in 2015, so I wrote some basic physics and visualisation code, connecting it all via Gstreamer. It wasn’t some ultra-realistic rocket science; just a code that fit for testing my computer vision solution without needing to fly a drone each time. Then, in 2016, it was turbulent; the team I was part of was running numerous pivots and looking for investors. Then we created Twist Robotics with Rostyslav, and I focused on applying my computer vision solution to managing industrial warehouses, not flying drones. The code just sat there waiting for better times,” recalls Viktor. Over time, other teams also suggested using computer-generated video footage to train various computer vision models and algorithms.
People Instead of AI Simulation: How the Full-Scale Invasion Forced the Obriy Team to Change Its Approach
“The better times” were not so “better.” However, when russia invaded on a full scale in February 2022, it soon became evident that drones would be the primary weapon of the Ukrainian resistance. Viktor dug up one of his old office hard drives and found the code he had written in 2015. “It was evident some major changes happened to ArduPilot throughout those 7 years. However, I had no time to reflect on those changes. We created several parallel computer vision solutions for various units and drone manufacturers. Serhii took it on since then and turned my quick-and-dirty temporary solution into military grade software.”
Initially, the project had an old good “SITL tune”: the team needed a virtual testing environment to crash fewer drones and do more coding instead of test-flying in the sun, rain, or snow. On top of that, the regular russian air raids using cheap Iran-made Shahed drones made airtime for test flights very limited as Ukrainian air defence banned all drones (including friendly ones) from the airspace during the air raids and the ensuing air battles.
Over time, the team started training the first customers for what would later become the Saker Scout and Saker Hunter families of UAVs. “People asked if we could provide some simulator for newcomers, so as not to risk the expensive equipment,” explains Viktor. The Saker team replied that they could and asked the Obriy team next door if they could “borrow” the software for their virtual test flights from their colleagues. Suddenly, Obriy Sim was training real humans instead of neural networks. It created new challenges.
The Docker Containers Appear
Serhii is a quiet, smiling introvert you would mistake for a college kid on a street in downtown Kyiv. But under those chill, millennial vibes, a skilled software architect resides. Under his leadership, Obriy became the software used (and loved) by thousands of pilots across the Ukrainian frontline.
The code that Viktor and his colleagues created throughout the 2010-s was an excellent achievement for its time. However, it required Viktor’s Linux desktop to work (and it was the only machine in the entire World on which it would run properly). Sometimes it functioned with Viktor running commands via the MAVLink protocol in the background for real-time debugging.
It came as no surprise. In 2015 Viktor meant his solution for machines and neural networks, not humans. You could not just download and install it from Steam. Serhii, Ivan, Kateryna, and Dmytro refurbished the code, creating neat Docker containers. However, the next step was determining the underlying architecture, which would shape Obriy Sim’s ability to work well on new, powerful GPUs and the Obriy team’s ability to introduce new features quickly.
Serhii and the Team Try Gazebo
Developed by the Player Project (and then Open Robotics) since 2002, “Gazebo underwent some major changes right before the full-scale russian invasion had started,” explains Serhii. In 2018, the ArduPilot team launched an official Gazebo plugin, which allowed what Viktor was looking for back in 2015 — running SITL-simulated firmware in a Gazebo 3D environment. After testing it in Gazebo, Ivan created a simple script to automate the repetitive tasks of debugging the drone firmware.
Jaeysoung Lim gives a demo of the Ignition Gazebo framework under PX4 SITL. Launched at ETH Zurich in 2011, PX4 is another autopilot software, developed independently of ArduPilot. The Dronecode Foundation, a part of the Linux Foundation, currently maintains it. Like ArduPilot, PX4 also uses a software-in-the-loop mechanism. To make things even more complicated (as it sometimes happens in the world of open-source software, Santa Barbara), from 2018 to 2022, “Ignition” was the name for a next-generation fork of the original Gazebo. Then, following the trademark dispute, the dev team dropped the “Ignition” name, and this new project became just “Gazebo.” In contrast, the older original Gazebo project was renamed “Gazebo Classic” to be ultimately discontinued in 2025.
As the development of Obriy Sim continued, it soon became clear that while Gazebo was an excellent tool for robotic simulations, it wasn’t meant for strike drone missions. The rendering engine options were limited at that time — at some point, the Obriy team struggled to create trees and bushes in Gazebo.
Also, at that time, Gazebo required Linux and ArduPilot “to play the sim.” While this was okay for experienced pilots looking to acquire new skills, rookies at military training centers compared it unfavorably to the commercial FPV simulators like Liftoff, The DRL Sim, or FPV Freerider (which run on Windows or Mac and can be easily installed from Steam, just like any other game). Those people were unhappy about having to set up additional software, and we have heard them.
Obriy Experiments with Microsoft AirSim
In 2018, Microsoft Research released AirSim — an open-source robotics simulation platform based on the robust and ultra-realistic Unreal Engine developed by Epic Games, the studio behind Fortnite. In 2019, during Google’s Summer of Code, software developer Rajat Singhal undertook a project to integrate SITL with AirSim. “Upon discovering AirSim, we started playing around with it, improving physics and trying to make it simulate the firmware for our drones,” recalls Serhii.
Andrew Tridgell gives a demonstration of AirSim on SITL. In 2022, the Obriy team used AirSim as a reference for creating the first version of what would become Obriy Sim.
Switching to the Unreal Engine 5 and Going Beyond AirSim
Microsoft based AirSim on Unreal Engine 4. In 2022, four years after AirSim’s initial release, Epic Games introduced UE5. Serhii and the team decided Obriy needed UE5 for several reasons. Firstly, UE5 not only takes advantage of newer and more advanced GPUs, but it also allows artists and developers to use more advanced and efficient tools, to a certain extent. This allowed for creating more beautiful and detailed visualizations on a broader range of hardware.
William Faucher, the VFX/CGI artist in Marvel’s Black Panther and HBO’s Watchmen, explains how Nanite smartly revolutionizes the LOD (level of detail) of the simulated world without requiring a lot of computational power behind it. We were looking for that computational economy in Obriy Sim!
Secondly, the Nanite virtualized geometry feature of UE5 allows creating and importing complex environments with cinematographic quality in a matter of seconds. Importing is key here. That’s how Obriy Sim got thousands of square kilometers of photorealistic Ukrainian warzone area, which we exported from the actual satellite images. “We still haven’t used even half of the features available via Nanite for Obriy Sim. However, the team is slowly but steadfastly implementing something new each month,” explains Serhii.
Clouds Do Matter
Thirdly, UE5 leverages the Academy Color Encoding System (ACES) for its filmic tonemapper and rendering pipeline. This ensures precise color consistency across different environments, devices, and in-game cameras. As a result, we’re able to render highly realistic skies and cloud formations.
This instructional video explains how Unreal Engine’s ACES tonemapper supports realistic sky and cloud rendering — an essential technique we used in the Obriy Sim to train drone operators in visually complex combat zones.
Why do clouds matter? Not the IT kind — actual atmospheric clouds. Their appearance varies significantly depending on geographic location, and even more so in combat zones, where smoke, fire, and chemical trails (no conspiracy intended) from rockets and munitions alter the sky’s visual characteristics. Accurately generating these skies helps Obriy Sim users develop visual orientation skills in complex environments — an especially critical capability for air defense drone operators.
The sky over Ukraine is painfully beautiful. And we continue the Obriy Sim story to help defend this beauty. This is a real render of sky over Gorgany mountain ridge in Obriy Sim via UE5.
Gazebo Strikes Back! Welcome ROS, Its Younger Sibling
Despite the Obriy team having fun with its new love, the Unreal Engine, our story with Open Robotics is far from over. While Obriy Sim team took a different route with AirSim, then UE5 and beyond, the engineers in charge of developing actual drones at Saker Aerospace, our parent company, really liked the Robot Operating System (ROS) created by the Willow Garage, Stanford AI Lab and OSRF (later it became Open Robotics). However, that’s a story for another blog. Still, let’s briefly cover what happened with our corporate parent from 2022 to 2025.
Team Launches Saker Scout, Saker Hunter Combat Drones, and Oscar Autonomous Navigation Module
“Like many other Ukrainian teams, in February 2022, we started by buying some basic equipment and drones and offering technology support and advice on setting up this equipment. Our efforts covered various friends and family members who joined the defence forces of Ukraine,” recalls Viktor.
At first, there was hope that the full-scale invasion wouldn’t last more than a year or two, after the russian original blitzkrieg plan flopped so miserably. As time passed, it became clear that the war would last for years, if not decades. Given the team’s experience in machine learning and building UAVs, launching a manufacturing business for strike drones was only a matter of time. However, it took a lot of team effort to focus, as in Autumn 2022, the team became involved in more projects than it could deliver.
In 2023, Saker Aerospace launched three distinct products: the Saker Scout multicopter, the Saker Hunter fixed-wing drone, and the Oscar autonomous navigation module. Obriy became a standalone project with a dedicated team.
How Does Obriy Sim Now Fit Into the Saker Ecosystem?
“Owning the entire technology chain — from R&D and manufacturing to software development and learning environments allows us to create the unique end user experience and make sure pilots and units using Scout, Hunter, and Oscar know how to deploy all features and tools we created for them,” explains Rostyslav. With the dedicated version of Obriy Sim for Saker Scout and Saker Hunter, pilots can experience all the tools they will use in real combat via a simulated digital environment.
“We are using the new architecture of Obriy Sim in tasks for which we created it in 2015 — improving the autonomous flight machine learning algorithms for all three hardware products we currently offer,” explains Viktor. Moreover, the Ukrainian military and volunteers started using Obriy Sim as a documentation tool and then as a reference guide for sharing the best drone tactics for individual pilots and drone swarms.
Obriy Sim as a Tool for Recording and Sharing Digital Combat Experiences
“Some of Obriy Sim pilots are not here anymore, they are K.I.A.,” Viktor admits quietly. “And the best memory for these people would be to preserve the tactical tricks and wisdom they came up with for the future generations of Ukrainian pilots to learn and apply.” However, how do you “record” an experience that involves drone firmware hacks, piloting tricks, and is influenced by multiple external factors, such as weather, terrain, and the enemy EW?
Based in Bratislava, Slovakia, a photogrammetry startup, Capturing Reality, launched its RealityCapture software public beta in 2016. RealityCapture and its mobile app, RealityScan, help any willing smartphone owner create a 3D object consisting of thousands of tiny triangular polygons, collectively referred to as a “mesh.” Ironically but not surprisingly, in 2021, Epic Games (yes, the maker of the Unreal Engine, which we use for Obriy Sim) acquired Capturing Reality. Epic currently owns a vast ecosystem for distributing photogrammetric models to game developers, film studios, industrial designers, architects, and interactive storytellers. “In the future, Obriy Sim might be running its own ‘megascans’ and ‘blueprints,’ for learning, recording, and sharing the strike drone combat tactics,” explains Serhii proudly.
With SITL, advanced physics, photorealistic CGI engine, weather, and EW simulations, pilots can use Obriy Sim to train themselves in randomly generated situations, as well as replicate the actual combat missions and hardware setups for learning and post-mission analysis. “Think about Obriy Sim as a Ranger Handbook or Encyclopedia Britannica — but one you can practice with, not just read,” explains Viktor. The team is now working on tools to simplify and improve the “environment scanning” process, which is somewhat similar to what RealityCapture does with photogrammetry and would allow replicating the actual mission environment in Obriy Sim. We plan to cover it all in this blog. Stay tuned!
I Support Ukraine and Its Fight for Freedom and I Want to Help Obriy Sim Grow and Flourish. What Are My Next Steps?
The usual “call to action” part of the corporate blog stories we used to write in the past would end with a sales email and a link to the company’s contacts section on its website. However, we can’t do that for Obriy Sim for apparent reasons. Being under constant enemy surveillance, we also can’t offer you “to book the demo” or “download a trial version.” Moreover, while we already have a security clearance and vetting mechanism for dealing with the new people inside Ukraine, there is no such mechanism for the international partners yet.
Nevertheless, trust between people appears over time. To start the conversation, you can write to inquiries@obriy.airforce or send us a message via Signal or WhatsApp to +38050.567.4381 (please note that voice or video calls will not be accepted). We are particularly interested in presenting our product and building a cooperation with the tech military units from the NATO countries.
Research, interviews, and storytelling by Block(chain). Edited with care using Grammarly and OpenAI. Narrative guidance by Ms. Cynthia M. and Mr. Artem H.