Federal Project Grant Award Summary

The University of Michigan received a $367,296 Project Grant award from the National Science Foundation's Division of Computing and Communication Foundations under the Computer and Information Science and Engineering (CISE) program (CFDA 47.070). The CAREER award, effective May 1, 2026 through April 30, 2031, supports development of generative artificial intelligence (AI) systems that create fast, physically realistic four-dimensional simulations of dynamic environments. The research addresses critical limitations in current AI models that consume excessive computational resources and frequently violate physical laws—flaws that render them unsafe for applications requiring precision, such as medical surgery, robotics, and scientific simulations. The project will develop conditional sparse neural fields and autoregressive networks that optimize computational efficiency by sampling only critical spatial regions while maintaining full fidelity in areas of high physical importance. Integration of physical constraints based on partial differential equations, such as momentum conservation, ensures simulations remain physically plausible.

The award includes a significant educational component, with deliverables encompassing creation of a new university course on computer graphics at the University of Michigan, mentoring of undergraduate researchers, and summer camps introducing high school students to artificial intelligence. The research findings will be applied to cancer radiation therapy, enabling real-time tumor movement prediction to improve treatment precision while minimizing damage to healthy tissue. This work advances U.S. scientific leadership in AI-driven simulation and contributes to enhanced national health outcomes through safer medical treatment protocols.