Research map / Detection and segmentation / Open-set and few-shot detection
CoToGrasp: Contact-Topology-Conditioned Dexterous Grasp Synthesis via Canonical Workspace Learning
Julien Mérand, Boris Meden, Liming Chen, Mathieu GROSSARD
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 |
|---|---|---|---|
| add label-consistency check before optimization (ablation) | CoToGrasp · DexGraspNet test (taxonomy-aware) | TC (%): 14.45 -> 17.18 | Improved |
| add force-closure validation with latent resampling (ablation) | CoToGrasp · DexGraspNet test (taxonomy-aware) | TC (%): 16.26 -> 17.18 | Improved |
| add force-closure validation (ablation) | CoToGrasp · DexGraspNet test (taxonomy-aware) | avg topology SR (%): 22.08 -> 26.72 | Improved |
| add both validation checks (vs No Check) (ablation) | CoToGrasp · DexGraspNet test (taxonomy-aware) | avg topology SR (%): 21.06 -> 26.72 | Improved |
| condition on contact topology, object-agnostic, vs Dexonomy | grasp synthesis · DexGraspNet test (precision grasps) | SR (%): 12.36 -> 22.71 | Improved |
| use CoToGrasp instead of Dexonomy | taxonomy-aware grasp synthesis · DexGraspNet test (taxonomy-aware) | TC (%): 14.28 -> 17.18 | Improved |
| use topology-conditioned CoToGrasp vs unconditioned planners | MultiDex subset · MultiDex subset (Shadow Hand) | semantic entropy H_TC: 0.7389 -> 0.83 | Improved |
| use topology-conditioned CoToGrasp vs unconditioned planners | MultiDex subset · MultiDex subset (Shadow Hand) | SR (%): 77.90 -> 36.94 | Hurt |
| CoToGrasp vs Dexonomy on per-topology SR | per-topology comparison · DexGraspNet test, topologies M2/M6/M11/M13/M18/M21 | SR (%): M2 10.5 -> 30.3; M11 60.5 -> 29.8 | Mixed |
| project features onto canonical workspace via kNN aggregation (ablation) | DGCNN features · Feature alignment, matched pairs | cosine similarity: 0.35 -> 0.61 | Improved |
| add topology embedding and Transformer attention (ablation) | workspace features · Feature alignment, matched pairs | cosine similarity: 0.61 -> 0.66 | Improved |
Closest papers by meaning
- Contact Map Transfer with Conditional Diffusion Model for Generalizable Dexterous Grasp Generation (NeurIPS 2025 · similarity 0.89)
- CADGrasp: Learning Contact and Collision Aware General Dexterous Grasping in Cluttered Scenes (NeurIPS 2025 · similarity 0.89)
- ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning (CVPR 2025 · similarity 0.89)
- Grasp2Grasp: Vision-Based Dexterous Grasp Translation via Schrödinger Bridges (NeurIPS 2025 · similarity 0.89)
- DexGrasp Anything: Towards Universal Robotic Dexterous Grasping with Physics Awareness (CVPR 2025 · similarity 0.89)
- UniGraspTransformer: Simplified Policy Distillation for Scalable Dexterous Robotic Grasping (CVPR 2025 · similarity 0.89)
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.
