Building evaluation for human-centered flood management: integrating foundation models and value-based methods

Virginia Polytechnic Institute & State University (Virginia Tech) received a $550,000 project grant from the National Science Foundation (NSF) Division of Civil, Mechanical, and Manufacturing Innovation under the Engineering program (CFDA 47.041), awarded June 15, 2025, with completion scheduled for May 31, 2028. This research initiative delivers integrated evaluation methodologies and digital tools designed to improve building design and operational resilience during flood events, with particular focus on individuals with mobility challenges. The primary deliverables include automated feature extraction and analysis systems leveraging foundation models—large deep learning models trained on extensive datasets of images and text—to identify critical building design characteristics. These technological outputs are validated against need hierarchies derived from a curated database of lived experiences documented during past flood events, ensuring that building design recommendations directly address the practical evacuation and accessibility barriers faced by people who rely on mobility aids.

Beyond technological outputs, the project delivers educational and knowledge-advancement products, including new interdisciplinary course modules integrating engineering, humanities, and design perspectives on disaster resilience. The research generates findings and guidance applicable across multiple disaster types, built environment scales, and population groups, contributing to the foundation of human-centered building design standards. Virginia Tech is planning subawards as part of this project's execution, leveraging collaborative institutional expertise to execute the comprehensive research agenda addressing the critical gap between current building evaluation practices and the needs of vulnerable populations during disasters.

Abstract

The objective of this project is to support research that looks to examine building design and operation during disasters, focusing on people who have navigational challenges that may adversely affect their evacuation. IT provides new insights into building features from the perspective of those who rely on mobility aids. The outcomes potentially lead to responsible, resilient buildings and contribute to knowledge at the interaction of disasters, built environment, and populations. Additionally, this project promotes interdisciplinary education in engineering, humanities, and design through new course modules and outreach initiatives. Building designers and operators often overlook unique obstacles posed during disasters due to a lack of evaluation methods that can integrate architectural design principles, the needs of people with mobility support requirements, and disaster resilience goals. This GAP is further rooted in the absence of voices from people who use mobility aids and efficient mechanisms to incorporate their needs into digital platforms. This project, by leveraging the power of foundation models, which are large deep learning models trained on extensive datasets of images and text, looks to automate extraction and analysis of key building features. Furthermore, value of the developed models will be evaluated by assessing how well such features align with need hierarchies derived from a curated database of lived experiences during past flood events. The findings are expected to inform design decisions across disaster types, built environment scales, and population groups. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are planned for this award.