EUVIMEDEUVIMED entdecken
Crossref

Adaptive Biased Estimation in Fault-Tolerant MEMS-Based Inertial Navigation System

Adrian Muzychenko, Roman Dunets

Advances in Cyber-Physical Systems · 29.05.2026 · 11 (1) · 67-73

DOI: 10.23939/acps2026.01.067 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Advances in Cyber-Physical Systems, online erschienen am 25.05.2026. Verfasst von Adrian Muzychenko und 1 weiteren Autor:innen. Die Arbeit nennt 10 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Low-cost MEMS inertial sensors enable redundant architectures in UAVs and autonomous systems; however, their stochastic instability and gradual degradation limit the reliability of classical fusion methods. Conventional unbiased fusion techniques, such as averaging and weighted least squares, are variance-optimal under stationary noise assumptions, but degrade under non-stationary conditions. This paper proposes an Adaptive Biased Weighted Least Squares (AB-WLS) fusion framework that explicitly exploits the bias–variance trade-off to improve robustness and overall estimation accuracy. The method introduces controlled shrinkage of individual sensor contributions toward a robust group reference, combined with residual-aware adaptive weighting. This approach enables continuous down-weighting of degraded sensors without explicit fault detection and isolation. Single-axis Monte Carlo simulations of a synthetic redundant five-sensor MEMS gyroscope array (sampling rate 500 Hz, duration 300 s) under soft fault conditions (linear bias drift and variance inflation) indicate that the proposed method limits root mean squared error degradation to below 10%, whereas conventional averaging and static weighted fusion exhibit severe error growth. Under nominal (fault-free) operation, AB-WLS incurs a small 3% RMSE penalty relative to the BLUE-optimal WLS, which is a deliberate cost of the bias–variance trade-off exploited to achieve fault tolerance. The results confirm that biased estimation provides a principled and computationally efficient pathway toward fault-tolerant inertial sensor fusion in resource-constrained autonomous systems.

Zitieren

APA

Muzychenko, A., & Dunets, R. (2026). Adaptive Biased Estimation in Fault-Tolerant MEMS-Based Inertial Navigation System. Advances in Cyber-Physical Systems, 11(1), 67-73. https://doi.org/10.23939/acps2026.01.067

Vancouver

Muzychenko A, Dunets R. Adaptive Biased Estimation in Fault-Tolerant MEMS-Based Inertial Navigation System. Advances in Cyber-Physical Systems. 2026;11(1):67-73. doi:10.23939/acps2026.01.067

BibTeX
@article{Muzychenko2026,
  author = {Muzychenko, Adrian and Dunets, Roman},
  title = {Adaptive Biased Estimation in Fault-Tolerant MEMS-Based Inertial Navigation System},
  journal = {Advances in Cyber-Physical Systems},
  year = {2026},
  volume = {11},
  number = {1},
  pages = {67--73},
  doi = {10.23939/acps2026.01.067}
}
Kostenlos starten →