Adaptive Human-machine Systems with Automated Control of Large Language Models for Unmanned Aerial Vehicle Control
Elektronne modelûvannâ · 25.05.2026 · 48 (3) · 3-17
DOI: 10.15407/elmodel.48.03.003 ↗Kurz gefasst
Fachartikel in der Zeitschrift Elektronne modelûvannâ, online erschienen am 01.06.2026. Verfasst von D. Vasyk und 1 weiteren Autor:innen. Die Arbeit nennt 14 Literaturangaben.
Abstract (Originalsprache)
An Intelligent Assistance Systems (IAS) study was conducted on the integration of Large Lan-guage Models (LLMs) into control loops for a safety-critical systems context, specifically Unmanned Aerial Vehicles (UAVs) performing relevant missions. The key drawbacks of using generative artificial intelligence in operational control tasks were analyzed, among which the stochastic nature of the models, susceptibility to hallucinations, and insufficient predictability of responses were highlighted. To address the identified issues, a method for organizing adap-tive automatic control is proposed, implemented as a separate architectural layer of the AI sys-tem. This approach ensures deterministic filtering of the language modelʼs output data, ena-bling the safe transformation of LLM results into clear executive commands. A set of metrics for analyzing system performance indicators and determining the limits of LLM capabilities in interpreting commands and context was identified.
Zitieren
APA
Vasyk, D., & Kharchenko, V. (2026). Adaptive Human-machine Systems with Automated Control of Large Language Models for Unmanned Aerial Vehicle Control. Elektronne modelûvannâ, 48(3), 3-17. https://doi.org/10.15407/elmodel.48.03.003
Vancouver
Vasyk D, Kharchenko V. Adaptive Human-machine Systems with Automated Control of Large Language Models for Unmanned Aerial Vehicle Control. Elektronne modelûvannâ. 2026;48(3):3-17. doi:10.15407/elmodel.48.03.003
BibTeX
@article{Vasyk2026,
author = {Vasyk, D. and Kharchenko, V.},
title = {Adaptive Human-machine Systems with Automated Control of Large Language Models for Unmanned Aerial Vehicle Control},
journal = {Elektronne modelûvannâ},
year = {2026},
volume = {48},
number = {3},
pages = {3--17},
doi = {10.15407/elmodel.48.03.003}
}
EUVIMED