Star Tracker
A joint ASRI&TPD/TNO project

Star tracker is a strapdown stellar sensor that takes the image of an arbitrary region of the sky, recognizes the stars and determinates its own attitude by comparing the apparent star positions with those stored in the star catalog. In contrast with most attitude sensors, it needs neither a specific reference, nor a specific orientation. Another major advantage of a star tracker is its high accuracy. Even the medium precision, wide field instrument we dealt with provides the accuracy of »10¸20 arcseconds. Instruments equipped with optics that is more powerful and a finer matrix detector will reach the precision of a few arc sec.

The star field sensors were pioneered in early 1970^th. Since then, an enormous progress has been achieved, both in hardware and in software. Use of solid-state CCD matrices instead of bulky vacuum photodetectors allowed the decrease of mass and power consumption by two orders of magnitude. Growth of available computer resources and progress in image processing algorithms made a star tracker an irreplaceable element of modern spacecraft, especially of mini and micro satellites.

Wise Observatory in Negev The goal of the joint ASRI & TPD/TNO project is development, manufacturing, and ground- -based tests of a medium-precision star tracker for near-Earth and deep space missions. TPD/TNO is responsible for the sensor head (CCD camera), while electronics, onboard computer and software are the responsibilities of ASRI.
A breadboard of the whole system was manufactured in March 2000. After some delays, the tests were carried out in Wise Astronomical Observatory near Mitzpe Ramon.

The tests lasted four nights from July 3 until July 6, 2000. In spite of unfavorable atmospheric conditions, more than 4000 frames of the near-zenith region of the sky were taken during this period.

Schematic block diagram of the star tracker

The star tracker consists of three major parts: Optical Head (OH), Frame Grabber (FG), and Processor Unit (PU). OH provides FG with an analog video signal. FG converts it to the digital form and stores the frame in a buffer memory accessible to PU. Apart from A/D conversion, FG carries out some other kinds of data preprocessing synchronously with the image readout.

Outlook of the optical head with baffle.

PU implements deep imagery processing, including separation of star images, star pattern recognition, catalog handling operations, filtering, and attitude evaluation.

Deep processing was carried out in "quasi-real time". During the observations, the frames were just stored on hard disk. In Technion, the frames were played back one by one (at the same rate as they were taken) and subject to complete processing.

A special program shell was prepared to run the imagery processing software in the interactive mode. This is the graphic display of the most complicated part of it, the star pattern recognition. It this frame, it succeeded. Black dots and numbers are thestar positions and ordinal numbers in the catalog: blue ones are their observed counterparts.

A priori estimates predicted the following end-to-end performance of the star tracker.

Sampling Rate< 3 Hz

Precision of Attitude Determination (1s)

Across the Boresight60mrad

About the Boresight300mrad

Operability at Angular Rates< 3 mrad/s

In the course of processing the collected observations,

·Hardware and software were successfully tested,

·Anticipated precision was attained,

·Ways were proposed to further improvement of the instrument performance.

In particular, a detailed map of distortions stemming from all sorts of geometric imperfections was extracted from cross-correlation of frames. Digital cancellation of these errors from the measurements will provide a further precision improvement.

For more information, contact Dr. Alex Kogan