Research map / 3D vision / Human pose estimation
ActiveGrasp: Information-Guided Active Grasping with Calibrated Energy-based Model
Boshu Lei, Wen Jiang, Kostas Daniilidis
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 |
|---|---|---|---|
| calibrate the energy-based model (Se3diff Calib vs Se3diff Scene) with ActiveGrasp view selection (ablation) | ActiveGrasp · Simulated clutter grasping | success rate: 73.00% -> 79.00% | Improved |
| use ActiveGrasp info-gain view selection instead of random with calibrated EBM (ablation) | Se3diff Calib · Simulated clutter grasping | success rate: 74.25% -> 79.00% | Improved |
| take 8 extra random views instead of active selection (ablation) | Se3diff Calib + Random · Simulated clutter grasping | success rate: 74.25% -> 76.50% | Improved |
| alternative NBV methods (Breyer, FisherRF, ActiveNGF, ACE) vs random | Se3diff Calib · Simulated clutter grasping | success rate: 73.50-74.00% vs 74.25% random | No clear effect |
| add scene information to Se3diff (ablation) | Se3diff · ACRONYM validation | AP: 34.69 -> 66.00 | Improved |
| add AP loss (ablation) | Se3diff+Scene · ACRONYM validation | AP: 66.00 -> 83.31 | Improved |
| add failure-grasp score matching (ablation) | Se3diff+Scene+AP · Bullet-executed grasps | ECE (Bullet): 0.17 -> 0.08 | Improved |
| add failure-grasp score matching (ablation) | Se3diff+Scene+AP · ACRONYM validation | Ang (acc.): 0.3517 -> 0.4119 | Hurt |
| add learnable temperature (ablation) | Se3diff+Scene+FSM+AP · ACRONYM validation | AP: 83.13 -> 87.84 | Improved |
| add learnable temperature (ablation) | Se3diff+Scene+FSM+AP · Bullet-executed grasps | ECE (Bullet): 0.08 -> 0.05 | Improved |
| ActiveGrasp vs ACE view selection | Se3diff Calib · Real-world setup 2 (red chip bottle) | successes out of 10: 4/10 -> 8/10 | Improved |
| ActiveGrasp vs best baseline in real world | Se3diff Calib · Real-world setup 3 (red tomato can) | successes out of 10: 8/10 -> 9/10 | Improved |
Closest papers by meaning
- Active Perception for Grasp Detection via Neural Graspness Field (NeurIPS 2024 · similarity 0.87)
- DexFuncGrasp: A Robotic Dexterous Functional Grasp Dataset Constructed from a Cost-Effective Real-Simulation Annotation System (AAAI 2024 · similarity 0.86)
- ZeroGrasp: Zero-Shot Shape Reconstruction Enabled Robotic Grasping (CVPR 2025 · similarity 0.86)
- CADGrasp: Learning Contact and Collision Aware General Dexterous Grasping in Cluttered Scenes (NeurIPS 2025 · similarity 0.85)
- Grasp2Grasp: Vision-Based Dexterous Grasp Translation via Schrödinger Bridges (NeurIPS 2025 · similarity 0.85)
- DexGrasp Anything: Towards Universal Robotic Dexterous Grasping with Physics Awareness (CVPR 2025 · similarity 0.85)
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.
