UAV Navigation Using Modified Neural Networks

• Published in: Journal Europeen des Systemes Automatises Vol. 58; no. 6; p. 1147
• Main Authors: Ahmed Hameed Reja, Hamzah, Mazin Abdulaali
• Format: Journal Article
• Language: English
• Published: Edmonton International Information and Engineering Technology Association (IIETA) 01.06.2025
• Subjects: Aircraft; Artificial intelligence; Classification; Computer vision; Neural networks; Personal computers; Semantics; Surveillance; Vehicles
• ISSN: 1269-6935; 2116-7087
• DOI: 10.18280/jesa.580606

With the development of modern communications systems, and the increased need for effective air defense and surveillance systems to improve tracking, command and guidance systems for unmanned aerial vehicles (UAVs), the need to improve and develop aerial monitoring devices to assist in the process of detecting moving targets with high accuracy and efficiency has increased. Since air surveillance systems and radars require human monitoring, they remain subject to error and inaccuracy. As a result of relying on this human factor, aircraft surveillance and navigation systems cannot be completely relied on human effort as their performance varies depending on the efficiency of the operators. In this study, an air surveillance system for navigation and radar devices was proposed that works with smart technologies to detect moving targets and control air navigation. Artificial neural networks (ANNs) and conventional neural network (CNN) technologies are used to automatically identify and classify moving targets in the navigation system. The data of the moving object through reflected signals and radar images is fed into the training module of the Artificial Intelligence (AI) system, which is an algorithm to plot and track the path of the moving object based on the reflected radar signals. The accuracy of the results of the AI system depends on the accuracy of the radar signals reflected from the moving object to represent the data output to the ANN. Simulation results showed that intelligent navigation can accurately identify various targets and chart their path with high efficiency. Through the results obtained by implementing AI algorithms, it is possible to control air navigation by following and detecting moving targets. The simulation results of the CNNs technology showed a high efficiency in controlling and tracking targets, reaching 97%, with robust implementation for moving target detection and tracking simulation using efficient deep learning FRCNN algorithm updated by applying MATLAB and tested on a set of data for images of various moving UAVs. Also, the results of several tests showed the success of the applied algorithm design in identifying and detecting moving targets at small error rate of 0.01 with a good training speed for the tested data set of 12 seconds.