Federal Grant Award Summary

Award Details: Massachusetts Institute of Technology (MIT) received a $360,000 CAREER Project Grant from the National Science Foundation (NSF) Division of Computing and Communication Foundations under the Computer and Information Science and Engineering (CISE) program (CFDA 47.070). The five-year award, effective July 1, 2026 through June 30, 2031, supports research and educational initiatives in self-supervised video representation learning for machine perception.

Products and Services: The grant funds development of a self-supervised learning framework that extracts action-relevant visual representations from raw video and sensor data without requiring dense human annotation. The research delivers advances in machine perception capabilities by creating unified perception, prediction, and planning systems that enable artificial intelligence (AI) systems to understand environmental dynamics, anticipate action consequences, and support decision-making in physical environments. Key deliverables include novel learning methodologies and architectures supporting long-context prediction and action-conditioned world modeling; compact scene representations applicable to 3D reconstruction, motion estimation, and segmentation tasks; and open educational materials and mentorship programs training students in vision, robotics, and machine learning. These outputs aim to advance embodied intelligence systems with applications in robotics, scientific discovery, and other domains requiring reliable machine perception. No subawards are planned for this project.