.png){kind=icon}
.png){kind=icon}
Return
ResAlignNet: A data-driven approach for INS/DVL alignment
.png){kind=icon}
OA
AI
DOI：10.1016/j.oceaneng.2026.125277.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A data-driven approach specifically optimized with residual connections for deep and stable INS/DVL alignment. • The method achieves effective alignment using only INS and DVL sensors. It requires no external sensors (like GNSS) or complex, specific vehicle maneuvers/trajectories. • ResAlignNet significantly reduces the required time for alignment, providing a substantial real-world practical advantage and faster mission start. • The approach demonstrates successful simulation-to-reality (Sim2Real) transfer learning, allowing for training on synthetic data and deployment on operational measurements, reducing the need for extensive real-world data collection.
Keywords:
Doppler velocity log
Inertial navigation system
Autonomous underwater vehicle
Information aiding
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
5.5
Papers:
5.8K
Citations:
7.6W
Organization
No organization information available