People Forecasting: Research Project Description

My thesis work began as part of project Rescue (http://www.itr-rescue.org) whose goal is to help emergency personnel better respond to natural and man-made disasters. I titled my Rescue subproject “People Forecasting”. Just as a weather forecast helps people make more informed decisions about what to wear, where to go and when to go there; the goal of People Forecasting is that emergency personnel would use our models to discover information about population density and movement city-wide at any instant in time, resulting in increased speed and improved quality of their decisions regarding response efforts during a disaster. Human activity sensors, such as pictured in Figure 1, are becoming more common and are forming an increasingly dense network, especially in urban areas. They provide realtime information about the presence and movement of people that could be valuable in emergency response, transportation planning, and security applications. The aim of my research project is to extract useful information about human activity and behavior that is hidden in these simple measurements. This project is particularly appealing to me because I believe our research has potential to be of great value to the emergency response community. The existing state-of-the-art methodology for occupancy estimation on a large-scale; for example, relies on census information (e.g, in transportation and urban planning studies), sometimes with a constant multiplier for daytime for industrial and residential areas. How-