PUGS: Perceptual Uncertainty for Grasp Selection in Underwater Environments

2025-02-13

AI Summary

To address the challenge of grasp pose selection under underwater weak-perception conditions—characterized by partial observability, occlusions, and sensor noise—this paper proposes an uncertainty-driven autonomous grasping method. The core methodological innovation lies in explicitly modeling and propagating voxel occupancy uncertainty from multi-view 3D reconstruction (e.g., via NeRF or TSDF) into the grasp pose evaluation pipeline, thereby departing from conventional equal-weighting assumptions. Specifically, it comprises: (i) uncertainty estimation of volumetric reconstructions, (ii) a cross-view uncertainty propagation mechanism, and (iii) an uncertainty-weighted grasp scoring function. Extensive evaluation on both simulated and real-world underwater datasets demonstrates substantial improvements in grasp success rates, particularly under severe occlusion and high noise. The approach delivers robust, interpretable, and generalizable decision-making for autonomous manipulation in low-observability environments.

Technology Category

• Reasoning under Uncertainty: Uncertainty Representations
• Intelligent Robots: State Estimation
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

When navigating and interacting in challenging environments where sensory information is imperfect and incomplete, robots must make decisions that account for these shortcomings. We propose a novel method for quantifying and representing such perceptual uncertainty in 3D reconstruction through occupancy uncertainty estimation. We develop a framework to incorporate it into grasp selection for autonomous manipulation in underwater environments. Instead of treating each measurement equally when deciding which location to grasp from, we present a framework that propagates uncertainty inherent in the multi-view reconstruction process into the grasp selection. We evaluate our method with both simulated and the real world data, showing that by accounting for uncertainty, the grasp selection becomes robust against partial and noisy measurements. Code will be made available at https://onurbagoren.github.io/PUGS/

Problem

• Quantify perceptual uncertainty in 3D reconstruction
• Incorporate uncertainty into underwater grasp selection
• Enhance robustness against noisy measurements

Innovation

• Quantifies perceptual uncertainty in 3D reconstruction
• Incorporates uncertainty into underwater grasp selection
• Propagates reconstruction uncertainty into grasp decisions