Drone chasers and archive recorded in glass: what was shown at the AI exhibition in Minsk

The exhibition within the framework of the V International Forum IT-Academgrad "Artificial Intelligence in Belarus" brought together at the site of the National Academy of Sciences leading domestic and foreign experts and developers in the field of artificial intelligence on October 13, the correspondent of the agency "Minsk-Novosti" reports.

At the exhibition, members of the scientific community demonstrate advanced achievements and developments in a variety of fields – from automation of industrial and military processes to healthcare, agriculture, education and information protection.

Representatives of the National Cadastre Agency presented the service of automatic valuation of residential real estate, which is based on the current prices of real transactions from the state “Price Register”. The development takes into account 15 factors and, depending on them, gives the final cost. Among them, for example, the area of the apartment, the number of rooms, the material of the walls, the year of commissioning of the building, the distance to the metro, the neighborhood.

“The service is completely free and has already won the trust of citizens – since the launch in October 2025, more than 300 thousand user requests have been recorded,” said Elena Dalidovich, deputy head of the cadastral assessment department of the National Cadastral Agency. Our neural network helps sellers and buyers to determine a fair price and significantly accelerates the conclusion of transactions, and also serves as a reliable reference point for realtors, mortgage services of banks and independent appraisers.

Innovative aircraft vertical takeoff and landing series "VTOL-mini" was presented by representatives of the Scientific and Production Center of multifunctional unmanned systems. They are equipped with visual navigation systems for autonomous flights in the absence of satellite communications. Also among the developments there is an unmanned aerial vehicle for automatic detection and tracking of ground and air objects.

- Our optical system with automatic tracking of moving objects can be used, for example, for regulation on roads. Conventionally, the drone operator found the intruder’s car, turned on auto escort, and then the aircraft already follows this mobile object in pursuit mode, said Maria Maximova, head of the marketing and business development department of the SPC of multifunctional unmanned complexes of the National Academy of Sciences of Belarus. Our technologies are also used in the Forest and Invasive Plants project. We compiled a database of shrinking trees, woodlands and invasive plants such as Sosnowski hogweed and goldenrod. The drone, equipped with a camera, flies around the site, and then the neural network processes the received video materials and automatically determines the necessary places.

Visitors of the forum and the scientific project “5D Archive of Cultural Heritage of Belarus” were interested. It is aimed at creating a fundamentally new type of ultra-durable archival memory using optical data recording inside special glass elements. Such carriers are able to withstand extreme temperature loads up to 1.5 thousand degrees.

- Today, with the rapid development of artificial intelligence, the load on digital libraries and archives has increased tens of thousands of times, and traditional hard drives are rapidly wearing out. Recording data into the glass structure at the level of optical defects, we achieve a colossal storage density, said Mikhail Lobanok, associate professor of the Department of Physical Electronics and Nanotechnology of BSU. Last year, within the framework of the contest “100 Ideas for Belarus”, we successfully tested the technology, bringing information from Belarusian museums about the monument to Euphrosyne of Polotsk into a glass matrix. The plans are to scale the development and implement it in the leading museums of the country.

The development of the military sphere also differed. In particular, a high-performance hardware and software complex, which is designed to enhance security and operational monitoring. It is capable of processing up to 50 video streams from surveillance cameras in real time and automatically detecting 12 types of ground military equipment.

The main difficulty was not the development of the interface, but the creation and detailed markup of a unique dataset. If civilian databases are freely available, then it took us about three months to train the neural network to recognize military equipment, commented Lieutenant Colonel Maxim Kasperovich, head of the research center of the Military Academy. Moreover, the source of the video signal does not matter: the complex works equally effectively with stationary hidden cameras and dynamic video streams from drones. In the future, we plan to increase the number of types of targets and expand the system’s capabilities for working with air objects.

Photo by Alexander Morunov