Federal Project Grant Award Summary

The University of North Carolina at Chapel Hill received a $328,476 CAREER award from the National Science Foundation's Division of Computing and Communication Foundations (CFDA 47.070: Computer and Information Science and Engineering program) beginning June 1, 2026, and extending through May 31, 2031. This project grant funds research to develop advanced computer vision systems capable of inferring physical properties of objects from images and videos through inverse rendering and physics-based simulation enhanced with generative priors. The deliverables include a generalizable machine perception framework that jointly infers three-dimensional object shape, material properties, physical parameters, and external physical entities from sparse visual inputs by combining physics-based modeling with statistical priors from generative models.

Beyond core research outputs, the project supports significant educational and workforce development objectives through curriculum development in physical understanding using artificial intelligence, provision of undergraduate research experiences, and targeted outreach to high school students and individuals from non-computing backgrounds. The research has direct applications to healthcare robotics for minimally invasive surgical systems and manufacturing automation for handling delicate or flexible materials. By addressing the technical challenge of resolving ambiguity in estimating multiple spatial and physical properties from limited video data, this CAREER project advances fundamental capabilities in machine perception while simultaneously building human capital in the computing and information science engineering workforce.