Protecting soft targets and crowded spaces requires agile surveillance systems with a diverse suite of sensors, capable of detecting emerging threats by noticing contextual anomalies. These sensors must be managed dynamically to identify the threat, focusing multimodal sensing on the areas where the threat is emerging. Once the threat is localized and identified, these systems should support the selection of appropriate mitigation actions that account for the psychology of the humans involved (would an evacuation order lead to panic that results in casualties? Would it be ignored?) and the specific architectural features of the venue (e.g., rapid exit options, safe areas.) The tasks of creating and operating these systems are complicated by the diversity of threats, the potential rapid tempo of future attacks involving autonomous vehicles, and the complexity of a multi-modal network of sensors needed to provide observability of the STCP space. The goals are to develop decision support systems that assist decision-makers in the deployment and control of multi-modal, surveillance systems and attack mitigation measures that optimally combine the capabilities of available sensors, that are tailored to architectural features of the crowd space environment and that account for human behavior in that environment. These decision support systems must support threat detection in dynamic, data-deluged situations, as well as advise on mitigation strategies and likely outcomes, allowing decision makers at several levels to quickly explore what-if scenarios.
This project will collaborate with others in this research area to design experiments involving human actors at facilities such as the Guardian Centers campus testbed, which will be used to evaluate models developed for decision support systems for threat mitigation, as well as interact with the other research areas within SENTRY.
