Abstract
Advancements in virtual and live training technologies are providing new opportunities to integrate multimodal data—video, audio, sensor, geographical positioning, behavioral ratings —into training evaluations to enable data-driven insights. Using multi-modal data to assess team performance and inform learning analytics is a critical dependency for providing squads and Soldiers with timely feedback and insights that aid in improving performance, which is a key focus of the U.S. Army’s training modernization program. Achieving this end-state requires a robust data infrastructure to reliably capture, synchronize, and interpret multimodal sources, an evidence-centered framework for converting raw data into actionable insights that can support performance assessment, and functions to support data visualizations and performance modeling across time. While multimodal data promises unprecedented insights into squad performance, harnessing its full potential presents significant challenges. This paper discusses how we have systematically extended the Synthetic Training Environment Experiential Learning for Readiness (STEEL-R) architecture to interoperate with a state-of-the-art mesh network to capture live instrumented training data and align to a squad competency framework in near real-time. By combining diverse data sources−motion capture sensors, movement data, orientation, GPS, communication logs, and fire effectiveness−STEEL-R aims to provide faster and more accurately aligned insights into the factors and processes impacting small unit performance. The infrastructure also incorporates Observer-Controllers/Trainer ratings to ensure assessments are data-guided and inform contextual understanding. We outline the architectural requirements for accommodating multimodal data streams, discuss an assessment model we developed that turns raw data into insights for competency modeling, and describe enhancements to STEEL-R's data visualization capabilities to enrich after-action review experiences. Finally, we highlight lessons learned from applying STEEL-R to assess squad performance during a field training exercise and highlight how extending its architecture can significantly contribute to enhanced training analytics to improve squad readiness.