QUEST and The Anti-QUEST: Good and Evil Attitude Estimation
- PDF / 2,339,240 Bytes
- 15 Pages / 496.08 x 722.88 pts Page_size
- 50 Downloads / 204 Views
QUEST and The Anti-QUEST: Good and Evil Attitude Estimation Yang Cheng! and Malcolm D. Shuster
Abstract
An attitude determination algorithm, proposed 25 years ago as a back-up algorithm in case the QUEST algorithm failed to function properly on its maiden mission, is finally compared with the QUEST algorithm. A comparison of the attitude determination accuracies of both algorithms is carried out using the QUEST measurement model. The back-up algorithm, which we have chosen to call "The Anti-QUEST," while it lacks the special features that have made QUEST so popular today and is rather clumsy, slow and deficient in many ways, works well under the almost ideal conditions of the Magsat mission, but not generally. Further comparisons of QUEST and The Anti-QUEST provide useful insights into the different behaviors of optimal and deterministic attitude estimators. This work also presents useful practical techniques for the covariance analysis of nonoptimal algorithms.
Introduction: QUEST When the QUEST algorithm [1-3] was first developed between August 1977 and October 1978, no one, especially not QUEST's author, anticipated that it would gain the popularity it enjoys today. QUEST came at a critical time in NASA attitude mission support, when accuracy requirements and attitude computation frequency requirements for the proposed Magsat mission threatened to overwhelm the computational capacity of Mission & Data Operations at NASA Goddard Space Flight Center. No one had a good idea what to do, and the expectation was that NASA would spend a lot of time computing spacecraft attitude (on ground-based main-frame computers) and the project scientists would all have long white beards by the time they received precisely referenced magnetic field data. Fortunately, early in the development of the Magsat attitude software an algorithm, QUEST (for QUaternion ESTimator), was developed which was lightning fast and had other good properties which greatly streamlined attitude mission 'Postdoctoral fellow, Department of Mechanical and Aerospace Engineering, University at Buffalo, State University of New York, Amherst, New York, 14260. email: [email protected]. 2Director of Research, Acme Spacecraft Company, 13017 Wisteria Drive, Box 328, Germantown, Maryland 20874. email: [email protected].
337
338
Cheng and Shuster
operations. The story of the development of QUEST is an interesting story in itself, which the reader can find in reference [4]. QUEST used somewhat unfamiliar mathematics to achieve its speed, and because of this there was some trepidation at NASA3 that it might not perform as well with real data as in simulation. For that reason, about six months before the Magsat launch on October 30, 1979, it was decided to have a back-up algorithm just in case. The back-up algorithm was clumsy, slow, ugly, obviously less accurate than QUEST, and it could not do half of what QUEST could do, but no one doubted that it would work. As it turned out, QUEST performed remarkably well from the first frame of data, and the back-up a
Data Loading...
