Decision Making Tools to be Developed for USAF Threat Response

Charles River Analytics is set to develop integrated tools to support decision-making related to installation security and risk assessment for the U.S. Air Force. 

This comes after the company received Phase II Small Business Innovation Research (SBIR) funding from the Air Force Civil Engineer Center (AFCEC). 

The resultant system, Real-Time Assessment and Decision Interactive Toolkit (READI), delivers relevant information tailored to the needs of each user in the security hierarchy. 

READI uses probabilistic programming and deep learning techniques, along with ecological situation awareness visualizations, to provide insights into securing installations for mission execution. Using hybrid artificial intelligence (AI) techniques, READI characterizes potential threats and provides real-time risk assessments.

READI complements risk assessment plans already in place. It draws on information from disparate sources then paints a more comprehensive picture of risk, while factoring in critical elements to support situational awareness and threat response decision-making.

Because noise degrades much of real-time data, READI separates the signal from the noise. “It’s helping a security officer understand what should be acted on and in what order to prioritize response activities,” stated Dr. Nicolette McGeorge, Human Factors Engineering Scientist and Principal Investigator on the READI project. 

The result can then update the risk assessment as well, leading to a more robust model. Charles River Analytics assert that probabilistic modeling efficiently handles dynamically noisy and uncertain environments, facilitating READI to provide rapid, explainable updates.

The foundation of READI is a deep study of the work, and cognitive support needs to understand how users make decisions when interacting with complex systems. As a result, the tool is able to provide easily visualized contextual information.

Phase I of READI involved characterizing threats in the environment, creating probabilistic models of risk assessment and course of action using Charles River’s probabilistic programming framework, Scruff™, and then creating visualizations to support decision-making for security personnel.

Potentially disparate pieces of information, when viewed together in context, can lead to critical insights, McGeorge explains. By itself, an individual piece of information need not signify a threat, but when stitched together with others, it might. Risk assessments change accordingly, as do recommended courses of action.

The company state that a similar approach will likely apply to the evaluation of natural disasters, while factoring in environmental data, and is expected to form the basis of the next phase of the READI project. Phase II will focus on a wider range of threats caused by natural disasters such as hurricanes, floods, and fires. Under a related NASA-funded effort called WIMPLE, Charles River has created a hybrid AI-based decision support tool for wildfire risk assessment and mitigation.

Updating risk assessments according to new and emerging threats, as READI does, helps security and emergency management teams respond faster and more efficiently. “There’s an opportunity to catch problems sooner than you might have otherwise,” McGeorge added.

Charles River Analytics report that commercial applications could extend to use by emergency management organizations of municipalities, and even for security management of large events.