A robust position estimation scheme using sun sensor

A robust estimation scheme to locate accurately the position of the sun is proposed using a sun sensor equipped to produce four output signals: each signal is proportional to the square of the distance of the focused light from one of the four corners of the sensor plane. Intermediate measurements, termed pseudomeasurements, are derived by computing the difference between outputs taken two at a time so that the sun position is linearly related to the pseudomeasurements. Both random noise and bias in the measurements are considered. A Kalman filter-based algorithm is used to compute an optimal estimate of the position of the focused light with respect to a reference frame. A model for the signal and for the noise are incorporated into the Kalman filter so that the signal is enhanced and the noise attenuated. The redundancy in the measurement is exploited to detect and isolate erroneous sensor measurements, which are then removed. The proposed scheme is evaluated on the simulated data as well as on the actual data obtained from the sun.