NASA Technical Reports Server (NTRS) 20140002370: Kalman Filter Input Processor for Boresight Calibration

0.07 MB·English

by  NASA Technical Reports Server (NTRS)

input(s) to a system and the resultant frozen and the algorithm embedded in Inquiries concerning rights for the commer-
output(s) in real time or a posteriori, or an application. cial use of this invention should be addressed to
from software-generated data sets, were This work was done by Michael J. Krasowski NASA Glenn Research Center, Innovative
presented to the system, which gener- and Norman F. Prokop of Glenn Research Cen- Partnerships Office, Attn: Steven Fedor, Mail
ated outputs. Once a system is learned, ter. Further information is contained in a Stop 4–8, 21000 Brookpark Road, Cleveland,
the coefficients and constants can be TSP (see page 1). Ohio 44135. Refer to LEW-18887-1.
Kalman Filter Input Processor for Boresight Calibration
The new software brings this technology to the industrial level.
NASA’s Jet Propulsion Laboratory, Pasadena, California
Ka-band ranging provides the phase neuver each spacecraft to make neces- poses. This innovation is capable of com-
center (PC) to phase center range, sary attitude corrections. bining the input data from several cali-
which needs to be converted to the cen- While elements of this approach have bration maneuvers, evaluating individ-
ter of mass (CM) to center of mass been used for the boresight calibration ual range biases, and compressing the
range. Nominally, both PC and CM lie in the GRACE project, the new software time stamps. It uses Lagrange interpola-
on the line connecting the spacecraft brings this technology to the industrial tion for the orbit data, and a unique
GRAIL A and GRAIL B. In this case, the level. It is now fully documented and can quaternion-interpolating algorithm for
conversion should be done simply by be used by people other than its devel- interpolating the attitude data. As a re-
adding the CM-to-PC distance L to the opers. This innovation provides graphic sult, data files with different data rates
measured range for both spacecraft. outputs and log files that are critical for and independent time stamps can be
However, due to various technical rea- quick analysis and troubleshooting. In handled together.
sons, such as displacement of the true addition to the line of sight direction, This work was done by Dmitry V. Strekalov,
CM from its nominal position in the the software allows one to evaluate the Gerhard L. Kruizinga, Meegyeong Paik, Dah-
SRF, or spacecraft attitude fluctuations, CM-PC base length, which is important Ning Yuan, and Sami W. Asmar of Caltech for
the PC and CM define a unit vector that when the PM location is subject to varia- NASA’s Jet Propulsion Laboratory. For more in-
may be different from the nominal line tions (e.g., due to fuel depletion). formation, please contact Brian Morrison at
of sight. The objectives of the software This software is implemented in [email protected].
are to determine the actual line of sight Python and offers excellent cross-plat- This software is available for commercial li-
direction for each spacecraft and correct form porting possibilities. It is very versa- censing. Please contact Dan Broderick at
the previously recorded range data, and tile, and may be applied under various [email protected]. Refer to
to provide instructions for how to ma- circumstances and for other related pur- NPO-48479.
Organizing Compression of Hyperspectral Imagery to Allow
Efficient Parallel Decompression
Higher compression factors can be attained.
NASA’s Jet Propulsion Laboratory, Pasadena, California
A family of schemes has been devised pendently. As an example, for a com- much of the image data will go into a
for organizing the output of an algo- pressor that does not make use of inter- chunk, and in fact a chunk may contain
rithm for predictive data compression of band correlation, a piece could be de- incomplete portions of encoded sam-
hyperspectral imagery so as to allow effi- fined to be an individual spectral band, ples (at the chunk’s start or end). The
cient parallelization in both the com- or a fixed number of bands. In the tech- compressor iterates the process of decid-
pressor and decompressor. In these nique, the compressed output for a ing on chunk sizes and producing
schemes, the compressor performs a piece is comprised of multiple “chunks.” chunks for each piece of the requested
number of iterations, during each of The concatenated chunks for a given size, until compression of each piece is
which a portion of the data is com- piece form the compressed output for almost finished. At that point, the re-
pressed via parallel threads operating on the piece. Most of the compressed image mainder of the pieces is compressed se-
independent portions of the data. The is produced in multiple iterations, where rially without a target chunk size.
general idea is that for each iteration it is during a given iteration, one chunk is Typically, the chunk size calculation
predetermined how much compressed produced for each piece. Prior to the should seek to balance the progress
data will be produced from each thread. start of an iteration, chunk sizes are cal- through each piece, i.e., to leave equal
A simple version of this technique is culated for each piece. The chunks can numbers of samples remaining in each
applicable when the image is divided be produced or decompressed in paral- piece; a suggested procedure has this
into “pieces” that are compressed inde- lel. It is noted that it is not specified how aim. A key requirement on the chunk
24 NASA Tech Briefs, January 2014