2021 – Present
This project addresses the critical need for decision-support systems that effectively use multimodal sensors to help decision-makers identify threats, predict crowd behavior, and suggest strategies to influence and guide crowds to safety. The primary objective is to develop a new class of real-time decision-support systems to advise decision-makers on effective mitigation strategies and likely outcomes. These systems will help enable threat detection in dynamic situations, allowing decision-makers at several levels to explore what-if scenarios and select appropriate actions.
The project is developing theoretical and algorithmic foundations to support creating new real-time decision-support systems based on combinations of physical and data-driven models. The project will utilize publicly available databases, conduct surveys, and develop scenario simulations to collect behavioral data for civilian responses to threat cues at various soft targets, including schools, shopping malls, and public entertainment venues. Researchers will use this data to develop new algorithms for designing and operating layered surveillance systems that employ available remote sensors and the venue’s design features. The project will implement its findings by creating a real-time decision support system that provides decision-makers with tools to explore hypothetical scenarios, improve security best practices, and inform decisions to select appropriate actions for threat mitigation.
Existing decision support systems are very limited in their capabilities to assist decision-makers in managing distributed multimodal sensors, evaluating the information provided, or selecting appropriate strategies needed to mitigate the impact of intelligent threats. This project aims to develop new algorithms capable of adaptive sensor management and advanced crowd behavior prediction. This project seeks to optimize sensor networks to prioritize critical information, enabling earlier threat detection and mitigation.
In 2024, the project began data collection and algorithm development. Surveys on threat perceptions and responses were expanded to various venues, and virtual environment behavioral experiments were conducted to inform agent-based simulations. Additionally, methods for optimizing sensor control and validating crowd behavior models were advanced.
Researchers are collaborating with the Real-Time Video Surveillance for Threat Detection and Mitigation project for video analytics and multi-sensor platforms and the Dynamic Digital Twins for Secure and Smart Civic Space project to help validate findings and explore potential applications to the Virtual Sentry Framework.
