Predictive Framework for In-Time Safety Monitoring

Our innovation offers NASA and the aviation community a robust, extensible data processing and analysis infrastructure that supports predicting the near-future status of terminal airspace to aid decision-making. The Robust Analysis for Predictive In-time Decision Support (RAPIDS) will generate short-term horizon forecasts (10-20 minutes) for the state of the airspace. RAPIDS combines known procedural structures of the airspace with historical operational data to predict future aircraft trajectories in the terminal airspace. The predicted trajectories will then be used to identify operational/safety stress in the airspace and support mitigation actions. Most of the prevalent research and tools for predictive safety monitoring focus on reactive or tactical predictions (several seconds to minutes ahead). As the forecast horizon increases, the uncertainty in the projected states increases. RAPIDS aims to reduce the uncertainty in the tactical and strategic levels by integrating machine learning algorithms trained on historic operations. Commercial aircraft operational patterns are repetitive due to the highly structured operational procedures used in terminal airspace, thus allowing machine learning models to extract the underlying operational patterns.