用大模型生成可执行的CAD脚本，让自然语言直接操控几何形状。
Towards High-Fidelity CAD Generation via LLM-Driven Program Generation and Text-Based B-Rep Primitive Grounding
- 通过大模型理解自然语言指令，自动生成可运行的CadQuery代码。
- 在真实工业设计数据集上达到当前最佳生成精度，支持复杂特征如倒角和圆角。
- 适合需要快速原型设计的工程师或想用文字建模的初学者。
计算机辅助设计（CAD）生成近年来取得显著进展，现有方法主要分为参数化建模与直接边界表示（B-Rep）合成两类。在现代特征驱动的CAD系统中，参数化建模与B-Rep本质关联：高级参数操作（如倒角、斜角）需显式选择B-Rep几何体，而B-Rep本身也由参数化操作生成。这一范式差距仍是限制人工智能在复杂工业产品设计中应用的关键瓶颈。本文提出FutureCAD，一种新颖的文本到CAD框架，融合大语言模型（LLM）与B-Rep接地变换器（BRepGround），实现高保真度的CAD生成。该方法生成可执行的CadQuery脚本，并引入基于文本的查询机制，使大模型能通过自然语言指定几何选择，BRepGround则将其精准定位至目标几何体。为训练该框架，我们构建了一个包含真实世界CAD模型的新数据集。对大模型采用监督微调（SFT）建立基础生成能力，再通过强化学习（RL）提升泛化性能。实验表明，FutureCAD在多项指标上达到当前最优表现。代码与数据集已开源：https://github.com/JohanStackk/FutureCAD。
原文摘要 · Abstract (English)
The field of Computer-Aided Design (CAD) generation has made significant progress in recent years. Existing methods typically fall into two separate categories: parametric CAD modeling and direct boundary representation (B-Rep) synthesis. In modern feature-based CAD systems, parametric modeling and B-Rep are inherently intertwined, as advanced parametric operations (e.g., fillet and chamfer) require explicit selection of B-Rep geometric primitives, and the B-Rep itself is derived from parametric operations. Consequently, this paradigm gap remains a critical factor limiting AI-driven CAD modeling for complex industrial product design. This paper presents FutureCAD, a novel text-to-CAD framework that leverages large language models (LLMs) and a B-Rep grounding transformer (BRepGround) for high-fidelity CAD generation. Our method generates executable CadQuery scripts, and introduces a text-based query mechanism that enables the LLM to specify geometric selections via natural language, which BRepGround then grounds to the target primitives. To train our framework, we construct a new dataset comprising real-world CAD models. For the LLM, we apply supervised fine-tuning (SFT) to establish fundamental CAD generation capabilities, followed by reinforcement learning (RL) to improve generalization. Experiments show that FutureCAD achieves state-of-the-art CAD generation performance. Code and dataset are available at: https://github.com/JohanStackk/FutureCAD
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。