This paper aims to present the development and performance evaluation of an attitude and rate estimation algorithm using an extended Kalman filter structure based on a body‐referenced representation of the state.
The algorithm requires only geomagnetic field data and can be used as a low‐cost alternative or as a back‐up estimator in the case of attitude sensor failures. The satellite rate is estimated as a part of the filter state and thus no gyroscope is necessary. The assessment of the algorithm performance is realized through a Monte Carlo simulation using a low‐Earth orbit, nadir‐pointing satellite.
Given some attitude and rate error requirements, the range of admissible initial errors on the filter state and the effect of un‐modelled disturbance torque are determined, along with the achievable attitude and rate accuracies.
Because the simulation set‐up is clearly stated, the results of this evaluation can be used as a benchmark for other estimation algorithms.
The necessary assumptions and approximations used to derive the filter equations are explicitly pointed out for the benefit of the readers. Well‐defined filter initial conditions are used in an extensive series of tests resulting into a unique set of findings.
