Infrastructure Systems: Digital Storytelling
Led by Jason O. Hallstrom, Ph.D.
This project again builds upon the West Palm Beach testbed (see above), with a focus on digital storytelling to support urban design and community engagement. Human-centered urban design begins by establishing an understanding of how people interact with public spaces. The approach often relies on volunteers to collect information about the streetscape (e.g., pedestrian activity, mobility routes, social interactions, dog walking), typically via a mobile app. In the City of West Palm Beach, this occurs on a semi-regular schedule, and the results are used to inform streetscape planning and design activities. The process relies on web dashboards to 鈥渢ell the story鈥 of the streetscape. The planned project begins with the observation that these analytics dashboards do not tell a story in a format that resonates with most residents; residents are simply not versed in the diagrammatic notations, charts, or figures. This project seeks to explore the application of generative AI tools to create public-facing textual and graphical narratives that summarize the streetscape interactions captured by the West Palm Beach testbed. The intellectual merit of this project derives from (1) the exploration of applying generative AI tools to synthesize human-centered textual and graphical narratives based on both historical and near-real-time streetscape data; and (2) the design of an efficient system pipeline to support these facilities. The broader impacts of this project are again anticipated to be significant, providing new mechanisms for connecting communities to their streetscapes and better engaging communities in the art and science of placemaking.
Under the leadership of Dr. Hallstrom, the project will provide a research experience for up to two REU participants. The first participant will focus on the application of large language models (e.g., ChatGPT, LLama) to synthesize narrative summaries of streetscape activities based on sensed data. The second participant will explore the application of diffusion models (e.g., Sora) to synthesize corresponding visual summaries.