Unmanned aerial vehicles (UAVs) every year demonstrate more and more advanced capabilities in detecting and avoiding obstacles, as well as in identifying and pursuing targets. The use of computer vision systems, artificial intelligence methods is an urgent task for ensuring the future of mobility and autonomous transport. The article discusses the analysis of existing image databases to prepare data for decision-making during the autonomous flight of an unmanned aerial vehicle, even in the absence of navigation systems. The choice of the coefficient for evaluating the quality of the semantic segmentation of video stream objects is carried out, a review of the methods of training the neural network based on the available data is carried out. The applied goal is the development of architecture and implementation of semantic segmentation methods for small UAVs used for research and training of automatic flight capabilities. The paper proposes to use the data obtained during the flight with the operator of the Parrot Mambo mini UAV, then manual segmentation of individual images of the received video is performed. After segmentation, the resulting data set is transferred to the neural network for training. The completed subsystem is used to simulate flight using the Cannon block in the Simulink support package for Parrot Minidrones. As a result of several flights, IoU = 0.63 was obtained, which is an acceptable result for the study of SS methods and further application in UAV automatic flight algorithms.

Analysis of the Architecture of Perceiving a Dynamic Environment for an Unmanned Aerial Vehicle / Pohudina, Olha; Kovalevskyi, Mykhailo; Naso, David; Bartolo, Rossella (LECTURE NOTES IN NETWORKS AND SYSTEMS). - In: Integrated Computer Technologies in Mechanical Engineering, 2022 : Synergetic Engineering / [a cura di] Mykola Nechyporuk, Vladimir Pavlikov, Dmitriy Kritskiy. - STAMPA. - Cham, CH : Springer, 2023. - ISBN 978-3-031-36200-2. - pp. 601-610 [10.1007/978-3-031-36201-9_50]

Analysis of the Architecture of Perceiving a Dynamic Environment for an Unmanned Aerial Vehicle

Olha Pohudina
;
David Naso;Rossella Bartolo
2023-01-01

Abstract

Unmanned aerial vehicles (UAVs) every year demonstrate more and more advanced capabilities in detecting and avoiding obstacles, as well as in identifying and pursuing targets. The use of computer vision systems, artificial intelligence methods is an urgent task for ensuring the future of mobility and autonomous transport. The article discusses the analysis of existing image databases to prepare data for decision-making during the autonomous flight of an unmanned aerial vehicle, even in the absence of navigation systems. The choice of the coefficient for evaluating the quality of the semantic segmentation of video stream objects is carried out, a review of the methods of training the neural network based on the available data is carried out. The applied goal is the development of architecture and implementation of semantic segmentation methods for small UAVs used for research and training of automatic flight capabilities. The paper proposes to use the data obtained during the flight with the operator of the Parrot Mambo mini UAV, then manual segmentation of individual images of the received video is performed. After segmentation, the resulting data set is transferred to the neural network for training. The completed subsystem is used to simulate flight using the Cannon block in the Simulink support package for Parrot Minidrones. As a result of several flights, IoU = 0.63 was obtained, which is an acceptable result for the study of SS methods and further application in UAV automatic flight algorithms.
2023
Integrated Computer Technologies in Mechanical Engineering, 2022 : Synergetic Engineering
978-3-031-36200-2
Springer
Analysis of the Architecture of Perceiving a Dynamic Environment for an Unmanned Aerial Vehicle / Pohudina, Olha; Kovalevskyi, Mykhailo; Naso, David; Bartolo, Rossella (LECTURE NOTES IN NETWORKS AND SYSTEMS). - In: Integrated Computer Technologies in Mechanical Engineering, 2022 : Synergetic Engineering / [a cura di] Mykola Nechyporuk, Vladimir Pavlikov, Dmitriy Kritskiy. - STAMPA. - Cham, CH : Springer, 2023. - ISBN 978-3-031-36200-2. - pp. 601-610 [10.1007/978-3-031-36201-9_50]
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11589/259983
Citazioni
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact