Aura: An Automated System for the Real-Time Evaluation of Flight Maneuver Performance

The automated evaluation of flight maneuver performance in civilian and military aircraft enhances pilot proficiency and aircraft safety. In this article, we present Aura, an inaircraft flight maneuver training system that provides real-time performance feedback to pilots. Aura consists of sequential flight data capture, flight data analysis, and flight data visualization modules. The flight data capture module uses a pipeline of computer vision techniques and object detection algorithms to collect flight data optically. The flight data analysis module applies a transformer-based classifier network, trained on a custom dataset, to identify flight maneuvers in real-time. The flight data visualization module displays processed data to pilots on a hardware display in a task-specific layout. Aura does not need a physical interface with aircraft avionics, thereby circumventing data access issues in military aircraft. Aura is ground tested using UH-60M Black Hawk helicopter flight simulator recordings and is flight tested for several hours in a Cessna 172S G1000 to validate effectiveness.