Student EEICT 2025: General Papers

Permanent URI for this collection

https://hdl.handle.net/11012/255236

Title: Proceedings I of the 31th Conference STUDENT EEICT 2025
Subtitle: General Papers

Publisher: Brno University of Technology, Faculty of Electrical Engineering and Communication
Editor: Assoc. Prof. Vítězslav Novák
Place and year: Brno, 2025

ISBN: 978-80-214-6321-9
https://www.eeict.cz/

Browse

Now showing 1 - 5 of 80

Driver inattention state: an overview of distraction, drowsiness, visual perception and infrared imaging
(Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2025) Boujenfa, Sabrina
The identification of driver inattention is of critical importance in the context of road safety, given that inattention constitutes one of the primary causes of road traffic accidents. This paper examines several factors that contribute to driver inattention, with a particular focus on distraction, drowsiness, visual perception and infrared imaging. Distraction can be triggered by external (e.g. mobile phones) or internal (e.g. cognitive load) factors, while drowsiness significantly reduces responsiveness and attention. Visual perception is a key indicator of the driver's state of alertness. Modern driver monitoring systems increasingly use infrared imaging and eye-tracking technologies to analyse eye movements and detect signs of distraction or drowsiness at an early stage. This article provides a general overview of these aspects and shows how technological developments can help to improve the detection of inattention and increase driving safety.
Image Reconstruction in Electrical Impedance Tomography through 1D-Convolutional Neural Network
(Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2025) Nomvussi, Serge Ayme Kouakouo; Mikulka, Jan
This paper presents a comparative analysis of image reconstruction performance using a 1D-Convolutional Neural Network (1D-CNN) against the Total Variation and the Gauss- Newton methods. The evaluation, conducted across multiple tests conditions, demonstrates that the 1D-CNN consistently outperforms both conventional methods in terms of correlation coefficient and structural similarity index (SSIM). In noise-free scenarios, the 1D-CNN achieves significantly higher correlation and SSIM values, indicating superior reconstruction accuracy. Furthermore, in the presence of noise (30 dB and 60 dB), the performance of the Total Variation and Gauss-Newton methods deteriorates considerably, whereas the 1D-CNN maintains high correlation and SSIM values, demonstrating strong robustness to noise. These findings highlight the effectiveness of deep learningbased approaches for image reconstruction, making the 1D-CNN a promising alternative to traditional reconstruction techniques.
General Solution of Linear Planar Weakly Delayed Discrete Systems with Multiple Delays and with a Singular Matrix of Non-delayed Terms
(Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2025) Hartmanová, Marie; Diblík, Josef
A planar weakly delayed linear discrete system with multiple delays is considered. Assuming that the conditions for weakly delayed systems are fulfilled, the general solution of the system is constructed in the previously not considered case of the matrix of nondelayed terms having a single nonzero eigenvalue with the second one being zero. Then, there exists a non-delayed planar discrete system that, for all sufficiently large values of the independent variable, has the same general solution. New findings are illustrated with an example and discussed in relation to the previously known investigations. The formulas derived can be useful in digital signal processing.
The Methods of Accurate Measurement of Tilt Azimuth within Small Angles of Tilt
(Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2025) Junasová, Veronika; Levek, Vladimír
This article focusses on the determination of the tilt azimuth within the small angles of tilt using accelerometer. Azimuth is usually acquired using magnetometers and is then called the magnetic azimuth. In practice, magnetometers are commonly accompanied by other inertial sensors or filtering. The azimuth can be relatively easily acquired using a single magnetometer, which has to be properly calibrated against the magnetic materials in its surroundings. For further improvement in accuracy of measurement, a sensor fusion or filtering is necessary. The advantages and drawbacks of using magnetometer and filtering are discussed. In contrast to these more complex techniques, a different configuration of measurement is used further in the article. The main focus is to present a low-cost and easy-to-integrate method of measurement. It consists of the definition of the accelerometer position, which is rotated by 45°. In this configuration, the accelerometer alone can be used to determine the azimuth of the tilt and the tilt itself. That is given by the better distribution of gravitational force among accelerometer axes and their sensitivity. This is prooved by mathematically derived equations. In the final results, the measurement with the rotated accelerometer is shown, and the improvement in accuracy and statistics is determined.
Review of Autonomous UAV Methods in GNSS-Challenging Environments
(Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2025) Prokop, Šimon; Marcoň, Petr
Interest in autonomous UAVs has been growing due to the need in many different industries to seek a robust and efficient system that can work even in remote areas without any other intervention. This paper provides a comprehensive review of recent advancements in autonomous UAV methodologies, with a particular focus on three key areas: planning, navigation, and AI-driven algorithms. The review examines the strengths and limitations of traditional approaches, such as Kalman filters and SLAM-based methods, while also exploring the potential of AI-driven techniques, particularly deep reinforcement learning (DRL), in enhancing UAV autonomy. Although recent developments show promising results, challenges remain in scalability, computational efficiency, and adaptability to complex environments. The findings suggest future research directions toward hybrid methodologies that integrate classical and AIbased techniques to improve UAV performance in real world scenarios.

Browse

Recent Submissions