EUVIMEDEUVIMED entdecken
Crossref

Particle Filter-Based Cascaded Velocity Scale and Yaw Gyroscope Bias Estimation for Map-Based Localization

Gyumin Oh, Soomok Lee, Seung-Woo Seo

IEEE Robotics and Automation Letters · 2026-12 · 11 (12) · 14037-14044

DOI: 10.1109/lra.2026.3746076 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Robotics and Automation Letters, 2026-12. Verfasst von Gyumin Oh und 2 weiteren Autor:innen. Die Arbeit nennt 19 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Particle-filter (PF) map localization under unknown odometry scale or yaw gyroscope bias must stay anchored to a fixed prior map while recovering internally correct calibration parameters. Methods meeting only the first, by augmenting the particle state or resampling on pose-level scores, can stay map-localized yet produce severely wrong parameter estimates. We introduce the Uncertainty-Feedback Particle Filter (UFPF), a lightweight adaptive-noise mechanism in which a low-dimensional Kalman filter (KF) tracks velocity scale and gyro bias from PF posterior statistics and reuses its scale and yaw-bias covariances to modulate the PF translational and yaw process noises, so particle dispersion follows current calibration uncertainty, not a hand-tuned fixed level. On HeLiPR across six environments and three heterogeneous LiDAR sensors, UFPF holds map-matched translation accuracy at the nominal floor while keeping parameter RMSE low against the frame-matched injected targets. An implementation-specific map-off diagnostic scores each complete estimator in the DR-onset body frame: at $H\!=\!100$ m and $s_{\text{inj}}\!=\!0.85$ , median drift is 1.8 m for UFPF against 3.0–14.0 m for the dual-cloud PF and the shrinkage AugPF, and 12–37 m for the naive PF and the bootstrap AugPF. The added KF update incurs $\! ms per cycle on a single-thread Intel i7-14700K.

Zitieren

APA

Oh, G., Lee, S., & Seo, S. (2026). Particle Filter-Based Cascaded Velocity Scale and Yaw Gyroscope Bias Estimation for Map-Based Localization. IEEE Robotics and Automation Letters, 11(12), 14037-14044. https://doi.org/10.1109/lra.2026.3746076

Vancouver

Oh G, Lee S, Seo S. Particle Filter-Based Cascaded Velocity Scale and Yaw Gyroscope Bias Estimation for Map-Based Localization. IEEE Robotics and Automation Letters. 2026;11(12):14037-14044. doi:10.1109/lra.2026.3746076

BibTeX
@article{Oh2026,
  author = {Oh, Gyumin and Lee, Soomok and Seo, Seung-Woo},
  title = {Particle Filter-Based Cascaded Velocity Scale and Yaw Gyroscope Bias Estimation for Map-Based Localization},
  journal = {IEEE Robotics and Automation Letters},
  year = {2026},
  volume = {11},
  number = {12},
  pages = {14037--14044},
  doi = {10.1109/lra.2026.3746076}
}
Kostenlos starten →