Blindfolded Flight: A Novel Approach for Secure Drone Flight

This study proposes a positioning system that allows a drone to determine its location independently of external data sources. The system is one of the key factors in achieving blindfolded flight, which is the ability to follow a predetermined flight route using only the aircraft’s internal data, even in the presence of spoofing, interference, or unavailability of any positioning systems. The system remains operational even when the sensor signals are of low quality. The GPS spoofing detection method based on this system effectively manages slow drift attacks, and its stability is demonstrated by the small coefficients of variation (CVs) of detection delays. The prediction of the current position is based on deep learning rather than motion equations. Existing approaches to blindfolded flight are constrained by cost, flight time, or flight distance. The proposed system does not significantly suffer from these limitations. The flight time limitations of the latest studies are approximately 100 s. The system demonstrates no tendency for prediction error to increase over time during the 360-second flight, regardless of the presence of GPS spoofing attacks.