Innovating Beyond Boundaries to Advance the Chem-Bio Defense Mission

Demo: CBRN Extended Reality (XR) Training

Noblis Team: Dr. Lance McLean, Jack Daus, Stephanie Andino, Quang Vo, Barry Ross, Jared Haas, Dr. Amanda Fond

Our CBRN XR training application offers a safe-to-fail environment while still providing realism. The scenario shown is a radiation search mission, but this technology can be applied to the entire CBRN field. The app includes features like networked multiplayer (multiple trainees and a trainer), radiation search (detector, place markers, dose stay times), trainer desktop view and after-action reporting.

Demo: 3D Mapping

Noblis Team: Jack Daus, Dr. Lance McLean, Dr. Amanda Fond, Bay Wiggins

Noblis is developing a prototype application that enables users to automatically create digital 3D scene reconstructions using state-of-the-art machine learning (ML) algorithms. Users capture photos of a scene on their phone and upload the images to the cloud where they are automatically converted into a high-quality 3D model. Using a new technique called 3D Gaussian Splatting, users can view a digital replica of the scene in as little as 30 minutes directly from a web browser. Our research team is continuing to explore extended applications of this technology, including immersive virtual reality (VR), automatic 3D object segmentation and embedded annotations.

Predictive Surveillance of Infectious Disease Outbreaks (PSIDO)

Noblis Team: T.C. Folkedal, Dr. Katharine Jennings, Blaise Arbogast, Dr. Fanwei Zeng, Lauren Moore, Dr. David Ashford, Dr. Loren Shaffer, Dr. Anna Swiatecka

Noblis has developed a model to predict the risk of mosquito blooms correlated with infectious disease outbreaks to improve defense mission readiness and reduce morbidity and mortality. This project aimed to develop and demonstrate a model that accounts for often overlooked weather-related variables associated with arbovirus infectious disease outbreaks. The model provides risk prediction for mosquito blooms likely to coincide with weather sufficiently far enough in advance to enable better preparations and prescriptive actions to mitigate the risks.

Computer Vision-Enabled Point-of-Care Biothreat Detection

Noblis Team: Dr. Bradley Abramson, Dr. Daniel Negron, Dr. Katharine Jennings, Bryan Necciai, Shanmuga Sozhamannan, Dr. Kelsey Hauser

Fast biothreat detection in resource-constrained environments is essential for fast countermeasure deployment to give warfighters unprecedented advantages. Blindspot Chips and loop-mediated isothermal amplification (LAMP) assays are portable, low-cost solutions for rapid detection of multiple diverse targets simultaneously. These tests, however, are still prone to subjective interpretation, lack clear guidance for next steps and require costly, vendor-specific hardware. Learn how we overcame these challenges and developed an application to analyze solutions using computer vision for in-field interpretation – ultimately enabling faster, easier and more accessible biothreat detection at the point-of-care for our warfighters.

Applying ML and Statistical Models to Low-Cost Aerosol Sensors for Anomaly Detection

Noblis Team: Dr. Justin Taylor, Brice Ballesteros, Dr. John Helmsen, Dr. Sean Kinahan, Dr. Oscar Olmedo, Cody Rutherford, Ahmid Said, Nathan Spivy, Riley White

The U.S. is investing in multiple programs to improve monitoring for airborne releases of chemical or biological agents using sensors and anomaly-detection data analytics. But common environmental material can emit a signal that increases false alarm rates. Recent developments in ML techniques offer the ability to move from traditional classification of detection events to an approach fusing data from multiple sensors and sources. Using data from the PurpleAir sensor network, Noblis researchers developed synthetic data and overlaid that data with PurpleAir data. Our results show that, overall, smarter data fusion could play a larger role in the spectrum of biological incident detection and response.