Research map / 3D vision / Human pose estimation
6-DOF GraspNet: Variational Grasp Generation for Object Manipulation
Arsalan Mousavian, Clemens Eppner, Dieter Fox
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use 6-DOF GraspNet instead of GPD in real-world grasping | 6-DOF GraspNet · Real-world grasping | overall success rate: 47% -> 88% | Improved |
| use 6-DOF GraspNet instead of GPD on mugs | 6-DOF GraspNet · Real-world grasping, Mug | grasp success rate: 6% -> 86% | Improved |
| use 6-DOF GraspNet instead of GPD on boxes | 6-DOF GraspNet · Real-world grasping, Box | grasp success rate: 50% -> 83% | Improved |
| increase latent dimensionality from one to two (ablation) | VAE grasp sampler · Simulated grasp evaluation (held-out objects) | success-coverage AUC | Improved |
| increase latent dimensionality from two to three or four (ablation) | VAE grasp sampler · Simulated grasp evaluation (held-out objects) | success-coverage AUC | Hurt |
| add gradient-based grasp refinement iterations using the evaluator (ablation) | 6-DOF GraspNet · Simulated grasp evaluation (held-out objects) | success and coverage rate | Improved |
| continue refinement beyond 10 iterations (ablation) | 6-DOF GraspNet · Simulated grasp evaluation (held-out objects) | success-coverage AUC | No clear effect |
| sample more latent values (2000 instead of 200 grasps) (ablation) | VAE grasp sampler · Simulated grasp evaluation (held-out objects) | coverage rate | Improved |
| use learned VAE sampler (latent size 2) instead of geometric normal-based sampler (ablation) | 6-DOF GraspNet · Simulated grasp evaluation (held-out objects) | success and coverage rate | Improved |
Later papers that compared with this method
1 later accepted paper reports 6-DOF GraspNet as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- TransGrasp: Grasp Pose Estimation of a Category of Objects by Transferring Grasps from Only One Labeled Instance (ECCV 2022 · better on 7 of 9 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Grasp Proposal Networks: An End-to-End Solution for Visual Learning of Robotic Grasps (NeurIPS 2020 · similarity 0.90)
- Generalizing 6-DoF Grasp Detection via Domain Prior Knowledge (CVPR 2024 · similarity 0.87)
- Contact Map Transfer with Conditional Diffusion Model for Generalizable Dexterous Grasp Generation (NeurIPS 2025 · similarity 0.87)
- LatentHOI: On the Generalizable Hand Object Motion Generation with Latent Hand Diffusion. (CVPR 2025 · similarity 0.86)
- RAGG: Retrieval-Augmented Grasp Generation Model (AAAI 2025 · similarity 0.86)
- DexFuncGrasp: A Robotic Dexterous Functional Grasp Dataset Constructed from a Cost-Effective Real-Simulation Annotation System (AAAI 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
