vMF-Contact: Uncertainty-aware Evidential Learning for Probabilistic Contact-grasp in Noisy Clutter

2024-11-06

arXiv.org

AI Summary

Addressing the challenge of uncertainty modeling for 6-DoF grasp detection under noisy conditions—including occlusion, sensor noise, and out-of-distribution (OOD) objects—this paper proposes vMF-Contact, the first uncertainty-aware framework that probabilistically models grasp orientation using the von Mises–Fisher (vMF) distribution. Methodologically, it integrates evidential deep learning to disentangle epistemic and aleatoric uncertainties, derives a theoretically grounded second-order objective to ensure reliable evidence calibration, and incorporates partial point cloud reconstruction as an auxiliary task to enhance generalization. Evaluated on real robotic platforms, vMF-Contact achieves a 39% improvement in clutter-clearing success rate over baseline methods, supports single-pass real-time inference, and significantly improves robustness to OOD objects and reliability of contact-based grasping.

Technology Category

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

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Grasp learning in noisy environments, such as occlusions, sensor noise, and out-of-distribution (OOD) objects, poses significant challenges. Recent learning-based approaches focus primarily on capturing aleatoric uncertainty from inherent data noise. The epistemic uncertainty, which represents the OOD recognition, is often addressed by ensembles with multiple forward paths, limiting real-time application. In this paper, we propose an uncertainty-aware approach for 6-DoF grasp detection using evidential learning to comprehensively capture both uncertainties in real-world robotic grasping. As a key contribution, we introduce vMF-Contact, a novel architecture for learning hierarchical contact grasp representations with probabilistic modeling of directional uncertainty as von Mises-Fisher (vMF) distribution. To achieve this, we analyze the theoretical formulation of the second-order objective on the posterior parametrization, providing formal guarantees for the model's ability to quantify uncertainty and improve grasp prediction performance. Moreover, we enhance feature expressiveness by applying partial point reconstructions as an auxiliary task, improving the comprehension of uncertainty quantification as well as the generalization to unseen objects. In the real-world experiments, our method demonstrates a significant improvement by 39% in the overall clearance rate compared to the baselines. The code is available under: https://github.com/YitianShi/vMF-Contact/

Problem

• Addresses grasp learning in noisy, cluttered environments.
• Captures both aleatoric and epistemic uncertainties in robotic grasping.
• Improves grasp prediction and generalization to unseen objects.

Innovation

• Evidential learning for 6-DoF grasp detection
• vMF-Contact architecture with vMF distribution
• Partial point reconstructions enhance feature expressiveness