AI Paper Atlas

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

ICML 2026 · Large language models › Agents, code and math · approach: Long-horizon LLM agents · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add support-contact loss L_sup (ablation)Physics Corrector · Physics Corrector ablationfloating rate: 9.8 -> 0Improved
add support-contact loss L_sup (ablation)Physics Corrector · Physics Corrector ablationobject collisions: 4.7 -> 0Improved
add object-table collision loss (ablation)Physics Corrector · Physics Corrector ablationobject collisions: 13.6 -> 0Improved
add object-table collision loss (ablation)Physics Corrector · Physics Corrector ablationfloating rate: 5.4 -> 0Mixed
add object-object collision loss (ablation)Physics Corrector · Physics Corrector ablationobject collisions: 11.9 -> 0Improved
add object-object collision loss (ablation)Physics Corrector · Physics Corrector ablationfloating rate: 15.8 -> 0Improved
use progressive scene construction instead of one-shot generation (ablation)Semantic Reasoner · Semantic Reasoner ablationAwT (%): 89.9 -> 99.4Improved
use progressive scene construction instead of one-shot generation (ablation)Semantic Reasoner · Semantic Reasoner ablationdistractor rate: 78.6 -> 86.1Improved
replace post-hoc optimization (MesaTask, Steerable) with learned Physics CorrectorSTABLE · collision-resolution robustnesscollision-free recovery at 20-40 initial collisions: fails -> succeeds at 20-30 and 30-40 collisionsImproved
replace Steerable post-processing with STABLE dual systemSteerable · MesaTask-10KAwS (%): 91.1 -> 99.0Improved
use STABLE Physics Corrector instead of StructDiffusionStructDiffusion · Tabletop rearrangementEMD to GT: 0.23 -> 0.08Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy