A High Performance Star Sensor System for Full Attitude Determination on a Microsatellite

ABSTRACT A high precision star sensor system is presented for accurate attitude determination on the SUNSAT microsatellite. The design constraints were low cost, mass, volume and power. The principle of operation is to image stars and match the observed constellation to a star catalogue. A set of corresponding observed and reference vectors are used to feed an extended Kalman filter for full attitude knowledge. The stars are extracted reliably from a noisy image by a region growing method. The probability of mistaking a false object as a star is minimized by verifying that the shape and size of the object match that of a star. A well known matching algorithm is used to match any imaged constellation to a star catalogue within a few seconds. A tracking algorithm is then initialized to provide measurement and reference star vectors to the Kalman filter at one second intervals. The star sensor uses a matrix CCD device and a dedicated transputer processor for image processing. Calibration of the star sensor is essential for the accurate determination of the angular distance between stars and to pin point the star centroids. The calibration procedure determines the optical centre on the CCD and the exact focal length. A distortion model is also derived to account for lens aberrations and CCD tilt.