.png){kind=icon}
.png){kind=icon}
Return
Zero-Shot Sim-to-Real 6-DoF Pose Estimation for Underwater Vehicles Based on Uncertainty-Guided Dense Correspondence
.png){kind=icon}
PRE
AI
Q
杨
X
Z
C
Q
Z
DOI：10.1109/tro.2026.3701577.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Accurate 6-DoF relative pose estimation is essential for multi-autonomous underwater vehicle (AUV) cooperative tasks. However, pose-annotated underwater data are difficult to obtain, limiting learning-based methods. In this article, we present ZUPose, a monocular pose estimator trained entirely on synthetic data and deployed directly in real underwater scenes. A key challenge is prediction noise introduced by the sim-to-real gap and underwater optical degradation. To address it, we adopt an algorithm-data co-design strategy. At the algorithm level, we develop an uncertainty-guided dense-correspondence framework in which the network jointly predicts dense correspondences and per-pixel uncertainty under a Laplace-based probabilistic formulation. The predicted uncertainty acts as a learned scale parameter to model correspondence noise and guide pose optimization. At the data level, we construct a physics-guided simulation pipeline to model underwater optical degradation and generate diverse synthetic images. In real turbid water, ZUPose achieves translation and rotation errors of 6.7 cm and 7.7<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^\circ$</tex-math></inline-formula>, with both reduced by about half compared with the best-performing baseline. The method remains stable under overexposure and long-range observation, and dual-AUV navigation experiments further validate its practical viability.
Keywords:
6-DoF pose estimation
sim-to-real transfer
uncertainty quantification
underwater robotics
Journal
IF:
10.5
Papers:
3.3K
Citations:
2.8W
Organization
Cited Papers