2024 – Present
This project seeks to understand how architectural design affects human behavior during emergencies, particularly during active shooter incidents. Using the information realized from analyzing human behavior in emergencies specifically related to architectural design elements, this research will address knowledge gaps in current crowd simulation tools, which fail to capture the complexity of human behavior in these critical situations. The primary objective is to assess the influence of architectural design on emergency behavior during active shooter incidents, ultimately creating a decision-support tool for improving safety practices, policy guidelines, and architectural standards across diverse building types.
The project utilizes a combination of data-driven modeling, reinforcement learning, and agent-based simulations. It leverages data from previous research involving human behavior during emergencies to create mathematical models that inform crowd behavior in simulations. The proposed crowd simulation model will demonstrate the dynamic behavior of an active shooter, validated against real-world law enforcement reports to ensure accuracy. The project intends to integrate the decision-support tool it generates into existing DHS operations and systems to apply its findings in practical scenarios.
This project improves upon current research by employing advanced clustering analysis to identify behavior patterns among participants in active shooter scenarios, integrating a comprehensive spectrum of personal, social, and environmental variables. This approach offers a more nuanced understanding of crowd behavior compared to current analysis done using demographic segmentation.
In 2024, the project conducted preliminary clustering work on human behaviors, enriching datasets through video analysis, and developing reinforcement learning models that simulate active shooter incidents within office environments. It also explored building exit parameters and worked to develop representative models.
Researchers collaborate with various SENTRY projects in the Layered Security Architectural Design and Simulation Research Area, and will work across all four SENTRY research areas to support the center’s vision of the Virtual Sentry Framework. These collaborative efforts are crucial for enhancing the project’s success by providing access to resources, validating hypotheses through expert feedback, and ensuring the practical relevance of the research outputs.
