Registration Number
0225U004994,
(0121U113068)
, R & D reports
Title
To fulfill the tasks of the long-term development plan of the scientific direction "Technical Sciences" of V. N. Karazin Kharkiv National University
popup.stage_title
На виконання завдань перспективного плану розвитку наукового напряму «Технічні науки» Харківського національного університету імені В.Н.Каразіна на 2025 рік
Head
Tolstoluzka Olena H., Доктор технічних наук
Registration Date
23-12-2025
Organization
V.N. Karazin Kharkiv National University
popup.description1
Development and experimental-statistical research of new effective algebraic methods and means of formation and digital processing of signals, algorithms of automatic radio control and radio monitoring, as well as prototypes of multifactor biometric systems for authentication of mobile service users.
popup.description2
The object of the study is the process of creating and operating applied architecturally-oriented time-parameterized parallel software for computer systems based on robotic platforms
The subject of the study is models, methods and tools for developing highly efficient parallel software that take into account time parameters, architectural features of computing devices and use artificial intelligence methods for processing multimedia data
The purpose of the study is to increase the efficiency of robotic environmental monitoring systems by developing a technology for creating applied time-parameterized parallel software that guarantees real-time data processing under conditions of limited resources.
Research results:
Theoretical justification: The concept and models of architecturally-oriented time-parameterized software have been developed. It has been proven that explicit consideration of the time factor and dynamic planning allow to eliminate delays caused by the stochastic nature of artificial intelligence algorithms in robotic systems.
Hardware optimization: CPU inefficiency for real-time computer vision tasks has been experimentally established. It has been proven that the use of graphics accelerators and the Apple Silicon (MPS) architecture provides data processing acceleration up to 25 times, guaranteeing stable latency and energy efficiency.
Development methodology: The feasibility of a hybrid approach to code autogeneration has been substantiated, where the combination of the Claude (for logic and orchestration) and Gemini (for infrastructure) models ensures the reliability of the development of complex parallel systems.
Practical implementation: A software package has been created for collecting and analyzing multimedia data using the YOLOv11 and BLIP algorithms. Parallel data compression algorithms and cloud architecture (Docker, Elasticsearch) have been implemented, which ensured system scalability and integration with equipment such as DJI Goggles
Product Description
popup.authors
Serhii V. Yakovlev
Maksym М. Khruslov
Viktoriia Y. Strilets
Volodymyr A. Skrynnik
Vladyslav S. Korshenko
Volodymyr V. Kopeikin
popup.nrat_date
2025-12-23
