.png){kind=icon}
.png){kind=icon}
Return
A physics-informed residual transformer framework with sparse GNSS constraints for underwater SINS/DVL trajectory reconstruction
.png){kind=icon}
PRE
AI
X
H
H
R
J
DOI：10.1016/j.oceaneng.2026.127530.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Physics-Informed Residual learning for underwater SINS/DVL trajectory reconstruction. • Physics-informed losses enable ground-truth-free learning of current residuals. • Uncertainty-aware three-factor gating safely injects learned priors into the filter. • Closed-form DVL scale-factor self-calibration within sparse GNSS windows. • 86–94% position error reduction over standard SINS/DVL with 4.44% GNSS.
Keywords:
Underwater trajectory reconstruction
Sparse GNSS
Physics-informed neural network
Transformer
Error-state filtering
RTS smoothing
Journal
IF:
5.5
Papers:
5.5K
Citations:
7.6W
Organization
No organization information available
Cited Papers