Skeleton-Guided Geometry-Aware Auto-Labeling for Robotic Grasping: From 4-DoF Learning to 6-DoF Execution
Skeleton-Guided Geometry-Aware Auto-Labeling for Robotic Grasping: From 4-DoF Learning to 6-DoF Execution
This research presents a skeleton-based algorithm for generating grasp keypoints and identifying stable grasp regions for robotic grasping. The proposed SkelGG-CNN model achieves high accuracy in both 4-DoF and 6-DoF grasping tasks, with performance comparable to state-of-the-art methods. The approach enables efficient training of lightweight models on small datasets and provides scalable annotations for cluttered 3D scenes.
This paper introduces a novel auto-labeling framework for robotic grasping that transitions from 4-DoF learning to 6-DoF execution, achieving high accuracy in both simulated and real-world settings. The method leverages skeleton-based guidance and point cloud processing to generate geometry-aware annotations, outperforming existing approaches without domain adaptation.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S4
Reported numbers
Reported success rate
98.4%
View original evidence
nd Cornell, respectively. On the GraspNet novel split, 6-DoF auto-labeling achieves an average precision of 18.33, comparable to state-of-the-art model-based approaches. In simulation, the 6-DoF method achieves success rates of 98.4% on Dex-Net and 89.1% on EGAD!. Real-world experiments with a Delta parallel robot show success rates of 97.4% on single household objects and 95.0% in cluttered scenes, outperforming sta
Open source S4Reported success rate
89.1%
View original evidence
nd Cornell, respectively. On the GraspNet novel split, 6-DoF auto-labeling achieves an average precision of 18.33, comparable to state-of-the-art model-based approaches. In simulation, the 6-DoF method achieves success rates of 98.4% on Dex-Net and 89.1% on EGAD!. Real-world experiments with a Delta parallel robot show success rates of 97.4% on single household objects and 95.0% in cluttered scenes, outperforming sta
Open source S4Reported success rate
97.4%
View original evidence
nd Cornell, respectively. On the GraspNet novel split, 6-DoF auto-labeling achieves an average precision of 18.33, comparable to state-of-the-art model-based approaches. In simulation, the 6-DoF method achieves success rates of 98.4% on Dex-Net and 89.1% on EGAD!. Real-world experiments with a Delta parallel robot show success rates of 97.4% on single household objects and 95.0% in cluttered scenes, outperforming sta
Open source S4Reported success rate
95%
View original evidence
nd Cornell, respectively. On the GraspNet novel split, 6-DoF auto-labeling achieves an average precision of 18.33, comparable to state-of-the-art model-based approaches. In simulation, the 6-DoF method achieves success rates of 98.4% on Dex-Net and 89.1% on EGAD!. Real-world experiments with a Delta parallel robot show success rates of 97.4% on single household objects and 95.0% in cluttered scenes, outperforming sta
Open source S4
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
What remains unknown
Not established in the collected evidence: Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
nd Cornell, respectively. On the GraspNet novel split, 6-DoF auto-labeling achieves an average precision of 18.33, comparable to state-of-the-art model-based approaches. In simulation, the 6-DoF method achieves success rates of 98.4% on Dex-Net and 89.1% on EGAD!. Real-world experiments with a Delta parallel robot show success rates of 97.4% on single household objects and 95.0% in cluttered scenes, outperforming sta
Open source S4
ative Grasp Convolutional Neural Network (SkelGG-CNN), a 4-DoF architecture that learns geometric grasp features from skeleton-based guidance. To extend auto-labeling to 6-DoF grasping, we develop a generalized quasi-3D skeletonization algorithm that operates directly on point clouds. Using automatically generated 4-DoF labels, SkelGG-CNN achieves accuracies of 95.42%, 96.55%, and 97.1% on Jacquard V1, Jacquard V2, a
Open source S3
Source：arXiv Robotics — research abstracts · arxiv.org