Fine Tuned Biometric Models for Warfighters in Austere Settings
Abstract
a.Technical description: High workloads, fatigue, and stress can negatively impact decision-making and peak physical performance. To enhance endurance and resilience in military personnel, it is essential to assess their physiological readiness in operational andtraining contexts. Effective monitoring of physiological state allows for individualized training, ensuring appropriate intensity levels through real-time adjustments. Sensors can facilitate this monitoring, increasing awareness of mental states. However, many existing software algorithms lack sophistication, resulting in inflexible and imprecise feedback. There is a pressing need for more accurate algorithms that provide personalized adjustments to optimize training outcomes and enhance decision-making capabilities.Developing these algorithms requires comprehensive datasets that account for individual responses to conditions like high workload, fatigue, stress, and diverse environmental factors such as elevated temperatures. Incorporating data from diverse backgrounds and cultures is crucial, as these variables significantly influence physiological responses. Collecting diverse data is vital for creating unbiased algorithms. University of South Australia#s (UNISA) Behaviour-Brain-Body Research Centre have unique expertise and facilities to collect these datasets. Fortifyedge has developed innovative algorithms tailored for the next generation of wearable and sensor technologies expected to be released within 12-24 months. Advances in hardware will enable on-device intelligence through dedicated micro-Neural Processing Units, allowing real-time deep learning inference with techniques such as Generative AI, Transformers, convolutional neural networks (CNN), and recurrent neural networks (RNN), all optimized for ultra-low power consumption. Fortifyedge#s capabilities represent a significant advancement in performance and efficiency for context-aware intelligence on devices and sensors, allowing fine-tuning to the unique characteristics of individual users and their variable environments. This integration of human and environmental context will support more individualized algorithms, fostering trust in human-machine collaboration. This project willleverage expertise at UNISA to assist Fortifyedge in refining and validating their novel algorithms by establishing a reliable ground truth dataset utilizing resources and the unique laboratory at the UNISA#s Behaviour-Brain-Body Research Centre.b.Specify relevance to ONR:This project aligns closely with ONR s research priorities of advancing human performance, training, and education throughresearch into decision-making, expertise development, warrior resilience, and augmented warfighter operations. Furthermore, the research addresses stress response by investigating individual variations in stress reactions, circadian rhythm impacts in 24/7 operations, and biomarkers distinguishing stress resilience from vulnerability. The project also resonates with DARPA s recent Biological Technologies Office initiative, focusing on enhancing team training assessment through bio-behavioral signals. This initiative aims to establish objective metrics for predicting team effectiveness, crucial for future defense training requirements. The project s application scenarios encompass USMC FiST Training, US Air Force air battle management (ABM) team training, and U.S. Navy submarine navigation (SPAN) team training.c.US Collaborators: Army Futures Command Indo-Pacific, Army Combat Capabilities Development Command - Soldier Center.d.Outcomes: 1.Establish ground truth data sets: Collect data at UniSA to create a novel, bias-free dataset on a range of individual and team tasks during dynamic simulated environment conditions.2.Fine tune multi-modal deep learning algorithms that Fortifyedge have developed so they are individualized and personalised using the ground truth dataset.3.Assess effectiveness of Machine Learning Operation
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Nov 09, 2024
- Source ID
- N629092412111
Entities
People
- Camilla Liddy
Organizations
- Office of Naval Research
- United States Navy
- University of South Australia