National Science Foundation Engineering Research Center GeoRealism GeoRealism Expanding the human ability to comprehend a larger geo space Cyrus Shahabi University of Southern California Los Angeles, CA
National Science Foundation Engineering Research Center What is solved? Virtual Worlds Earth visualization (EV) platforms, Microsoft Virtual Earth & Google Earth, display georealistic 3D worlds created from aerial imagery. Geographic Information Systems (GIS) offer query and analysis of static worlds. Games engage users in interactive worlds based on fictional data. Yet no one system combines interactivity, realism, and query & analysis for both static and dynamic world data in a unified system to create what we call GeoRealism
National Science Foundation Engineering Research Center What is missing? Vision: GeoRealism GeoRealism is a new computing paradigm that gives humans the ability to capture, model, integrate, and query static and dynamic data from the real world in real time at fidelities not currently available A GeoReal world is an ultra-high resolution virtual world that is geographically accurate, enables efficient querying and analyses of dynamic data, and supports interactivity from a variety of interface modalities.
National Science Foundation Engineering Research Center High risk and needed topics Mismatch between the available geo-data and the information humans require to successfully comprehend a larger geo-space is the main impediment to realizing the GeoRealism paradigm. Data mismatch problems -- GeoRealism requires data be available: 1.At the right space, time, resolution and quality –Freeways vs. schools; fire; microclimate; GPS –Actuation; human sensors; web sensing 2.At the right abstraction level –Point clouds vs. 3d models; video vs. events; trajectories vs. patterns 3.In an integrated fashion –Vector with imagery; text (news) with maps; 4.As fast as we need it –Not just rendering data structures but search and access (on land surface, road, for dynamic data)
National Science Foundation Engineering Research Center More specifics (w/ bias towards my own research :) What is almost solved? –Spatial queries in constrained spaces Road-Network (VLDB’04; SIGMOD’08, Samet et. al) Land surface (VLDB’08) –Variations of spatial queries Spatial skyline (VLDB’06) The optimal sequenced route query (VLDB-J’08) –Private spatial queries Cloaking approaches (VLDB’06, Mokbel et. al) PIR-based approaches (SIGMOD’08) What has failed? –Integration of spatial queries & data structures into commercial products, e.g., Google maps, Yahoo! maps –An intuitive general-purpose GUI for complex geospatial and temporal queries