SBIR Phase I: Designing the Future: Generative Configuration Design

这个小企业创新研究（SBIR）第一阶段项目开发生成式人工智能（AI）算法，帮助工程师和设计师开发新产品。作为一种可持续的软件即服务（SaaS）业务模型，工程师可以比“头脑风暴”更快地进行设计，从而发现新颖、高效的解决方案。对于航空航天和国防部门，该技术将用于快速设计创新解决方案，以应对潜在的近对手日益增长的威胁。较短的设计周期减少了重新设计解决方案以适应快速变化的项目需求的浪费。这项技术也可以用于航空旅行的脱碳。通过降低进入设计工程的门槛，该项目将使更广泛的美国民众参与设计、工程、产品开发和发明。该项目的成果可以加速在各个行业推广更安全，更高效，更具成本效益的产品。这个小型企业创新研究（SBIR）第一阶段项目推进了生成人工智能的最新技术，特别是关于算法处理复杂非媒体数据类型的能力，以及在没有大型预先存在的数据库的情况下开发可以生成新型网络物理系统架构的方法。目前，计算机辅助工程软件擅长于为给定系统体系结构的动态提供精确的分析结果，但几乎没有提供关于能够满足性能要求的各种体系结构的信息。与此同时，生成式人工智能目前在不受性能、物理或逻辑限制的情况下，擅长生成媒体产品。该项目将开发一个框架，将基于仿真的物理信息整合到生成算法中，使工程师能够创建物理上可实现的系统。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。

This Small Business Innovation Research (SBIR) Phase I project develops generative artificial intelligence (AI) algorithms that assist engineers and designers in developing new products. As a sustainable Software-as-a-Service (SaaS) business model, engineers can design more quickly than “brainstorming” and thereby discover novel, high performing solutions. For the aerospace and defense sector, this technology will be used to quickly design innovative solutions to counter growing threats from potential near-peers. The short design cycles lead to less wasted effort in reengineering solutions to fit within rapidly changing program requirements. The technology may also be used to decarbonize air travel. By lowering the barriers to entry for design engineering, this project will enable a broader cross-section of the American populace to engage with design, engineering, product development, and invention. The results of this project can accelerate the promotion of safer, more efficient, and more cost-effective products in various industries.This Small Business Innovation Research (SBIR) Phase I project advances the state of the art in generative artificial intelligence, particularly regarding algorithms’ ability to engage with complex non-media datatypes and develop methods that can generate novel cyber-physical system architectures in the absence of large pre-existing databases. Currently, computer-aided engineering software excels at rendering precise analytical results for the dynamics of a given system architecture but offers little to no information as to the variety of architectures that can satisfy performance requirements. Simultaneously, generative AI currently excels at generating media products without the constraint of performance, physics, or logic. This project will develop a framework for incorporating simulation-based physics information into generative algorithms to enable engineers to create physically realizable systems.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.