Saved in:
Bibliographic Details
Main Authors: Tang, Song, Zhao, Kaiyong, Li, Yuliang, Yan, Qingsong, Sun, Penglei, Zou, Junyi, Wang, Qiang, Chu, Xiaowen
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.27555
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914520225546240
author Tang, Song
Zhao, Kaiyong
Li, Yuliang
Yan, Qingsong
Sun, Penglei
Zou, Junyi
Wang, Qiang
Chu, Xiaowen
author_facet Tang, Song
Zhao, Kaiyong
Li, Yuliang
Yan, Qingsong
Sun, Penglei
Zou, Junyi
Wang, Qiang
Chu, Xiaowen
contents Automatically generating interactive 3D indoor scenes from natural language is crucial for virtual reality, gaming, and embodied AI. However, existing LLM-based approaches often suffer from spatial errors and collisions, in part because common scene representations-raw coordinates or verbose code-are difficult for models to reason about 3D spatial relationships and physical constraints. We propose SpatialGrammar, a domain-specific language that represents gravity-aligned indoor layouts as BEV grid placements with deterministic compilation to valid 3D geometry, enabling verifiable constraint checking. Building on this representation, we develop (1) SG-Agent, a closed-loop system that uses compiler feedback to iteratively refine scenes and enforce collision constraints, and (2) SG-Mini, a 104M-parameter model trained entirely on compiler-validated synthetic data. Across 159 test scenes spanning five scenarios of different complexity, SG-Agent improves spatial fidelity and physical plausibility over prior methods, while SG-Mini performs competitively against larger LLM-based baselines on single-shot generation scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2604_27555
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SpatialGrammar: A Domain-Specific Language for LLM-Based 3D Indoor Scene Generation
Tang, Song
Zhao, Kaiyong
Li, Yuliang
Yan, Qingsong
Sun, Penglei
Zou, Junyi
Wang, Qiang
Chu, Xiaowen
Artificial Intelligence
Automatically generating interactive 3D indoor scenes from natural language is crucial for virtual reality, gaming, and embodied AI. However, existing LLM-based approaches often suffer from spatial errors and collisions, in part because common scene representations-raw coordinates or verbose code-are difficult for models to reason about 3D spatial relationships and physical constraints. We propose SpatialGrammar, a domain-specific language that represents gravity-aligned indoor layouts as BEV grid placements with deterministic compilation to valid 3D geometry, enabling verifiable constraint checking. Building on this representation, we develop (1) SG-Agent, a closed-loop system that uses compiler feedback to iteratively refine scenes and enforce collision constraints, and (2) SG-Mini, a 104M-parameter model trained entirely on compiler-validated synthetic data. Across 159 test scenes spanning five scenarios of different complexity, SG-Agent improves spatial fidelity and physical plausibility over prior methods, while SG-Mini performs competitively against larger LLM-based baselines on single-shot generation scenarios.
title SpatialGrammar: A Domain-Specific Language for LLM-Based 3D Indoor Scene Generation
topic Artificial Intelligence
url https://arxiv.org/abs/2604.27555