Project Grant 2617314

The National Science Foundation Division of Computing and Communication Foundations awarded $300,000 to The Research Foundation For The State University Of New York (operating as SUNY at Binghamton) on September 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to advance large language model-based spatial-temporal forecasting for urban dynamics prediction.

The project establishes theoretical foundations, robustness guarantees, and interpretable reasoning capabilities for LLM-based models that forecast urban conditions by processing multi-sensor observations such as traffic speeds and travel demand. Current approaches lack explanation of when and why predictions succeed or fail, perform poorly under distributional shifts such as sensor failures and disruptive events, and provide no transparent reasoning identifying spatial regions, temporal contexts, and external events driving forecasts. The work unifies theory, learning, reasoning, and verification within a single framework to develop trustworthy, constraint-aware, interactive spatial-temporal foundation models for urban intelligence. The project integrates research with education and workforce development through curriculum innovation, student mentoring, and outreach activities.

Performance occurs in Binghamton, New York, with a period of performance from September 1, 2026, through August 31, 2029. The award is structured as a project grant, with $300,000 obligated at the time of award.