Flight Profile Effects on Factor Graph Magnetic Calibration Accuracy

Magnetic anomaly navigation has recently received significant research attention and shows the potential to provide alternatives to land and space-based signals navigation methods. Imperfections in the sensors as well as magnetic fields created by the vehicle the sensors are installed in create errors in the magnetic measurements. Magnetic anomalies are normally no larger than 500 nanotesla, whereas flight tests have shown measurement errors approaching 5,000 nanotesla when constraints prevent the use of booms commonly used on survey aircraft. With errors approaching an order of magnitude larger than the signal of interest, an effective compensation method is required for magnetic anomaly navigation. The Tolles-Lawson calibration procedure has been the standard method to compensate magnetic measurements from errors since its development in the 1940’s. Enhancements have been made; however, the fundamental procedure has remained: a series of maneuvers are flown with specific pitch, roll, and yaw rates to create errors with specific frequency content due to the aircraft orientation in Earth’s magnetic field. The resultant measurements are band pass filtered with the assumption that the resulting signal is a direct result of the installation errors which are then used to compensate the magnetic measurements. We’ve proposed a new method of compensation utilizing a factor graph estimation algorithm to determine magnetic measurement errors. Our initial work on factor graph calibrations utilized Tolles-Lawson maneuvers to compare calibration effectiveness of the two methods, however the factor graph method does not rely on the same assumptions as Tolles-Lawson calibration methods and therefore it is not constrained to utilize the same calibration profiles. This paper presents an analysis of the effect flight profiles have on magnetic sensor calibration accuracy utilizing a factor graph calibration method. While the typical Tolles-Lawson calibration profiles have proven effective with our factor graph calibration, they are not strictly required, and it would be advantageous in real world applications to reduce both the time and physical space required to perform sensor calibration when not constrained by flying a Tolles-Lawson calibration profile prior to every mission. Our analysis provides a baseline that informs the necessary trade space between an optimum calibration profile and airspace or mission constraints.