Research map / Large language models / Agents, code and math
STABLE: Simulation-Ready Tabletop Layout Generation via a Semantics–Physics Dual System
Zhen Luo, Yixuan Yang, Xudong XU, Jinkun Hao, Zhaoyang Lyu, Feng Zheng, Jiangmiao Pang, Yanwei Fu
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 support-contact loss L_sup (ablation) | Physics Corrector · Physics Corrector ablation | floating rate: 9.8 -> 0 | Improved |
| add support-contact loss L_sup (ablation) | Physics Corrector · Physics Corrector ablation | object collisions: 4.7 -> 0 | Improved |
| add object-table collision loss (ablation) | Physics Corrector · Physics Corrector ablation | object collisions: 13.6 -> 0 | Improved |
| add object-table collision loss (ablation) | Physics Corrector · Physics Corrector ablation | floating rate: 5.4 -> 0 | Mixed |
| add object-object collision loss (ablation) | Physics Corrector · Physics Corrector ablation | object collisions: 11.9 -> 0 | Improved |
| add object-object collision loss (ablation) | Physics Corrector · Physics Corrector ablation | floating rate: 15.8 -> 0 | Improved |
| use progressive scene construction instead of one-shot generation (ablation) | Semantic Reasoner · Semantic Reasoner ablation | AwT (%): 89.9 -> 99.4 | Improved |
| use progressive scene construction instead of one-shot generation (ablation) | Semantic Reasoner · Semantic Reasoner ablation | distractor rate: 78.6 -> 86.1 | Improved |
| replace post-hoc optimization (MesaTask, Steerable) with learned Physics Corrector | STABLE · collision-resolution robustness | collision-free recovery at 20-40 initial collisions: fails -> succeeds at 20-30 and 30-40 collisions | Improved |
| replace Steerable post-processing with STABLE dual system | Steerable · MesaTask-10K | AwS (%): 91.1 -> 99.0 | Improved |
| use STABLE Physics Corrector instead of StructDiffusion | StructDiffusion · Tabletop rearrangement | EMD to GT: 0.23 -> 0.08 | Improved |
Closest papers by meaning
- MesaTask: Towards Task-Driven Tabletop Scene Generation via 3D Spatial Reasoning (NeurIPS 2025 · similarity 0.91)
- Direct Numerical Layout Generation for 3D Indoor Scene Synthesis via Spatial Reasoning (NeurIPS 2025 · similarity 0.87)
- PhyScene: Physically Interactable 3D Scene Synthesis for Embodied AI (CVPR 2024 · similarity 0.87)
- SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent (NeurIPS 2025 · similarity 0.87)
- LayoutVLM: Differentiable Optimization of 3D Layout via Vision-Language Models (CVPR 2025 · similarity 0.86)
- MetaSpatial: Reinforcing 3D Spatial Reasoning in VLMs for the Metaverse (ICLR 2026 · 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.
