EUVIMEDEUVIMED entdecken
Crossref

A Decoupled Strategy for Attitude and Heading Estimation of MEMS IMU Under Acceleration and Magnetic Disturbances

Yiran Zhu, Shuai Chang, Kaixuan Wang, Hong Lu, Min Zhang, Hui Li

IEEE Sensors Journal · 01.07.2026 · 26 (13) · 20290-20303

DOI: 10.1109/jsen.2026.3697386 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.07.2026. Verfasst von Yiran Zhu und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 35 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Micro-electromechanical system (MEMS) inertial measurement units (IMUs) are widely used for motion control of autonomous robots and vehicles. However, their sensor observations are prone to contamination from external disturbances such as acceleration and magnetic anomalies, which seriously reduce the accuracy of attitude and heading estimates. To address these challenges, this article proposes a dual-stage strategy in a decoupled Extended Kalman Filter (EKF) framework and deals with the acceleration and magnetic disturbance independently. First, a hybrid long short-term memory and fourth-order Runge–Kutta algorithm (LSTM-RK4) is presented to improve the robustness of attitude estimation against acceleration disturbance. The gyroscope data are used in LSTM to predict attitude increments, and a weighted fusion of the LSTM prediction and Runge–Kutta integration results is performed to achieve stable and accurate attitude results. Second, to reduce the influence of magnetic interference on heading estimation, a Gaussian Process Regression (GPR)-assisted Decoupled EKF (GDEKF) is designed to incorporate the magnetic anomaly prediction in the second-stage filtering to achieve joint estimation of heading and magnetic anomalies. Field experiments validate the superiority of the proposed framework to traditional approaches. During severe continuous acceleration disturbance, the LSTM-RK4 method achieves attitude root-mean-square errors (RMSEs) as low as 0.14 $^{\circ}$ for pitch and 0.15 $^{\circ}$ for roll. Under strong magnetic disturbances, the GDEKF achieves a final heading RMSE of 1.48 $^{\circ}$ .

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Zhu, Y., Chang, S., Wang, K., Lu, H., Zhang, M., & Li, H. (2026). A Decoupled Strategy for Attitude and Heading Estimation of MEMS IMU Under Acceleration and Magnetic Disturbances. IEEE Sensors Journal, 26(13), 20290-20303. https://doi.org/10.1109/jsen.2026.3697386

Vancouver

Zhu Y, Chang S, Wang K, Lu H, Zhang M, Li H. A Decoupled Strategy for Attitude and Heading Estimation of MEMS IMU Under Acceleration and Magnetic Disturbances. IEEE Sensors Journal. 2026;26(13):20290-20303. doi:10.1109/jsen.2026.3697386

BibTeX
@article{Zhu2026,
  author = {Zhu, Yiran and Chang, Shuai and Wang, Kaixuan and Lu, Hong and Zhang, Min and Li, Hui},
  title = {A Decoupled Strategy for Attitude and Heading Estimation of MEMS IMU Under Acceleration and Magnetic Disturbances},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {13},
  pages = {20290--20303},
  doi = {10.1109/jsen.2026.3697386}
}
Kostenlos starten →