Region-aware Grasp Framework with Normalized Grasp Space for Efficient 6-DoF Grasping

TL;DR

This paper introduces a novel region-aware normalized grasp space and a corresponding network, RNGNet, which significantly improves 6-DoF grasp detection in cluttered scenes with real-time performance and robustness in dynamic scenarios.

Contribution

The paper proposes the Normalized Grasp Space and RNGNet, enhancing grasp detection accuracy and efficiency in cluttered environments, with improved generalizability and real-time capabilities.

Findings

Achieves over 20% performance improvement on benchmark datasets.

Operates at approximately 50 FPS in real-time.

Effective in real-world cluttered scene clearance and dynamic grasping scenarios.

Abstract

A series of region-based methods succeed in extracting regional features and enhancing grasp detection quality. However, faced with a cluttered scene with potential collision, the definition of the grasp-relevant region stays inconsistent, and the relationship between grasps and regional spaces remains incompletely investigated. In this paper, we propose Normalized Grasp Space (NGS) from a novel region-aware viewpoint, unifying the grasp representation within a normalized regional space and benefiting the generalizability of methods. Leveraging the NGS, we find that CNNs are underestimated for 3D feature extraction and 6-DoF grasp detection in clutter scenes and build a highly efficient Region-aware Normalized Grasp Network (RNGNet). Experiments on the public benchmark show that our method achieves significant >20% performance gains while attaining a real-time inference speed of approximately 50 FPS. Real-world cluttered scene clearance experiments underscore the effectiveness of our method. Further, human-to-robot handover and dynamic object grasping experiments demonstrate the potential of our proposed method for closed-loop grasping in dynamic scenarios.