Acoustic Neural 3D Reconstruction Under Pose Drift

2025-03-11

AI Summary

To address severe geometric distortion in 3D reconstruction caused by sonar sensor pose drift, this paper proposes the first end-to-end differentiable framework jointly optimizing neural implicit surfaces (SDF/NeRF-style) and 6-DoF sensor poses directly from acoustic images. Departing from conventional approaches relying on external pose priors or post-hoc correction, our method embeds learnable pose parameters into the neural implicit representation and designs a physics-informed neural renderer tailored to acoustic imaging characteristics. Geometry and pose are co-optimized via gradient-based backpropagation. Evaluated on both synthetic and real-world sonar datasets, the framework achieves high-fidelity reconstruction—preserving fine surface details and global structure—even under severe drift (>5° rotation and >10 cm translation). Results demonstrate substantial improvements in robustness and reconstruction accuracy over state-of-the-art methods.

Technology Category

Intelligent Robots: Multimodal Perception & Sensor FusionSearch and Optimization: Learning to SearchKnowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingGraph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphsUser Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

We consider the problem of optimizing neural implicit surfaces for 3D reconstruction using acoustic images collected with drifting sensor poses. The accuracy of current state-of-the-art 3D acoustic modeling algorithms is highly dependent on accurate pose estimation; small errors in sensor pose can lead to severe reconstruction artifacts. In this paper, we propose an algorithm that jointly optimizes the neural scene representation and sonar poses. Our algorithm does so by parameterizing the 6DoF poses as learnable parameters and backpropagating gradients through the neural renderer and implicit representation. We validated our algorithm on both real and simulated datasets. It produces high-fidelity 3D reconstructions even under significant pose drift.

Problem

• Optimizing neural implicit surfaces for 3D reconstruction using acoustic images.
• Addressing reconstruction artifacts caused by inaccurate sensor pose estimation.
• Jointly optimizing neural scene representation and sonar poses under pose drift.

Innovation

• Joint optimization of neural scene and sonar poses.
• Parameterizes 6DoF poses as learnable parameters.
• Backpropagates gradients through neural renderer.