NSF AI institute in dynamic systems

The National Science Foundation Division of Computer and Network Systems awarded the University of Washington a $1 million cooperative agreement on October 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop AI technologies for complex engineering system design, testing, manufacturing, and operation.

The institute will create physics-aware machine learning and AI tools to accelerate aerospace and advanced manufacturing development by enabling engineers to predict, optimize, and control systems governed by physical laws. Work is organized into four research thrusts addressing foundational mathematical and computational methods for learning and control applied to industrial-scale aerospace design optimization, testing, and manufacturing. The institute will pursue partnerships with industry and national laboratories to evaluate new technologies on real-world engineering challenges and will expand educational impact through freely available courses, open software, and public datasets to train the next generation of scientists and engineers nationwide.

The award runs through September 30, 2031, with place of performance in Seattle, Washington. The University of Washington Office of Sponsored Programs serves as the prime recipient, administering the cooperative agreement through its grant and contracts division.

Abstract

The NSF artificial intelligence (AI) institute in dynamic systems will develop new AI technologies to improve how complex engineering systems are designed, tested, manufactured, and operated. Modern aircraft and advanced manufacturing systems are among the most sophisticated technologies ever created, yet their development remains slow and expensive. The institute will accelerate these processes by creating AI tools that help engineers predict, optimize, and control systems governed by the laws of physics. The institute will advance fundamental research at the intersection of AI, engineering, mathematics, and the physical sciences. The institute will develop reliable and trustworthy methods for modeling and decision-making. This will reduce the time and cost required to bring new aerospace technologies from concept to deployment while improving safety, performance, and efficiency. These advances will strengthen U.S. Competitiveness in commercial aviation and national defense. The project serves the national interest by advancing U.S. Leadership in AI, advanced manufacturing, aerospace innovation, and scientific discovery. Through partnerships with industry and national laboratories, new technologies will be evaluated on real-world engineering challenges. The institute will further deliver sustained impact by expanding educational opportunities through freely available courses, open software, and public datasets. IT will train the next generation of scientists and engineers and provide access to AI education nationwide. The NSF artificial intelligence institute in dynamic systems will pursue a rigorous scientific agenda, addressing core challenges in learning and control for industrial-scale aerospace design optimization, testing, and manufacturing. The research program is organized into four tightly coupled thrusts. First, the institute will develop foundational mathematical and computational tools for physics-aware machine learning, embedding stability constraints and physical laws to produce interpretable models that generalize reliably for aerospace design and control. Second, the institute will advance uncertainty-aware, continuously updated virtual representations (e.g., digital twins) of engineered systems, integrating simulation, sensing, and data across the full design-test-manufacture lifecycle, to enable real-time decision support and adaptive control. Third, the institute will develop artificial intelligence copilots for scientific discovery and engineering design, including domain-specific large language models that serve as interfaces for automated reasoning, hypothesis generation, model validation, and data consistency checking. Fourth, the institute will construct core benchmark infrastructure for the evaluation of AI systems, spanning a range of dynamic systems, including turbulence and shape optimization. Together, these thrusts establish the mathematical and computational foundations for a new generation of trustworthy, physics-grounded artificial intelligence tools, deployed and stress-tested against real aerospace challenges in partnership with industry and national laboratories. 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.- subawards are planned for this award.