Metron proposes to develop a software service that monitors airspace message streams, makes predictions about flight trajectories, identifies anomalous behaviors, and generates alerts when certain risky or anomalous events occur. Our key innovations include novel neural network architectures and training methods designed to learn relevant flight behaviors and to detect anomalous deviations from normal behaviors.
We will address the technical questions of how to build a system that monitors the NAS continuously and automatically identifies flights that indicate safety or efficiency issues or precursors to such, while reducing the false alarm rate. Such a system will make predictions “in time” for ATC and pilots to take corrective action to minimize the effect of such events. To build this system, Metron will redesign and extend the Phase I neural network model and package it as a software service that makes predictions of important flight events.
Metron has teamed with ATAC who will provide subject matter expertise to visualize and assess the operational utility of the events identified by the system.
Our flight anomaly detection and risk prediction software will provide a key capability for NASA’s In-time System-wide Safety Assurance (ISSA) initiative. Integration into NASA’s Traffic Data Manager (TDM) will help pilots manage the increasingly congested and complex airspace by supplying factors used to determine the relevance of nearby aircraft. Additional integration into air traffic control systems via the consumption of SWIM data feed messages will provide controllers with operationally relevant alerts.
Terminal ATC users can be alerted both to flights exhibiting anomalous behavior and to predictions of various safety or efficiency related outcomes. We identified and plan to integrate into a specific overseas air traffic control system to improve the efficiency of airspace operations.