Federal Project Grant Award Summary

The National Science Foundation's Division of Computing and Communication Foundations awarded a Project Grant of $343,839 to the University of North Carolina at Chapel Hill effective August 1, 2026, through July 31, 2031, under the Computer and Information Science and Engineering (CISE) program (CFDA 47.070). This CAREER award supports the development of a unified framework for video perception, reasoning, and control grounded in natural language supervision. The project will create next-generation artificial intelligence systems capable of understanding complex, multi-step activities in video by leveraging narrated instructional video to develop structured, multi-level video representations. The system will model actions as learnable transformations applied to object representations, enabling recognition of novel action combinations and improved reasoning about causes and consequences of observed events.

Deliverables include development of AI methodologies applicable to robotics, healthcare, and manufacturing sectors where understanding dynamic visual scenes is essential. All software, trained models, and datasets produced will be released publicly to accelerate scientific progress and support responsible AI development. The award integrates substantial educational components, including development of new university courses, mentorship of undergraduate and high school students, and public outreach programs connecting artificial intelligence to accessible topics such as sports and skill learning. Performance will be conducted at the University of North Carolina at Chapel Hill, Chapel Hill, NC 27599.