SE(3)-PoseFlow: Estimating 6D Pose Distributions for Uncertainty-Aware Robotic Manipulation

2025-11-03

AI Summary

To address the ambiguity and multimodal uncertainty in 6D object pose estimation—arising from occlusion, symmetry, and other perceptual challenges in robotic manipulation—this paper introduces the first probabilistic pose estimation framework on the SE(3) manifold. It pioneers the application of flow matching to model continuous pose distributions directly on SE(3), enabling principled density estimation over rigid-body transformations. The method supports multi-hypothesis output and explicit uncertainty quantification, combining theoretical rigor with computational efficiency. Evaluated on Real275, YCB-Video, and LM-O benchmarks, it achieves state-of-the-art performance and successfully enables uncertainty-aware active perception and grasp planning. Key contributions include: (1) the first flow-matching-based probabilistic modeling paradigm on SE(3); (2) an end-to-end differentiable, geometrically consistent estimator for multimodal pose distributions; and (3) practical, closed-loop validation of uncertainty utilization in real-world robotic manipulation tasks.

Technology Category

• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Security and Privacy: Large-scale security measurements

Abstract

Object pose estimation is a fundamental problem in robotics and computer vision, yet it remains challenging due to partial observability, occlusions, and object symmetries, which inevitably lead to pose ambiguity and multiple hypotheses consistent with the same observation. While deterministic deep networks achieve impressive performance under well-constrained conditions, they are often overconfident and fail to capture the multi-modality of the underlying pose distribution. To address these challenges, we propose a novel probabilistic framework that leverages flow matching on the SE(3) manifold for estimating 6D object pose distributions. Unlike existing methods that regress a single deterministic output, our approach models the full pose distribution with a sample-based estimate and enables reasoning about uncertainty in ambiguous cases such as symmetric objects or severe occlusions. We achieve state-of-the-art results on Real275, YCB-V, and LM-O, and demonstrate how our sample-based pose estimates can be leveraged in downstream robotic manipulation tasks such as active perception for disambiguating uncertain viewpoints or guiding grasp synthesis in an uncertainty-aware manner.

Problem

• Estimating 6D object pose distributions under uncertainty
• Addressing pose ambiguity from occlusions and symmetries
• Enabling uncertainty-aware robotic manipulation and perception

Innovation

• Flow matching on SE(3) for 6D pose distributions
• Sample-based estimation of full pose distribution
• Uncertainty-aware reasoning for ambiguous robotic manipulation