Modeling human mobility
Modeling human mobility
The Location Intelligence Research Group at ORNL advances geospatial science to address challenges at the intersection of national, human, and energy security. The group’s scientific portfolio pursues inventive work in the areas of Human Mobility Science, Place-based Characterization, and Semantic Ontologies with Multimodal Data Conflation.
A team revealing global patterns of life
A team revealing global patterns of life
Location Intelligence researchers specialize in complex adaptive systems, human mobility modeling, geospatial simulation, transportation geography, spatial analysis, and more. Their impact spans federal agencies by modeling cross-border movement patterns using telematics and communications data and forecasting potential fuel shortages in the event of future natural or human-made disasters.
Focus areas
Human Mobility Science
Human Mobility Science
Understanding how and why people move is crucial for anticipating security threats, optimizing operational planning, and improving disaster response. The group employs movement-based patterns of life, trajectory analytics, and AI-driven pattern detection to analyze mobility patterns from sources such as transportation data, social dynamics, and communication networks. These insights inform border security, critical supply chain monitoring, and emergency response for a proactive intervention during crises.
Place-based Characterization
Place-based Characterization
Place-based characterization involves analyzing the unique attributes and functions of specific locations to understand human activities and land use patterns. This location intelligence approach integrates data from various sources, such as remote sensing imagery, social media, and geospatial information, to create detailed profiles of areas. By examining factors like spatial distribution, semantic meaning, and temporal dynamics, place-based characterization supports applications in urban planning, disaster response, and national security by providing insights into how places are utilized and how they evolve over time.
Knowledge Graph and Semantic Ontologies
Knowledge Graph and Semantic Ontologies
In geospatial intelligence, pulling useful insight from scattered and complex data is a constant challenge. Location Intelligence group focuses on using knowledge graphs and semantic ontologies to connect the dots, literally and figuratively, across diverse sources like satellite imagery, sensor feeds, and open data. We help decision-makers get a clearer, more complete picture by organizing this information in a way that machines can understand and reason over. Whether it’s for crisis response or monitoring global events, our work makes it easier to spot patterns, surface critical relationships, and support fast, informed action.
Projects
MapSpace: Multi-scale Global Land Use Modeling
MapSpace develops scalable models of global land use using Point of Interest (POI) data. By combining spatial distribution, semantic meaning, and temporal dynamics of POIs, this framework automatically characterizes land use across different spatial scales. Applications include disaster response, population modeling, and urban planning. The system integrates NLP, spatial network analysis, and deep learning to tailor models to geographic context.
PlanetSense
PlanetSense is a platform that harnesses both archived and real-time open-source data—such as social media, location services, and events—to generate geospatial intelligence on the fly. It integrates sophisticated data analytics algorithms and visualization tools, enabling real-time insights for applications like disaster response, critical infrastructure resilience, and land-use mapping.
SONET: Semantic Ontological Network for POI Categories
SONET addresses the challenge of data heterogeneity in crowd-sourced POI data by building a comprehensive semantic knowledge graph. The framework uses OpenStreetMap tags as a semantic bridge, organizing over 12,500 POI categories into a three-level hierarchy. It's foundational for rapid data fusion in emergency response, population modeling, and geospatial intelligence.
HumoNet
HumoNet is a comprehensive framework designed to model and simulate human mobility networks with high realism. By integrating various data sources and employing advanced simulation techniques, HumoNet enables the analysis of human movement patterns, which is crucial for applications in urban planning, transportation optimization, and emergency response strategies. The framework's ability to accurately represent mobility networks supports the development of more efficient and resilient infrastructures, directly contributing to national security and public safety initiatives.
TraKit
This project represents a novel initiative aimed at the extensive and scalable evaluation of human mobility datasets. TraKit is a comprehensive software and algorithmic library that systematically characterizes spatiotemporal human movement patterns using a broad range of scientifically rigorous metrics. It is built to serve diverse domains and missions by providing high-fidelity insights into five core categories of human behavior: Scale, Realism, Attribution, Patterns of Life (PoL), and Sociocultural Assessments
CORSAIR
The CORSAIR framework introduces a novel science capability that advances human mobility modeling by characterizing individual-place relationships and predicting transitions based on nuanced visit behavior. Unlike traditional models that treat all location visits uniformly, CORSAIR categorizes visits into types—such as casual, routine, anchor, and resettling—based on frequency and dwell time, providing a richer understanding of behavioral intent.
Contact
Gautam Malviya Thakur, Group Leader for Location Intelligence
See Gautam's bio
Gautam Malviya Thakur, Group Leader for Location Intelligence