Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
GraspGen-X: Cross-Embodiment 6-DOF Diffusion-based Grasping
Beining Han, Yu-Wei Chao, Erwin Coumans, Clemens Eppner, Jia Deng, Stan Birchfield, Adithyavairavan Murali
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 |
|---|---|---|---|
| retarget Franka grasps by approach offset instead of direct transfer | GraspGen (Franka) · Zero-shot novel grippers, all 10 | mAUC: 0.126 -> 0.398 | Improved |
| train cross-embodiment model conditioned on swept volume instead of retargeting | GraspGen-RTG · Zero-shot novel grippers, all 10 | mAUC: 0.398 -> 0.506 | Improved |
| use GraspGen-X vs retargeting on high-dof 3-finger grippers | GraspGen-RTG · Zero-shot high-dof 3F | mAUC: 0.503 -> 0.699 | Improved |
| use Swept Volume gripper encoding instead of TSDF (AdaGrasp) (ablation) | GraspGen-X · Gripper encoder comparison (453 objects) | mAUC: 0.432 -> 0.528 | Improved |
| use Swept Volume instead of PointNet++ gripper encoding (ablation) | GraspGen-X · Gripper encoder comparison (453 objects) | mAUC: 0.349 -> 0.528 | Improved |
| train on procedural grippers instead of 10 real grippers (ablation) | GraspGen-X · 10 novel test grippers | mAUC | Improved |
| use 50 procedural training grippers instead of 25 (ablation) | GraspGen-X · 10 novel test grippers | mAUC | Improved |
| condition on fully open and half open swept volume instead of fully open only (ablation) | GraspGen-X · 10 novel test grippers | mAUC | Improved |
| condition on swept volume instead of gripper type only or retarget offset only (ablation) | GraspGen-X · 10 novel test grippers | mAUC | Improved |
| add gripper type one-hot to swept volume input (ablation) | GraspGen-X · 10 novel test grippers | mAUC | No clear effect |
| initialize fine-tuning from GraspGen-X instead of Franka model or scratch | GraspGen generator · fine-tuning on novel grippers | recall / pose error | Improved |
| use GraspGen-X vs GraspGen-RTG on real cluttered shelf | GraspGen-RTG · Real robot Robotiq-2F140 clutter | grasp success rate: 57.1% -> 71.4% | Improved |
Closest papers by meaning
- Contact Map Transfer with Conditional Diffusion Model for Generalizable Dexterous Grasp Generation (NeurIPS 2025 · similarity 0.89)
- Generalizing 6-DoF Grasp Detection via Domain Prior Knowledge (CVPR 2024 · similarity 0.89)
- DexGrasp Anything: Towards Universal Robotic Dexterous Grasping with Physics Awareness (CVPR 2025 · similarity 0.88)
- ZeroGrasp: Zero-Shot Shape Reconstruction Enabled Robotic Grasping (CVPR 2025 · similarity 0.87)
- DexFuncGrasp: A Robotic Dexterous Functional Grasp Dataset Constructed from a Cost-Effective Real-Simulation Annotation System (AAAI 2024 · similarity 0.87)
- LatentHOI: On the Generalizable Hand Object Motion Generation with Latent Hand Diffusion. (CVPR 2025 · similarity 0.87)
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.
