CAREER: SmartCAD: Shaping The Next Revolution in Computer-Aided Design

人工智能（AI）的快速发展正在重塑各个行业，但制造设计领域并没有充分利用这些创新。该项目旨在通过开发将人工智能与传统设计算法和原则相结合的新设计工具来弥合这一差距。制造业中设计工具的核心挑战是需要同时培养创新能力，构思新颖的设计和分析能力，以批判性地评估和优化功能和生产。该项目的目标是解决这些挑战，并开发新颖的工具，使新手和经验丰富的设计师更容易创造出高质量，多样化的产品。这些工具不仅可以提高产品质量和多样性，还可以减少制造对环境的影响，促进可持续性。此外，该项目还包括一个教育部分，通过设计和制造使STEM领域更容易进入和吸引人，扩大这些领域的参与和创新。为了将设计精度与创造性探索结合起来，该项目将开发将神经抽象与形式推理结合起来的工具。将计算机辅助设计（CAD）模型视为生成几何图形的程序，该计划将设计解释为程序综合，受益于形式推理和人工智能驱动代码生成的进步。这种方法确保了可验证性和约束约束综合，这对精确的工程分析至关重要。该项目将开发一种新的领域特定语言（DSL），以更好地支持人工智能驱动的设计，创建算法，使用与几何知识集成的生成预训练变形器（GPT）等模型，从不同的输入生成DSL程序，并设计有效识别和探索设计变化的方法。此外，它将建立设计分解和重组的技术，增强DSL的灵活性和用户在设计生成和迭代中的创新能力。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The rapid advancement of artificial intelligence (AI) is reshaping various sectors, yet the field of design for manufacturing has not fully capitalized on these innovations. This project aims to bridge this gap by developing new design tools that combine AI with traditional design algorithms and principles. A central challenge for design tools in manufacturing is the need to simultaneously nurture the creative ability to conceive novel designs and the analytical prowess to critically evaluate and optimize functionality and production. The goal of this project is to address these challenges and develop novel tools that make it easier for both new and experienced designers to create high-quality, diverse products. These tools are expected to not only improve product quality and diversity but also reduce the environmental impact of manufacturing, promoting sustainability. Additionally, the project includes an educational component to make STEM fields more accessible and engaging through design and manufacturing, broadening participation and innovation in these areas.To unite design precision with creative exploration, this project will develop tools that combine neural abstractions with formal reasoning. Treating computer-aided design (CAD) models as programs that generate geometry, this initiative interprets design as program synthesis, benefiting from formal reasoning and advancements in AI-driven code generation. This approach ensures verifiability and constraint-adherent synthesis, essential for precise engineering analysis. The project will develop a new domain-specific language (DSL) to better support AI-driven design, create algorithms for generating DSL programs from diverse inputs using models like Generative Pre-trained Transformers (GPT) integrated with geometric knowledge, and devise methods to identify and explore design variations efficiently. Additionally, it will establish techniques for design decomposition and recomposition, enhancing the DSL's flexibility and users’ ability to innovate in design generation and iteration.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.