RII Track-2 FEC: Explainable and Adaptable Artificial Intelligence for Advanced Manufacturing

RII Track-2 FEC：用于先进制造的可解释且适应性强的人工智能

• 批准号: 2218063
• 负责人: Yifeng Zhu
• 金额: $ 600万
• 依托单位: University of Maine
• 依托单位国家: 美国
• 项目类别: Cooperative Agreement
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-08-01 至 2026-07-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2218063&HistoricalAwards=false
• 关键词: RII; Track; FEC; Explainable; Adaptable

Advanced technologies have radically transformed manufacturing and are essential to modern economic prosperity. The goal of this project is to leverage emerging technologies, i.e., artificial intelligence (AI), 3D metal printing, and robotics, to increase the quality, capability, safety, and sustainability of Advanced Manufacturing (AdvMfg) in northern New England. The project will also encourage the adoption of new technologies in industry to address manufacturing challenges facing the region. These two objectives will be accomplished by creating a scientifically- and geographically-interlinked team, i.e., Northeast Integrated Intelligent Manufacturing Lab (NIIM), consisting of members from the University of Maine, University of New Hampshire, University of Vermont, Dartmouth College, Southern Maine Community College, and Vermont Technical College communities. Although initial funding for NIIM is from a National Science Foundation (NSF) Research Infrastructure Improvement Track-2 Focused EPSCoR Collaboration (RII Track-2 FEC) award, NIIM will sustainably impact the EPSCoR jurisdictions of Maine (ME), New Hampshire (NH), and Vermont (VT) for years to come. NIIM will draw on the unique strengths and rich assets of each state, and fully leverage existing state and federal investments. The project's research team, led by early career faculty and senior mentors, will investigate how to integrate state-of-the-art AI techniques into modern manufacturing processes and systems. A proactive large-scale workforce and economic development assessment will identify the technological needs of firms in the region, which will inform project research and outreach activities, as well as identify skills gaps and opportunities for training and building career pathways in AdvMfg. The project will extend STEM experiences to undergraduates and graduates, especially those underrepresented in STEM fields. This project will also create new components for Upward Bound for low-income high school students, who will potentially be first-generation undergraduate students, and Northeast Passage for disabled students and workers at community and technical colleges. The team will work closely with the manufacturing extension partnership programs (MEPs) in the three states, an industrial advisory board, industry partners, and the US Economic Development Administration University Center for Economic Development. Working with these organizations will ensure that this Track-2 project remains closely tied to state and regional economic development priorities.In this era of Industry 4.0, intelligent tools and techniques are opening new dimensions to optimize manufacturing processes and systems. The Northeast Integrated Intelligent Manufacturing Lab (NIIM), established in this project, aims to create a new, explainable and adaptable AI framework that fills existing and future technology gaps in manufacturing, such as long and expensive experiments and simulations, lack of coordination among multiple machines, and difficulty in programming robots for complicated manufacturing tasks. Our convergent research teams across three EPSCoR jurisdictions (ME, NH and VT) will work closely with industry to create: (a) new AI models with intrinsic interpretability and increased adaptability to support Advanced Manufacturing (AdvMfg); (b) AI-guided design for additive manufacturing of metals that seamlessly connects multi-scale modeling and property predictions without unnecessary trial-and-error; (c) self-aware CNC machines that optimize the coordination and control in subtractive manufacturing; (d) industrial robots that efficiently and safely learn from video demonstrations for cellular manufacturing; (e) an industry-driven, unified hybrid manufacturing framework; and (f) an understanding of the factors that influence the adoption of new technologies by manufacturing businesses. The project anticipates specific outcomes that will be of immediate relevance to AdvMfg companies in Northern New England. For example, it is expected that the project will yield sample-efficient robot learning techniques that will enable factory workers to teach robots new skills through visual demonstrations, allow robots to learn from failure and request relevant demonstrations, and generate risk-bounded safe policies using uncertainty aware learning. This project will serve the northern New England manufacturing sector through relevant research, workforce development, and education. Diversity and inclusion efforts are integrated to remove barriers to STEM education for underrepresented, low income, potential first-generation, and/or disabled individuals.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.

先进的技术已经从根本上改变了制造业，对现代经济繁荣至关重要。该项目的目的是利用新兴技术（即人工智能（AI），3D金属印刷和机器人技术）来提高新英格兰北部高级制造业（ADVMFG）的质量，能力，安全性和可持续性。该项目还将鼓励在行业中采用新技术来应对该地区面临的制造挑战。这两个目标将通过创建一个科学和地理上的相互联系的团队，即东北综合智能制造实验室（NIIM），由缅因州，新罕布什尔大学，佛蒙特大学，达特茅斯大学，达特茅斯大学，缅因州社区学院和佛蒙特州技术学院社区组成。尽管NIIM的最初资金来自国家科学基金会（NSF）研究基础设施改进Track-2专注于EPSCOR合作（RII Track-2 FEC）奖，但NIIM将可持续影响缅因州的EPSCOR司法管辖区（ME），新汉普郡（NH）（NH）（NH）和佛蒙特州（VT）（VT）。 NIIM将利用每个州的独特优势和丰富的资产，并充分利用现有的州和联邦投资。该项目的研究团队由早期职业教师和高级导师领导，将调查如何将最先进的AI技术整合到现代制造过程和系统中。积极主动的大规模劳动力和经济发展评估将确定该地区公司的技术需求，这将为项目研究和外展活动提供依据，并确定技能差距以及在Advmfg中培训和建立职业途径的机会。该项目将把STEM的经验扩展到本科生和毕业生，尤其是在STEM领域的人数不足的人。该项目还将为低收入高中学生提供新的组成部分，他们可能是第一代本科生，以及东北通道，为社区和技术学院的残疾学生和工人提供东北通道。该团队将与三个州的制造扩展合作计划（MEP）紧密合作，工业顾问委员会，行业合作伙伴和美国经济发展管理局经济发展中心。与这些组织的合作将确保该Track-2项目与州和区域经济发展的重点密切相关。在这个行业4.0时代，智能工具和技术正在为优化制造过程和系统开放新的维度。该项目中建立的东北综合智能制造实验室（NIIM）旨在创建一个新的，可解释的和适应性的AI框架，填补了制造业中现有和未来的技术差距，例如长期且昂贵的实验和模拟，缺乏多个机器之间的协调以及在复杂制造任务的编程机器人方面的难度。我们在三个EPSCOR司法管辖区（ME，NH和VT）的收敛研究团队将与行业紧密合作，以创建：（a）具有内在可解释性的新型AI模型，并提高了支持高级制造（ADVMFG）的适应性； （b）用于无缝连接多尺度建模和财产预测的金属添加剂的AI引导设计，而无需不必要的试用和纠正； （c）在减法制造中优化协调和控制的自我了解的CNC机器； （d）有效，安全地从蜂窝制造的视频演示中学习的工业机器人； （e）行业驱动的，统一的混合制造框架； （f）了解影响通过制造业务采用新技术的因素。该项目可以预期与新英格兰北部的Advmfg公司直接相关的具体结果。例如，预计该项目将产生样本效率的机器人学习技术，这将使工厂工人能够通过视觉演示来教机器人新技能，使机器人能够从失败中学习并要求相关的演示，并使用不确定性学习学习产生风险结合的安全政策。该项目将通过相关研究，劳动力发展和教育为北部新英格兰制造业提供服务。多样性和包容性工作旨在消除STEM教育的障碍，以提供代表性不足，低收入，潜在的第一代和/或残疾人的个人。该奖项反映了NSF的法定任务，并被认为是值得通过基金会的知识分子优点和更广泛的影响审查标准来评估的。

项目成果

OmnImage: Evolving 1k Image Cliques for Few-Shot Learning

• DOI: 10.1145/3583131.3590430
• 发表时间: 2023-07
• 期刊: Proceedings of the Genetic and Evolutionary Computation Conference
• 作者: Lapo Frati;Neil Traft;Nick Cheney

Many-objective Optimization via Voting for Elites

• DOI: 10.1145/3583133.3590693
• 发表时间: 2023-07
• 期刊: Proceedings of the Companion Conference on Genetic and Evolutionary Computation
• 作者: Jackson Dean;Nick Cheney