Adviser: James W. CutlerWe present a method for on-orbit, attitude-independent magnetometer calibration that includes the effect of time-varying bias due to electronics on-board a spacecraft. The calibration estimates magnetometer scale factors, mis-alignments, and constant as well as time-varying bias. Time-varying effects are mitigated by including spacecraft telemetry in the measurement model and estimating constant parameters that map the telemetry data to magnetometer bias. The calibration is demonstrated by application to flight data from the Radio Aurora Explorer satellite and significantly reduces the uncertainty of off-the-shelf magnetometers embedded within the satellite and subject to spacecraftgenerated fields. This method simplifies the satellite design process by reducing the need for booms and strict magnetic cleanliness requirements.

No abstract

An online attitude determination filter is developed for a nano satellite that has no onboard attitude sensors or gyros. Specifically, the attitude of NASA Ames Research Center's O/OREOS, a passively magnetically stabilized 3U CubeSat, is determined using only an estimate of the solar vector obtained from solar panel currents. The filter is based upon the existing multiplicative extended Kalman filter (MEKF) but instead of relying on gyros to drive the motion model, the filter instead incorporates a model of the spacecraft's attitude dynamics in the motion model. An attitude determination accuracy of five degrees is demonstrated, a performance verified using flight data from the University of Michigan's RAX-1. Although the filter was designed for the specific problem of a satellite without gyros or attitude determination it could also be used to provide smoothing of noisy gyro signals or to provide a backup in the event of gyro failures.

No abstract

A method is presented to optimize the orientation of directional sensors and instruments in a vehicle body-fixed frame. Directional dependence is included by creating a uniformly distributed set of directions in the body-fixed frame and formulating the objective as a function of these directions. The method is demonstrated by application to photodiodes for sun sensing, for which the covariance of the sun vector estimate is derived as a function of the photodiode configuration. The measured sun vector angular accuracy is then minimized as a function of the configuration, which enables the most accurate sun sensing with the given hardware. This technique maximizes subsystem performance and provides a design method to replace traditional, iterative design approaches to sensor placement.