The use of magnetometers to augment orientation measurements
Abstract
The vectors of the local geomagnetic field constitute a unique natural frame of reference at any point on earth. When using magnetometers for vehicular and aerial navigation, an accurate measurement of the Earth's magnetic field vector is required. Our test bed of six magnetometers; one placed on each side of a cube, can determine the orientation of the vehicle. We show that measurement of the geomagnetic field at sufficient resolution uniquely identifies a location on the earth at a particular time. We calculate the accuracy of localization and orientation as a function of sensor resolution. We find that the sensor accuracy needed for localization and orientation is not far from available technology. We show the circuitry of our magnetic measurement prototype and the requirements for extracting the available information of the geomagnetic field. Additional calibration is needed due to the fabrication process of the magnetometers, since numerous non-linearities arise in the magnetometer model. The calibration enables the determination of additional information such as relative rotation and relative positioning. The calibration is performed using neural networks and external training data from a GPS system. Neural networks have the advantage of being able to update the accurate parameters of the magnetometer model by learning, as well as the capability to reject additional noise. The relative translational movement can be determined when the magnetometers provide sufficient resolution. The relative rotation, however, does not require as fine a resolution of the magnetometers; we show with a simple grid search and bisection method that the Euler angles can be estimated. As an example, the bisection method is applied to an F-16 model and accurate measurements for the body-axis roll, pitch and yaw rates are obtained.
Degree
M.S.M.E.
Advisors
Ariyur, Purdue University.
Subject Area
Electrical engineering|Mechanical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.