Research map / 3D vision / 3D generation
D-Grasp: Physically Plausible Dynamic Grasp Synthesis for Hand-Object Interactions
Sammy Christen, Muhammed Kocabas, Emre Aksan, Jemin Hwangbo, Jie Song, Otmar Hilliges
Check 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 |
|---|---|---|---|
| Separate grasping from motion synthesis hierarchically | D-Grasp · Dynamic grasp synthesis, 100 out-of-distribution target poses | MPE [mm]: 0.55 -> 0.08 | Improved |
| Keep the grasping policy active during motion synthesis | D-Grasp · Dynamic grasp synthesis, 100 out-of-distribution target poses | MPE [mm]: 0.45 -> 0.08 | Improved |
| Use closed-loop control instead of a learned motion synthesis policy | D-Grasp · Dynamic grasp synthesis, 100 out-of-distribution target poses | MPE [mm]: 0.30 -> 0.08 | Improved |
| Use the all-object policy rather than separate per-object policies | D-Grasp · DexYCB test split | Success: 0.42 -> 0.63 | Improved |
| Add the contact reward (ablation) | D-Grasp · DexYCB train split, grasping | Success: 0.0 -> 0.89 | Improved |
| Add the goal space (ablation) | D-Grasp · DexYCB train split, grasping | Success: 0.28 -> 0.89 | Improved |
| Add the feature extraction layer instead of using the original state space (ablation) | D-Grasp · DexYCB train split, grasping | Success: 0.47 -> 0.89 | Improved |
| Add wrist-guidance (ablation) | D-Grasp · DexYCB train split, grasping | Success: 0.58 -> 0.89 | Improved |
| Add the contact reward (ablation) | D-Grasp · DexYCB train split, grasping | Contact Ratio: 0.02 -> 0.43 | Improved |
| Add the goal space (ablation) | D-Grasp · DexYCB train split, grasping | Contact Ratio: 0.18 -> 0.43 | Improved |
| Add the feature extraction layer (ablation) | D-Grasp · DexYCB train split, grasping | Contact Ratio: 0.21 -> 0.43 | Improved |
| Add wrist-guidance (ablation) | D-Grasp · DexYCB train split, grasping | Contact Ratio: 0.28 -> 0.43 | Improved |
Closest papers by meaning
- Grasp’D: Differentiable Contact-Rich Grasp Synthesis for Multi-Fingered Hands (ECCV 2022 · similarity 0.93)
- GraspXL: Generating Grasping Motions for Diverse Objects at Scale (ECCV 2024 · similarity 0.91)
- SAGA: Stochastic Whole-Body Grasping with Contact (ECCV 2022 · similarity 0.90)
- DiffGrasp: Whole-Body Grasping Synthesis Guided by Object Motion Using a Diffusion Model (AAAI 2025 · similarity 0.90)
- Contact Map Transfer with Conditional Diffusion Model for Generalizable Dexterous Grasp Generation (NeurIPS 2025 · similarity 0.89)
- RAGG: Retrieval-Augmented Grasp Generation Model (AAAI 2025 · similarity 0.88)
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.
