CAD-Refiner：面向 CAD 生成与迭代编辑的统一框架
CAD-Refiner: A Unified Framework for CAD Generation and Iterative Editing
将多模态输入解析为拓扑结构图，并用统一解码器与序列注入策略支持 CAD 生成与迭代编辑，为 3D 表示的结构化生成与可控编辑提供可复用范式。
原文依据 we propose CAD-Refiner, a unified framework that supports free-form multimodal inputs and enables iterative refinement over previously generated results.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论计算机辅助设计（CAD）建模是众多工业应用的基础。在概念设计阶段，设计师通常会对初始方案进行迭代优化以达到预期结果。AI 辅助 CAD 的一个关键目标是支持从初始生成到迭代优化的完整建模流程。然而，现有方法大多将生成与编辑视为独立任务，这阻碍了真实场景中的连贯性与适应性。为解决这一局限，我们提出 CAD-Refiner，一个支持自由形式多模态输入并能对先前生成结果进行迭代优化的统一框架。具体而言，我们设计了一个名为 CAD Insighter 的智能体，将多模态输入解析为拓扑结构图，显式表示 CAD 对象中的基本元素及其关系。随后，我们提出精心设计的解码器架构和序列注入策略（SIS），以在统一建模框架内支持多种应用。此外，我们提出 CAD Checker，一个错误感知反馈模块，在优化过程中执行基于几何的奖励塑形，提升建模质量和几何有效性。另外，我们引入 MMCAD，这是 DeepCAD 的多模态扩展，专为 CAD 生成与编辑而设计。大量实验证明了 CAD-Refiner 在多个任务上的有效性。
英文原文摘要 1418 字符
Computer-Aided Design (CAD) modeling underpins a wide range of industrial applications. During the conceptual design phase, designers often refine initial solutions iteratively to achieve desired results. A key goal of AI-assisted CAD is to support the full modeling workflow from initial generation to iterative refinement. However, most existing approaches treat generation and editing as separate tasks, hindering coherence and adaptability in real-world scenarios. To address this limitation, we propose CAD-Refiner, a unified framework that supports free-form multimodal inputs and enables iterative refinement over previously generated results. Specifically, we design an agent named CAD Insighter that interprets multimodal inputs into topological structure graphs, which explicitly represent the fundamental elements and their relationships within CAD objects. We then propose a carefully designed decoder architecture and a Sequence Injection Strategy (SIS) to enable multiple applications within a unified modeling framework. Furthermore, we propose CAD Checker, an error-aware feedback module that performs geometry-based reward shaping during optimization, enhancing modeling quality and geometric validity. Additionally, we introduce MMCAD, a multimodal extension of DeepCAD tailored for CAD generation and editing. Extensive experiments demonstrate the effectiveness of CAD-Refiner across multiple tasks.