Calibration of an Autonomous Navigation System Using GPS and Machine Learning

GPS receivers are heavily used in aviation for both visual and instrument flights. There is a need for a backup system so that enroute and non-precision approaches can be flown without reliance on ground based systems should GPS be unavailable. We propose to use barometric pressure, airspeed, low-cost inertial sensors, three-axis magnetometer along with high-fidelity geomagnetic models, and an optical ground speed sensor to provide backup position and navigation solutions during GPS outages and improve the GPS solution when the signal is available. The optical ground speed sensor would be used to estimate the wind speed and direction, as this can be a significant source of error when considering inertial sensors alone. Should the ground be obscured, wind estimates can be provided externally by the pilot or via weather services such as Arnav¿s WxLink. The widespread use and acceptance of GPS with a backup system will promote the reduction of operating costs of the National Airspace System (NAS) through the shutdown of many ground based navigation systems. The integration of terrain information into such a system will also be considered to minimize the occurrences of Controlled Flight Into Terrain (CFIT)