AUTONOMOUS NAVIGATION ERRORS IN SMALL UNMANNED AIRCRAFT AND METHODS FOR OVERCOMING THEM
Випробування та сертифікація · 31.03.2026 · (1(11)) · 99-107
DOI: 10.37701/ts.11.2026.11 ↗Kurz gefasst
Fachartikel in der Zeitschrift Випробування та сертифікація, online erschienen am 31.03.2026. Verfasst von O. Kukla. Die Arbeit nennt 22 Literaturangaben.
Abstract (Originalsprache)
The article focuses on the complex processes of error generation and accumulation within the autonomous navigation systems of small unmanned aerial vehicles (UAVs), particularly when operating in environments characterized by restricted or denied access to Global Navigation Satellite System (GNSS) signals. The topic encompasses a critical evaluation of inertial sensor performance, the impact of environmental destabilizing factors, and the evolution of algorithmic correction methods. The primary purpose of the article is to conduct a comprehensive analytical review of the fundamental sources of navigation accuracy degradation and to substantiate the most promising compensation techniques, specifically emphasizing the integration of artificial intelligence technologies to ensure stable UAV performance in challenging interference conditions. The research methodology is built upon a systematic integration of general scientific and specialized technical methods. Systematic classification was employed to categorize the diverse types of errors inherent in micro-electromechanical systems (MEMS). A comparative analysis was conducted to evaluate the efficiency of classical filtering methods, such as the Extended Kalman Filter (EKF), against modern deep learning-based approaches. Furthermore, the method of theoretical generalization was applied to synthesize global trends in aviation transport and information technology. The study specifically analyzes the architectural features of recurrent neural networks (RNNs) and their capacity for processing multi-dimensional time-series data derived from flight telemetry. The analysis reveals that the primary challenge for small-scale UAVs is their reliance on low-cost MEMS sensors, which exhibit significant sensitivity to high-frequency structural vibrations and thermal drift. The study provides a detailed description of the mechanisms behind gyroscope zero-drift and scale factor instability, which lead to exponential growth in positioning errors over time. It is demonstrated that while multisensor data fusion is essential, its effectiveness is severely limited during prolonged GNSS outages due to the non-linear and non-stationary nature of sensor noise. The research identifies that the most effective tool for approximating these complex dependencies is the use of recurrent architectures, specifically Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks. These models excel at learning long-term temporal dependencies in sensor errors, allowing for the dynamic compensation of drift that traditional static models fail to capture.
Zitieren
APA
Kukla, O. (2026). AUTONOMOUS NAVIGATION ERRORS IN SMALL UNMANNED AIRCRAFT AND METHODS FOR OVERCOMING THEM. Випробування та сертифікація, 99-107. https://doi.org/10.37701/ts.11.2026.11
Vancouver
Kukla O. AUTONOMOUS NAVIGATION ERRORS IN SMALL UNMANNED AIRCRAFT AND METHODS FOR OVERCOMING THEM. Випробування та сертифікація. 2026:99-107. doi:10.37701/ts.11.2026.11
BibTeX
@article{Kukla2026,
author = {Kukla, O.},
title = {AUTONOMOUS NAVIGATION ERRORS IN SMALL UNMANNED AIRCRAFT AND METHODS FOR OVERCOMING THEM},
journal = {Випробування та сертифікація},
year = {2026},
number = {1(11)},
pages = {99--107},
doi = {10.37701/ts.11.2026.11}
}
EUVIMED