用语法化语言生成无碰撞3D室内场景，提升真实感与可验证性。
SpatialGrammar: A Domain-Specific Language for LLM-Based 3D Indoor Scene Generation
- 设计专用语法表示3D布局，支持可验证的空间约束检查
- 闭环系统迭代优化，生成159个场景中空间准确性显著提升
- 小模型仅104M参数却媲美大模型，适合资源受限场景
从自然语言自动生成可交互的3D室内场景对虚拟现实、游戏和具身AI至关重要。现有基于大模型的方法常出现空间错误与碰撞，部分原因在于常用场景表示（如原始坐标或冗长代码）难以让模型理解三维空间关系与物理约束。我们提出SpatialGrammar，一种领域专用语言，将重力对齐的室内布局表示为BEV网格放置，并可确定性编译为有效3D几何结构，实现可验证的约束检查。基于此表示，我们构建了(1) SG-Agent——利用编译器反馈进行闭环迭代优化，强制满足碰撞约束；(2) SG-Mini——一个在编译器验证的合成数据上训练的104M参数模型。在涵盖五种不同复杂度场景的159个测试用例中，SG-Agent在空间保真度和物理合理性上优于先前方法，而SG-Mini在单次生成任务中表现媲美更大的基于大模型的基线。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。