Modified Kalman filter based autonomous orbit determination using horizonscanner measurements

Abstract

With a pair of Horizon Scanners (HS) working as attitude sensors, four horizon points are obtained during each scan. A simplified non-linear model is developed for the horizon points using an appropriate choice of frame. The modified gain extended Kaiman filter is used to estimate the satellite position in the orbit reference frame based on the non-linear Horizon Scanner data. Through simulation studies it is shown that addition of one more scanner in the roll-yaw plane provides accurate estimates of the position. Under steady state conditions, simulation results show that estimates do not diverge even if there is a data break in the roll-yaw HS data. The achievable accuracies are of the order of 2 km with about 0.1 ° (1σ ) noise in the horizon points.