Research map / 3D vision / 3D generation
Fixed Reality, Diffused Possibility: Disentangling Stochastic and Deterministic Latent for Cluttered Grasping
Hritam Basak, Zhaozheng Yin
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 |
|---|---|---|---|
| replace single global latent with fixed latent only (ablation) | grasp model · DexGraspNet-2.0 GraspNet-1B Dense | success rate: 87.0 -> 86.5 | No clear effect |
| add diffused latent (Global Conditional Diffusion) to fixed latent (ablation) | split-latent model · DexGraspNet-2.0 GraspNet-1B Dense | success rate: 86.5 -> 88.5 | Improved |
| add diffused latent to fixed latent (ablation) | split-latent model · GraspNet-1Billion AP Novel | AP: 33.5 -> 35.8 | Improved |
| add Local Conditional Diffusion (ablation) | split-latent model · DexGraspNet-2.0 GraspNet-1B Loose | success rate: 69.8 -> 73.6 | Improved |
| add LCD (ablation) | split-latent model · GraspNet-1Billion AP Novel | AP: 35.8 -> 37.4 | Improved |
| add Harmonic Grasp Field (ablation) | split-latent + LCD · DexGraspNet-2.0 GraspNet-1B Dense | success rate: 89.3 -> 93.1 | Improved |
| add Harmonic Grasp Field (ablation) | split-latent + LCD · GraspNet-1Billion AP Novel | AP: 37.4 -> 40.9 | Improved |
| full model (fixed + diffused + LCD + HGF) vs single global latent (ablation) | full model · all benchmarks (Table 5) | success rate / AP | Improved |
| integrate Ours into A2-G-Place policy | A2-G-Place · A2 place Seen | success rate: 32.3 -> 35.7 | Improved |
| integrate Ours into A2 policy | A2 · A2 pick Seen | success rate: 95.3 -> 96.3 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GSNet — GSNet: Joint Vehicle Pose and Shape Reconstruction with Geometrical and Scene-aware Supervision (ECCV 2020) · better on 3 of 3 shared results
- AffordGrasp — AffordGrasp: Cross-Modal Diffusion for Affordance-Aware Grasp Synthesis (CVPR 2026) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · similarity 0.88)
- SeqAfford: Sequential 3D Affordance Reasoning via Multimodal Large Language Model (CVPR 2025 · similarity 0.87)
- G-HOP: Generative Hand-Object Prior for Interaction Reconstruction and Grasp Synthesis (CVPR 2024 · similarity 0.87)
- Grasp2Grasp: Vision-Based Dexterous Grasp Translation via Schrödinger Bridges (NeurIPS 2025 · similarity 0.86)
- LatentHOI: On the Generalizable Hand Object Motion Generation with Latent Hand Diffusion. (CVPR 2025 · similarity 0.86)
- Contact Map Transfer with Conditional Diffusion Model for Generalizable Dexterous Grasp Generation (NeurIPS 2025 · similarity 0.86)
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.
