SBIR Phase I: FlowStateVR

The broader/commercial impact of this Small Business Innovation Research (SBIR) Phase I project is to help address growing national pressures on the aviation ecosystem by developing a Virtual Reality (VR) training platform that personalizes training based on inferred pilot mental states (e.g., fatigue, stress, workload) to improve learning outcomes and reduce training costs. The United States faces substantial aviation workforce shortages, which impacts American companies and citizens every day. Current training systems cannot meet increasing demands due to reliance on costly flight simulators. This project operates within the intersection of applied cognitive neuroscience and aviation psychology and advances the scientific and technological understanding of human performance by developing new knowledge of pilot mental states and their influence on performance outcomes. This project offers a durable competitive advantage in commercial aviation training by developing new capabilities for VR, beyond what is possible using flight simulators and other VR devices. This project will be developed into Software as a Service, and Data as a Service business models giving commercial airlines new levels of understanding of their pilots. By year three, measurable outcomes include thousands of pilots trained, faster time-to-competency, improved performance metrics, and higher training throughput across the ecosystem. This Small Business Innovation Research (SBIR) Phase I project will develop and validate an adaptive virtual reality training system that: 1) integrates multimodal sensing and interpretable machine learning to infer cognitive microstates during aviation training tasks, and 2) adapts task demands to improve training experience and learning outcomes. The project addresses a core technological and engineering challenge in distinguishing multiple cognitive microstates that share overlapping data signatures across eye tracking, autonomic, and behavioral sensor modalities making them difficult to disentangle with existing multimodal or low-resolution systems. Objectives include: 1) multimodal feature extraction and fusion pipeline for an existing VR pilot training platform, 2) interpretable machine learning classifiers capable of microstate discrimination under sensor noise, and contextual ambiguity, and 3) develop a prototype pilot training flight scenario with varying difficulty levels that can be evaluated by aviation subject matter experts. Methodologies include feature engineering, ensemble methods, supervised machine learning, and dimension reduction approaches. Anticipated results include reliability and validity testing of microstate-relevant features and models, initial classifier performance benchmarks, and a functional VR prototype demonstrating stable real-time adaptation. These outcomes establish the scientific and technical foundation required to deploy a scalable, neuroadaptive simulation training device in operational aviation environments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.