AdaptiveTraining

This project focuses on developing an adaptive pilot training system that adjusts in real time to the pilot’s mental state and emotional responses. Traditional training methods do not adapt to fluctuations in a pilot's emotions or mental engagement, leading to inefficiencies in learning and increased chances of errors. By integrating real-time assessments of physiological indicators—such as heart rate variability and electrodermal activity—the proposed system will dynamically adjust its feedback to suit the pilot’s current mental and emotional conditions.

Unlike conventional systems, this adaptive training platform will provide personalized guidance through multimodal feedback (visual, auditory, and textual cues), tailoring the complexity and nature of the information based on the pilot’s needs. For instance, during periods of high stress or frustration, the system will simplify the feedback, whereas it will introduce more detailed instructions when engagement is low.

Expert pilots from Northrop Grumman Corporation will collaborate on the design and refinement of adaptive rules, ensuring the system benefits both novice and experienced pilots. By leveraging large language models (LLMs) for generating real-time feedback, the goal of this system is to enhance learning efficiency and emotional resilience, creating a more effective and responsive training environment.

More details will be released after publication
Shaoyue"Jewelina" Wen
Shaoyue"Jewelina" Wen
文劭玥

My research interest lies in understanding, predicting and adapting user behaviors.