Research map / 3D vision / 3D generation
AdaAfford: Learning to Adapt Manipulation Affordance for 3D Articulated Objects via Few-Shot Interactions
Yian Wang, Ruihai Wu, Kaichun Mo, Jiaqi Ke, Qingnan Fan, Leonidas J. Guibas, Hao Dong
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 iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pushing all (train cat.) | F-score (%): 71.21/72.64/73.16 -> 72.78/73.12/75.18 | Improved |
| add iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pushing all (test cat.) | F-score (%): 77.24/77.33/77.17 -> 77.58/77.63/78.42 | Improved |
| add iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pulling all (train cat.) | F-score (%): 41.19/42.10/42.81 -> 42.62/43.87/44.08 | Improved |
| add iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pulling all (test cat.) | F-score (%): 48.31/48.28/50.50 -> 49.51/50.00/51.33 | Improved |
| add iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pulling closed door | F-score (%): 56.74/64.88/80.64 -> 57.83/65.60/79.65 | Mixed |
| add iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pushing faucet | F-score (%): 73.81/83.03/84.32 -> 77.42/83.06/83.83 | Mixed |
| add iterative alternating training of AAP and AIP modules (ablation) | AdaAfford · pushing all (test cat.) | Sample-Succ (%): 31.03/33.89/38.83 -> 34.97/36.75/37.40 | Mixed |
| add the learned AIP module instead of random interaction proposals | AdaAfford · pushing all (train cat.) | F-score (%): 70.24/70.58/70.85 -> 72.78/73.12/75.18 | Improved |
| add the learned AIP module instead of FPS interaction proposals | AdaAfford · pushing all (train cat.) | F-score (%): 64.32/69.58/70.99 -> 72.78/73.12/75.18 | Improved |
| replace heuristic adaptation with the learned AAP network | AdaAfford · pushing all (train cat.) | F-score (%): 64.16/65.42/64.99 -> 72.78/73.12/75.18 | Improved |
| add sequential selected interactions instead of four interactions in one shot | AdaAfford · pushing all (train cat.) | F-score (%): 72.13 -> 72.78/73.12/75.18 | Improved |
| add actionability scores to FPS interaction sampling (ablation) | AdaAfford · pushing all (test cat.) | Sample-Succ (%): 34.15/32.60/35.06 -> 33.64/35.19/37.79 | Improved |
Closest papers by meaning
- IAAO: Interactive Affordance Learning for Articulated Objects in 3D Environments (CVPR 2025 · similarity 0.92)
- BiAssemble: Learning Collaborative Affordance for Bimanual Geometric Assembly (ICML 2025 · similarity 0.89)
- AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · similarity 0.88)
- VidBot: Learning Generalizable 3D Actions from In-the-Wild 2D Human Videos for Zero-Shot Robotic Manipulation (CVPR 2025 · similarity 0.88)
- AdaManip: Adaptive Articulated Object Manipulation Environments and Policy Learning (ICLR 2025 · similarity 0.88)
- GMAP: Generalized Manipulation of Articulated Objects in Robotic Using Pre-trained Model (AAAI 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.
