ESA GNC Conference Papers Repository

Recent investigations of the potential use of automated image-based navigation in the approach phase of small body missions indicate its' feasibility (initial orbit close to a prescribed, safe distance from the target body) under specific observability conditions (a Sun angle not larger than 135deg), using instruments with high angular resolution (~0.1mrad/pixel), and in case a suitable strategy (weekly manoeuvres) is implemented. The work presented here demonstrates the approach taken in order to obtain the above results. A detailed camera simulator (including high precision ephemerides, photon flux calculations, and electronic models including several noise effects) is used to generate realistic images for a variable exposure camera based on the HAS3 detector. Light curves representing center-of-brightness (CoB) to center-of-mass (CoM) displacements over variable sun angles and object rotation angle, for object of varying irregularity and at varying distances, are used to generate 2D maps representing the ground truth. Then, suitable image processing algorithms are implemented to extract the faint object location from within images of star fields, after which results are characterized in detail to yield a performance model of the CoB image processing algorithm. A hardware (including mixed CPU/FPGA) implementation of the CoB algorithm is described, and its' detection performance with respect to the software-only version of the same algorithm is assessed. Taking as an example the Asteroid Impact Mission approach sequence, a covariance analysis is then performed to derive the arrival performance in terms of the achieved orbit parameters around the Didymos near-Earth asteroid. Finally, Monte Carlo results are presented in support of the results achieved.