Search academic texts

Information × Registration Number 0224U033388, (0121U113590) , R & D reports Title Multisensor intelligent systems popup.stage_title Використання методів штучного інтелекту в безпілотних системах Head Olenych Ihor B., к.ф.-м.н. Registration Date 26-12-2024 Organization Ivan Franko National University of Lviv popup.description1 The aim of the project is to develop and implement a dynamically scalable architecture for the generation of prototypes of intelligent multiparameter biosensor systems, the principles of construction and operation of which are based on promising trends in biomedical technology. popup.description2 A B S T R A C T Report on the scientific research work: 19 pages, 2 figures. UNMANNED AERIAL VEHICLES, VISUAL ODOMETRY, COMPUTER VISION, MATHEMATICAL MODELING, MACHINE LEARNING, VIDEO IMAGE DIGITAL PROCESSING. Research object: models of unmanned aerial vehicle motion, flight path parameters, input streams of onboard video cameras, projective transformations of pixel images, and neural networks. The purpose of the work is to develop models and the navigation system prototype for unmanned aerial vehicles to perform autonomous flight along a given route, as well as to keep the target in the video stream and bring the aerial vehicle to the target. Research methods: mathematical modeling, computer vision and machine learning methods, programming platforms. Results obtained: new mathematical models of autonomous navigation of unmanned aerial vehicles based only on the video stream from the onboard camera and the inclination angles of unmanned aerial vehicles from the flight controller are proposed. In particular, a virtual video camera model and new algorithms for dynamically selecting computer vision parameter values that take into account illumination and terrain type have been developed. The neural network that determines the degree of similarity of image segments from a video stream with a given image has been built. The Lucas-Kanade method, which makes it possible to automate the process of holding a target in the incoming video stream and guiding unmanned aerial vehicles to this target, was improved using the Kalman filter. The program set for onboard computers has been developed that implement the real-time execution of the proposed algorithms.   Product Description popup.authors Ivanov Serhii A. Baranov Mykola V. Dyakonuk Liliya Mykolaivna Muzychuk Anatolii O. Filiak Yevhen Z. popup.nrat_date 2024-12-26 Close