Docstoc

User Interface Design for Moon-and-Star Night-Sky Observation

Document Sample
User Interface Design for Moon-and-Star Night-Sky Observation Powered By Docstoc
					                                                                                            AAS 03-128




                       User Interface Design for Moon-and-Star
                        Night-Sky Observation Experiments

                           Christian Bruccoleri∗ , and Daniele Mortari†

                                                Abstract
            The common ways to test new attitude sensors (engineering models, or prototype
            units) usually requires, very expensive ground test equipments. This is why the
            development of alternative lower cost validation hardware and software for night-sky
            equipment, has grown so fast. Many of the existing sensors for attitude estimation
            can be validated by night sky experiments. This paper deal with the development
            of a software interface used to drive experiments for a Star Sensor, and a Moon-Sun
            sensor. It can be easily extended to similar tasks. The software proposed, among
            others, performs star identification, attitude estimation, attitude control, and local
            time synchronization.



Introduction

Optical attitude sensors for spacecraft attitude determination involve several components.
Since the earlier versions, they have been improved in terms of both hardware and software.
Not only have individual sub-components changed and improved with time, but also whole
new attitude sensor types have been proposed and, some of them, also developed.
   The common ways to test these new ideas are, usually, very expensive. Ground test
equipment is very expensive, but also prototype units designed for space flight tests re-
quire big efforts in terms of funding and resources. This is why, during the last ten years,
the development of alternative lower cost validation hardware and software for night-sky
equipment, mainly based on cots components, has grown so fast. In particular, ground-
based night-sky experiments performed using commercial product (as notebooks, digital
DSP cameras, GPS receivers, equatorial motorized mounts, etc.) are used to prototype
new optical attitude sensors.
   With only a few exceptions (one in particular being the infrared Earth Sensor) most
of the available sensors for spacecraft attitude estimation can be validated by night-sky
experiments, as for instance those observing Sun, Moon or stars. Some of the associated
data processing, requires the knowledge of time and position of the observation site, and
   ∗
     Ph.D. Candidate, Department of Aerospace Engineering, H.R. Bright Building, Room 620c, Texas A&M
University, College Station, TX 77843-3141, Tel. (979) 458-0550, FAX (979) 845-6051, bruccoleri@tamu.edu
   †
     Associate Professor of Aerospace Engineering, Department of Aerospace Engineering, H.R. Bright Build-
ing, Room 741A, Texas A&M University, 3141 TAMU, College Station, Texas 77843-3141 Tel. (979) 845-
0734, FAX (979) 845-6051, mortari@tamu.edu


                                                     1
that can be easily obtained via inexpensive GPS receivers, while CCD digital cameras are
usually used to obtain visual observations.
   For instance, a new optical attitude sensor called the Moon-Sun sensor (Ref. [5]) which,
looking at the Moon, provides a self sufficient data set for a complete three-axis attitude
estimation (it mainly identifies the Moon and the Sun directions), represents one of the
possible instruments that can benefit from night-sky experiments. The proposed instru-
mentation and user interface, actually, have been designed for the feasibility study for this
sensor as well as for any single FOV star tracker.
   The software developed tests the capabilities and precision of an equatorial mount, car-
rying two digital cameras (VC51) whose relative orientation is fixed and known with respect
to the platform where they are mounted. The equatorial mount has two electrical motors
which, as for every standard equatorial mount used by astronomers for night-sky obser-
vations, set declination, and right ascension of the platform. These motors are directly
controlled by an external device (the Skysensor 2000-PC) which converts ASCII commands
(sent by the notebook computer) into the corresponding motor controls. The Skysensor
2000-PC, which is a small device with an on-board DSP and a serial interface for a PC
connection, is compatible with the LX200 command set (developed by Meade to control the
LX family of equatorial mounts). Communication with a GPS handheld device, in order to
read longitude, latitude, altitude, and correct time, is also provided via a serial port. To
achieve such goals, a software interface which combined all controls in a single program has
been developed; this provides great usability for operators and allows them to focus more
attention on their experiments, and less on the control tasks.
   The proposed paper illustrates the software developed and the hardware used for the
simulation of a Star Sensor and Moon-Sun sensor ground tests night-sky experiment. This
interface allows a notebook to communicate, via four different serial ports, with:

  1. A digital camera for Moon observation;

  2. A digital camera for star observation;

  3. A GPS receiver for timing and position determination;

  4. The two motors of an equatorial mount to set the orientation.

    In particular, for star sensor validation, the interface developed allows the user, just by
clicking, to send commands to the star observing camera in order to execute the following
useful tasks:

  1. Take a picture;

  2. Accomplish centroiding by a new approach which uses recursive functions;

  3. Accomplish the lost-in-space star identification process with the most robust Pyramid
     algorithm;

  4. Estimate attitude with either ESOQ2 algorithms;

                                              2
  5. Evaluate the reliability by means of the Attitude Error Estimator instead of the square
     root of the trace of the covariance matrix.

   while, for Moon-Sun sensor validation the user can:

  1. synchronize PC time with GPS time;

  2. evaluate the observation site position;

  3. change the attitude of the equatorial mount by a specific attitude in one of several
     representations (Euler-Rodriguez symmetric parameters, attitude matrix, Gibbs vec-
     tor, Principal axis and angle, etc.) as well as to move each single motor for a given
     angle;

  4. send commands to a digital camera to execute any specific task for Moon image
     processing.

   Since these tasks cover a wide range of possible applications, the characteristics and the
problematic details of this interface are therefore presented to a large readership.


Description of Instruments

Both the star sensor and the Moon sensor require image acquisition and processing. There-
fore, the hardware used to simulate these sensors can be shared, while the image processing,
and attitude determination software must be different. The use of digital (CCD) cameras
in astronomy is rather common in these years, and our choice was to use the Visual Com-
ponents VC51 camera for the image acquisition‡ .
   Usually one download the image taken by the camera into a desktop or laptop computer
for the image processing, but these computers have much more computation capability,
and memory, than the hardware commonly used on a spacecraft. For this reason we chose
a camera that has an integrated DSP, and a limited memory that can better simulate
an autonomous device. This is a more severe test for algorithms, and introduces some
programming constraints that must be taken into account. The results of this processing
are then transferred using a serial interface RS232-V24, and stored, on a laptop computer
that is a command center for the experiments. Because of the hardware and software
limitations of the serial port on the camera, long chunks of data, such as transferring an
image requires, are prone to flow control errors. Therefore, a special resilient transferring
software has been developed to solve this problem.
    The VC51 Camera has a CCD with 752×582 pixels while the pixel’s size is 6.5µm×6.25µm.
It is equipped with an ADSP2181 32-MHz digital processor, a memory of 16K×16 bits for
programs and 16K×24 bits of data. For the image storage, and processing, the DSP is
connected with a BUS to a DRAM of 8 MBytes. It is possible to save persistent data on
  ‡
     See Vision Components GmbH, http://www.vision-components.de/longinfo.html, for more Techni-
cal Information about VCxx cameras family.


                                               3
board the camera using a 2 Mbytes EPROM. The lenses used are common Nikon lenses,
connected through an adaptor. One critical point for the star identification process is that
the focal length of the objective must be known with good accuracy, therefore a calibration
technique must be used. In general, besides the ground camera calibration, you may need
to recalibrate the camera when the spacecraft is in orbit, and at this point you need to
do a “manual” identification of the stars to calculate the focal length, as explained in [11].
An alternative, and promising, method to avoid this manual identification of the stars is
proposed in [12].
   There is a specialized version of the GNU C compiler that compiles the C code for this
specific target environment. Unfortunately this compiler is not ANSI compliant, therefore
the code written specifically for the VC51 is not portable to other similar hardware devices.
On the other hand, the star identification Pyramid algorithm [7] does not rely on a specific
hardware, and therefore it can be easily ported to any platform.
   The Moon-Sun Sensor experiment requires a GPS device, as explained later, therefore we
used a portable device, Magellan GPS315§ . This device is connected to the laptop computer
also using a serial interface. It provides accurate timing, and location information.
   The Moon tracking experiment, also explained later, needs an automatic positioning of
the cameras, and thus we mounted them on an equatorial mount, VIXEN GP-DX, powered
with two motors, one for the Right Ascension, and the other for the Declination. This
equatorial mount must be controlled by the laptop computer, therefore it is connected to a
Skysensor 2000PC device, that act as a low level control device, while it is in turn controlled
by the software on the laptop using the serial interface.
    As mentioned through this description, a laptop computer is needed with a multiple
serial port PCMCIA interface. The software has been developed, so far, for the Microsoft
Windows (Win32) family of operating systems. A portable version of this software is under
consideration for the future development of this project. The whole hardware configuration
is described in Fig. 1


Star Tracker overview

A star tracker is an attitude sensor that uses the directions of “fixed” stars to evaluate the
attitude of the spacecraft with respect to the inertial reference frame. In Fig. 2 the basic
operation of a star tracker is shown. A star image is scanned by the Centroiding Algorithm
to identify the position of all candidate stars’ centroid on the CCD. Using this information,
and the focal length the direction of the stars in the camera reference frame is known.
    The Star identification algorithm finds matches between stars in the camera reference
frame, and the same stars in the inertial reference frame, using the on-board star catalog.
Then, if successful, it passes these matched unit vectors to the Attitude Determination Algo-
rithm. The latter evaluates the attitude, and outputs the attitude estimated - a quaternion,
for instance. The user interface has then the capability to convert this representation in
any other format needed by the user.
  §
   Magellan GPS, http://www.magellangps.com/en/products/product.asp?PRODID=2, Technical Infor-
mation about Magellan GPS 315.


                                              4
                                                 α



                                                     Camera support



                                VC                              51
                                  51                          VC




                            Equatorial mount

                                                                                          Multiple            Laptop
                                                                                          Serial Interface


     Power Line
     DC 9 V



                                                                                               Legend

                                                                                                                 Data Line

                                                                                                                 Power Line
                                                                                 GPS
                                       Skysensor 2000PC




Figure 1: Instruments description, wiring, and data connections




Image
                       Raster                                               N Centroids          Calculate "Object"
                       Image                   Centroiding                  of Objects           direction in Body Frame
Acquisition


                                                             N unit vectors {b} in the body frame




                                         Star Identification


                                                     n<=N matches <r,b>


                                                                                                                       Attitude
                                 No                                   Yes         Attitude Determination
                                           Succesfull ?                                                                Estimation




                  Figure 2: Overview of how a star tracker works




                                                        5
Software components

The software components used for this experiment can be described as pertaining to 4 major
tasks, namely

  1. Image processing - image acquisition, star detection, and centroiding;
  2. Attitude Estimation - star identification with Pyramid, ESOQ2 for attitude estima-
     tion;
  3. User Interface - control and show results of the previous tasks;
  4. Serial Communications - allow communication between cameras and laptop, in detail
     it permits image downloading for further analysis and validation.




Figure 3: Image of one dialog of the user interface used to control cameras, GPS and motors


   Image Processing, Star Identification, and Attitude Estimation simulate the software
on board the spacecraft, while the User Interface simulates the ground programs. The
communications system pertains to both.
   Each of these tasks would require several pages of detailed description therefore, the
interested reader should find more information in the given references. Here follows a short
description of each task, and of the problems involved.

                                            6
   The acquisition of the image, given the CCD, the lenses, and the focal length, has
the exposure time as parameter. Usually one choose a specific range of star magnitude to
observe, specifically tailored for the optical system, the CCD, and the mission profile. These
constraints are used to specify values for the parameters in the following phases i.e. which
stars are in the catalog, centroiding tolerance, key vector tuning, etc.).
   After an image has been stored ¶ it is processed. For each light spot on the CCD, all
the adjacent bright pixels above a given threshold are collected, and the centroid of the
star is calculated (Ref. [1]). For the Moon Sensor, the image processing techniques are
more elaborated, the objective in this case is to determine the center of the Moon (or of
the Earth, Sun) (Ref. [5],[6]).
   With the centroid of the stars available, and knowing the focal length with enough
accuracy (i.e. ±0.5 × 10−3 mm) it is possible to calculate the directions of the stars in the
camera reference frame. How to transform from the camera reference frame to any other
body frame of interest it is assumed to be also known. If (xi , yi ) is the position of the
i − th star centroid on the CCD plane, the star direction ˆ can be easily calculated using
                                                            b
the pin-hole model, where the optical axis offset has been assumed to be 0:
                                                                  
                                                              −xi 
                                     ˆi =            1            
                                     b                        −yi
                                                    2
                                             x2 + y i + f 2  f 
                                              i
                                                                 


   After the above steps have been completed, the Star Identification System comes into
play. For this experiment we assumed no prior knowledge of the attitude, therefore it
resembles the Lost in Space scenario. One can take advantage of the knowledge of the
Earth hemisphere where the experiment is running, ruling out at least half of the star
catalog, but we did not used this fact for this work, to not lose generality. Pyramid is
the algorithm used in this system for star identification. It has been already tested, with
success, on night-sky ground experiments, on the HETE Spacecraft [9], a custom version
of it will be used on the Inertial Stellar Compass developed by Draper Lab. [10], and it
has been selected for the GIFTS mission that firstly test the Multiple Field of View Star
Tracker [8] concept. The flow chart is shown in Fig. 4
   The heart of Pyramid is the k-vector (Ref. [13], [14], and [15]). This lookup table
technique cause the range searching required for star pair matching to be extremely fast.
Thus, even on the limited speed processor of the VC51 camera, the star identification is
accomplished in no more than a second. The attitude estimators used are ESOQ [3], and
ESOQ2 [4].


Moon-Sun Sensor

The dynamic of the experiment that the proposed software is designed to drive (Ref. [5, 6]) is
the following: when the instrumentation has been setup on the local Earth site the attitude
of the camera is completely unknown. Then one of the two digital cameras is used to shot
  ¶
      or while reading the CCD, without storing it



                                                         7
                                                               Start
           n: number of stars deteceted by centroiding
           Uses:
                Star Catalog
                                                              CASE
                K−Vector
                                                   n>3       n=3          n<3




                                                          Identify i, j, k
                      Get a Triangle i−j−k




                                                         Purge specular
                     Get a reference star    r           solutions




                     Confirmation phase:                     Unique             no
                     check pairs r−i, r−j, r−k               Solution ?

                                                                       Yes


                           Unique            Yes         Identify remaining
                           Solution ?                    stars, or spikes.

                                  No

                No         Last
                           triangle?

                                  Yes


                       Purge specular
                       solutions




                           Unique
                           Solution ?
                                            Yes

                                  No
                                                         Output Catalog
                       Failure: return                   Entries of                  Failure: return
                       error code                        Identified Stars            error code




                                                               End



Figure 4: This flow chart is a high level description of the Pyramid algorithm




                                                         8
a night sky picture and on board software (Pyramid ) is used to perform star identification
and therefore camera attitude is evaluated with respect to the inertial reference system.
   Now the relative position of a camera with respect to the other is known, and using
the GPS it is possible to evaluate the orientation of the cameras with respect to the local
reference system. At this point, using a Moon propagator, one can evaluate the trajectory
of the Moon as seen from the Earth site and the control software has enough information
to orientate cameras in order to lay on the same plane of the Moon trajectory. The Moon
propagator can be built using the NOVAS library . Two shot of the Moon are taken
at different times, without changing the attitude of the instruments. In this way one can
compare the accuracy of the Moon sensor with respect to a star tracker and also a differential
analysis with the same instrument can be done, within the same observation session.


Tests and results

The system was extensively tested with simulation before trying real night sky experiments.
Star Identification soundness, image transferring, and camera positioning were all validated
in the laboratory. In particular, for the star identification thousands of test, with perturbed
data, were executed, with a negligible failure rate.
   A star image obtained with our system is shown in Fig. 5, while part of the output of
the star identification from the camera is shown below.




                                Figure 5: A night sky picture



    See the WEB page http://www.usno.navy.mil/products.shtml of the Naval Observatory Vector As-
trometry Subroutines (NOVAS).




                                               9
Sensor data:
nH: 738 nV: 570 psH: 0.006500 psV: 0.006250 sigma: 10.000000
focal 24.3 mm Magnitude Threshold: 5.3
Number of stars loaded = 2306 Number of star from centroiding: 6

Centroiding coordinates:
1   459.22968, 299.25259
2   19.00713,   406.22389
3   115.65025, 706.99179
4   458.54876, 312.65150
5   388.72967, 443.83297
6   44.35253,   267.53226
Starting star identification..

Number of solutions: 1 Identification:
Star # Cat #    RA      DEC
 1 !! not identified
 2 263 05 16 41.368     45 59 52.050
 3 151 05 59 31.651     54 17 4.700
 4 149 05 23 27.838     57 32 39.710
 5 150 05 54 50.766     55 42 25.190
 6 !! not identified

Attitude (quaternion):
Q=<0.017, 0.307, 0.948, 0.082>


Conclusion

In this paper we focused our attention on how different technologies can be put together
to obtain a relatively cheap test environment for new Star Sensor. Not only it provides
the needed test capabilities using off-the shelf electronics and lenses, but also offer the
possibility, through software emulation, to compare different kind of sensors, and eventually
have a better tuning of the final prototype parameters. It is also a good teaching tools, to
allow students a better understanding of the principles of a Star Tracker design.
   The work done so far lays the basis for the development of a more comprehensive work-
bench for Star Sensors that should provide, for instance:

   • An improved graphical interface;

   • Support for the most common commercial cameras used for astronomy;

   • The possibility to add different modules as plug-in components, allowing the user to
     customize the system to meet its own specific needs.

  The high sensitivity of star identification to the variation of the focal length has been
a major time consuming issue for the experiments. A fully automated process for the

                                            10
calibration phase should be, therefore, included in the next developments.
   The night tests successfully accomplished guided us toward the refining of the Pyramid
algorithm, discovering several minor bugs, and eventually allowing the people of the HETE
mission [9], and of DRAPER laboratory[10], to use Pyramid on board of real spacecrafts.


Acknowledgment

We would warmly like to thank dr. Serena La Rosa, and all the people at Progetto S.
Marco, University of Rome, ”La Sapienza”. We feel indebted also to prof. John L. Junkins,
and prof. Thomas C. Pollock for their precious advice.


References
 [1] Mortari, D., Bruccoleri, C., La Rosa, S., and Junkins, L.J. “CCD Data Processing
     Improvements,” International Conference on Dynamics and Control of Systems and
     Structures in Space 2002, King College, Cambridge, England, July 14-18, 2002.
 [2] Mortari, D., “The Attitude Error Estimator,” International Conference on Dynam-
     ics and Control of Systems and Structures in Space 2002, King College, Cambridge,
     England, July 14-18, 2002.
 [3] Mortari, D. “ESOQ: A Closed-Form Solution to the Wahba Problem,” Journal of the
     Astronautical Sciences, Vol. 45, No. 2, April-June 1997, pp. 195-204.
 [4] Mortari, D. “Second Estimator of the Optimal Quaternion,” Journal of Guidance,
     Control, and Dynamics, Vol. 23, No. 5, Sept.-Oct. 2000, pp. 885-888.
 [5] Mortari, D. “Moon-Sun Attitude Sensor,” Journal of Spacecraft and Rockets, Vol. 34,
     No. 3, May-June 1997, pp. 360-364.
 [6] Mortari, D., and Gigli, S. “Earth-Sun Attitude Sensor: Hardware Design and Ground
     Tests,” Paper 99-438 of the AAS/AIAA Astrodynamics Specialist Conference, Gird-
     wood, AK, August 15-19, 1999.
 [7] Mortari, D., Junkins, J.L., and Samaan, M.A. “Lost-In-Space Pyramid Algorithm for
     Robust Star Pattern Recognition”, Paper AAS 01-004 Guidance and Control Confer-
     ence, Breckenridge, Colorado, 31 Jan. - 4 Feb. 2001.
 [8] Mortari, D., Pollock, T.C., and Junkins, J. L. “Towards the Most Accurate Attitude
     Determination System Using Star Trackers”, Advances in the Astronautical Sciences,
     Vol. 99, Pt. II, pp. 839-850.
 [9] Crew, G. B., Vanderspek, R., and Doty, J. “HETE Experience with the Pyramid
     Algorithm”, MIT Center for Space Research, Cambridge, MA, 02139 USA.
[10] Brady, T., Tillier, C., Brown, R., Jimenez, A., and Kourepenis, A. “The Inertial Stellar
     Compass A New Direction in Spacecraft Attitude Determination”, Paper SSC02-II-1,
     16 th Annual USU Conference on Small Satellites.

                                             11
[11] Samaan, M. A., Griffith, T., and Junkins, J.L. “Autonomous on-Orbit Calibration
     Of Star Trackers”, 2001 Core Technologies for Space Systems Conference, Colorado
     Springs, CO, 27-30 Nov. 2001.

[12] Samaan, M.A., Mortari, D., and Junkins, J.L. “Non Dimensional Star Identification
     for Uncalibrated Star Cameras”, Paper AAS 03-131 Space Flight Mechanics Meeting,
     Ponce, Puerto Rico, February 9-13, 2003.

[13] Mortari, D. “A Fast On-Board Autonomous Attitude Determination System based on
     a new Star-ID Technique for a Wide FOV Star Tracker,” Advances in the Astronautical
     Sciences, Vol. 93, Pt. II, pp. 893-903.

[14] Mortari, D. “Search-Less Algorithm for Star Pattern Recognition,” Journal of the
     Astronautical Sciences, Vol. 45, No. 2, April-June 1997, pp. 179-194.

[15] Mortari, D., and Neta, B. “k-vector Range Searching Techniques,” Paper AAS 00-128
     of the 10th Annual AIAA/AAS Space Flight Mechanics Meeting, Clearwaters, FL, Jan.
     23-26, 2000.




                                          12

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:109
posted:3/16/2010
language:English
pages:12