EUVIMEDEUVIMED entdecken
Crossref

Automatic Synchronization of IMU Gait Data via Cross-Segment Angular Velocity Prediction

Yi Chiew Han

IEEE Sensors Journal · 01.07.2026 · 26 (13) · 20242-20249

DOI: 10.1109/jsen.2026.3695136 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.07.2026. Verfasst von Yi Chiew Han. Die Arbeit nennt 19 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Temporal misalignment between inertial measurement units (IMUs) can compromise the accuracy of gait analysis in wearable motion monitoring. This article proposes a fully gyroscope-based method to synchronize multisegment IMU data by leveraging cross-segment angular velocity prediction. In particular, the angular velocity of the shank is predicted from the thigh angular velocity using multiple models, including a feedforward neural network (NN), a convolutional (CNN), long short-term memory (LSTM), bidirectional LSTM (BiLSTM), BiLSTM-Transformer, and CNN-BiLSTM-Transformer. The predicted shank signal is then aligned with the measured shank signal via cross correlation to estimate temporal shifts, eliminating the need for hardware triggers or communication-based synchronization. A leave-one-subject-out (LOSO) cross-validation scheme is employed to evaluate the generalization of the proposed method across subjects, ensuring that the models are tested on unseen participants. Experimental validation with six healthy adult subjects without gait impairments walking at a self-selected pace shows that all models achieve high correlation coefficients between the predicted and measured shank angular velocity for synchronization. Absolute synchronization errors remain low, with mean absolute errors ranging from 0.99 to 1.6 samples (9.9–16 ms), even under intentional shifts of ±300 samples (±3 s). Results indicate that simpler models, such as NN and LSTM, offer lightweight, practical solutions suitable for real-time synchronization, while more complex architectures like CNN-BiLSTM-Transformer achieve slightly higher synchronization accuracy.

Zitieren

APA

Han, Y. C. (2026). Automatic Synchronization of IMU Gait Data via Cross-Segment Angular Velocity Prediction. IEEE Sensors Journal, 26(13), 20242-20249. https://doi.org/10.1109/jsen.2026.3695136

Vancouver

Han YC. Automatic Synchronization of IMU Gait Data via Cross-Segment Angular Velocity Prediction. IEEE Sensors Journal. 2026;26(13):20242-20249. doi:10.1109/jsen.2026.3695136

BibTeX
@article{Han2026,
  author = {Han, Yi Chiew},
  title = {Automatic Synchronization of IMU Gait Data via Cross-Segment Angular Velocity Prediction},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {13},
  pages = {20242--20249},
  doi = {10.1109/jsen.2026.3695136}
}
Kostenlos starten →