Nevertheless, to our delight, news occasionally reaches us about intelligent students who have managed to impress academic and technological circles in Bulgaria and around the world with their knowledge, skills, and hard work. This is exactly what four students from the Technological College "Electronic Systems" (TCES) at the Technical University - Sofia did.
At the end of June, Kaloyan Miladinov, Ognyan Rassadov, Filostratos Titopoulos, and Aleksandar Mechkarov participated in the first edition of the European Hackathon for Defense Technology, part of the Future Festival Festival der Zukunft in Munich.
The TCES students (with mentors Petar Kirkov and Kirilka Angelova) are the only participants from Bulgaria and stand shoulder to shoulder with cybersecurity specialists and students from leading universities around the world. They impressed everyone with "Identification of anti-personnel mines using drones and artificial intelligence" - the innovative project with which the students won third place at one of the most authoritative technological events in Europe. But before we learn more details about their project, let's get to know its creators - Filostratos, Aleksandar, Ognyan, and Kaloyan.
The Wonderful Quartet
In the 7th grade Filostratos Titopoulos stumbles upon an exhibition at NDК dedicated to modern technologies. There, he meets people from TUES and instantly falls in love with the school's stories and the idea of studying there. Currently, he is in the "Computer Networks" parallel class; his favorite area in programming is automation, but he is also interested in cybersecurity. The modern scientists he admires are Elon Musk and Bjarne Stroustrup. Filostatos tries to make each of his days different. Over the past few weeks, he has been on industrial internships, where he encounters various interesting problems and tasks. "Outside of 'work hours,' I cook jams and savory dishes, try different coffee recipes, and work on some small projects," the young man says. Just like Filostatos, and Alexander Mechkarov is in the "Computer Networks" parallel class. He learned about TUES through additional math lessons. At that time, he had not engaged in any programming, but the idea of developing in the field of information technology and realizing his own ideas captivated him instantly. He is proud to study there. For Alexander, 2024 has been filled with many diverse activities, including the opportunity to be part of the organizational team for two emblematic events for his school: HackTUES and TUES Fest. Currently, he is completing his mandatory industrial internship at an IT company, where he further develops in the field of computer networks. Ognian Rassadov studies "System Programming." He chose TUES because he believes it is the best non-private school for technology-oriented students in Sofia. A typical day for him usually involves programming. "It depends on which projects I'm working on, but most often I sit in front of the computer and bang my head against the keyboard," Ognian shares with a smile. He does not hide that he enjoys programming and creating. He loves constantly expanding his worldview by traveling outside of Bulgaria, where he both gets to know different cultures and can enjoy diverse and delicious food. Even at the end of sixth grade, Kaloyan Miladinov knew that he wanted to pursue programming. He learned about TUES by chance during a visit to the dentist, but the name of the school was imprinted in his subconscious. By the end of seventh grade, the decision to study there was already clear. Currently, he is in the "System Programming" parallel class, and considers artificial intelligence and robotics as the most interesting fields in computer science. The scientists he admires are Sam Altman and Ilya Sutskever. "When I am not in front of the computer, I enjoy traveling within and outside the country, hiking in the mountains, going out with friends, or looking for something new and exciting to work on." Currently, Kaloyan is also completing his mandatory industrial practice in an IT company. However, he admits that over the past year he has been fortunate to have mostly diverse days in his routine - from organizing our school hackathon - Hack TUES, to participating in the one in Munich. How did the idea for the technology for identification of anti-personnel mines using drones and artificial intelligence? It happened after a meeting where we discussed various ideas on what to work on. We settled on identifying anti-personnel mines using drones and artificial intelligence, due to the simplicity of the working principle and the existence of existing studies. And thanks to the latter, within a few weeks, we managed to create a working prototype.
What is your project and how does it work?
Kaloyan: Our idea is to create a cheap solution for detecting and marking anti-personnel mines. Why exactly anti-personnel mines? Because they harm the civilian population the most, as they are designed to be very difficult to detect. Our project is a module that can be used with drones available on the market, thus eliminating a large part of the costs associated with the methods currently in use.
Philostrotos: On the side of search and recognition algorithms for mines, we use artificial intelligence algorithms, specifically Computer Vision algorithms. They work by finding "anomalies" in the recorded thermal image, which anomalies are caused by the presence of mines. Before we started creating any algorithms, we conducted an extensive study and data collection to train them on.
Alexander: I am involved in the development of embedded microcomputer electronics for the project and in establishing communication between the flying device and our module. At the hardware level, our project consists of several main components: a microcomputer (Raspberry Pi 5), a thermal (Flir One Pro) and a standard camera, which allow us to track the thermal image and the contours of anti-personnel mines. Obtaining the geographical location of anti-personnel mines can be achieved by integrating a GPS module. If the drone has a built-in GPS, we can directly obtain the coordinates through a connection between the two devices.
Ognian: We are working on an easy-to-use interface that visualizes the locations of the mines and the corresponding images. This will allow people with little training to use our system and will further facilitate its use.
How long did it take you to do all this?
We started working three weeks before the hackathon, researching the problem and conducting tests with inert mines. Thus, before the official start of the hackathon, we had confirmed our hypothesis about mine detection and had collected data as close to real-life as possible for training the algorithms. We managed to develop the project and the algorithms behind it during the hackathon, while also dealing with additional issues - our GPS module did not work, but we managed to integrate with the provided drone and use data from its sensors.
What are the three qualities that, in your opinion, distinguish your technology from those of the other participants at the European Defense Technology Hackathon 2024 in Munich?
- We have real data, resulting from scientific observation that we conducted.
- Clear targeted use and integration into existing systems and solutions. We are not reinventing the wheel and the entire infrastructure for its transmission, we only improve the tap.
- The prototype is built with easily accessible components, which allows for much better scalability of the idea in the future.
How much would it cost to manufacture such a module for drones?
We are still in the development phase of the project, so it is difficult to say with certainty. During the hackathon, we built the prototype with easily accessible components (under a 1000 BGN budget). Our goal is to make it efficient and many times more accessible than the current methods used for locating anti-personnel mines. According to the current World Bank statistics, the discovery of a single mine costs between 300 and 1000 dollars.
What do you think is the significance of your project?
Our technology represents one layer of the solution to a very painful and current problem - anti-personnel mines. With it, we hope to facilitate the teams that clear mines by locating the mines, which in most cases are no longer even on the battlefield, but where the peaceful population lives.
We aim for a low cost of manufacturing such a module and integration with existing systems - drones or visualization on maps that we are all used to using, because this makes the discovery and removal of mines easier and faster.
