F1G. 1. Gyrostabilized camera mount with K-37 camera.

ELSON C. VANCE

U. S. Naval Oceanographic OfficeWashington, D. C.

A Photo-Astro

Ship Locating SystemA method is developed and tested for determiningthe position of a vessel at sea from a passingsatellite using a gyrostabilized camera.

INTRODUCTlON

T HE METHODS OF GEODESY employed forestablishing control and mapping ter­

restrial masses of the earth are well knownand adequately documented. However, whenthe earth's surface is considered "in-total,"one cannot fail to observe that the oceans,which comprise about 72% of the earth'ssurface, are virtually a "geodetic desert" from

both informational and observational view­points. \\lith the exception of stations on somewidely spaced islands, the ocean areas arevoid of geodetic control.

When one considers the problem of map­ping the surface of the ocean floor, it soonbecomes apparent that the traditionalmethods of observational geodesy are, ingeneral, not usable at sea. Even the mostbasic and independent of classical geodetic

660

A PHOTO-ASTHO SHIP LOCATING SY TEM 661

survey techniques-that of establishing astro­nomic positions by multiple stellar observa­tions with a precise theodolite-is not prac­tical since one could not hope to establish asatisfactory vertical reference for each ofseveral observations taken from the movingdeck of a ship.

One might ask the question, why not useelectronic positioning systems? Actually thereare presently in use several electronic shippositioning systems which have the advan­tage of operating in all types of weather andare capable of providing real ti me posi tions.Most of these, however, are still su bj ect toerrors caused by a lack of knowledge of the

indeed! It thus appeared that the BallisticCamera approach offered a means of estab­lishing a geodetic con trol net for mapping theocean floor.

I t is the in ten t of this paper to presen t abrief background description of optical satel­Ii te geodesy and then to describe, in themain text, the details of developing and test­ing the equipment. The various phases willbe discussed as follows:

A. Preliminary Planl\ingB. De\"elopmcnt of HardwareC. Prelimillary TestingD. A 1Vl iniaturized Test of the System Using a

Ship i\lotioll Simulator

ABSTRACT: The purpose, design, fabrication, and testing of a prototype Photo­Astro Ship Positioning System are described. Results of Atlantic Missile Rangeund U. S. Naval Oceanographic OJfice tests of a gyrostabilized ballistic camerasystem indicated that a ship at sea was positioned within two feet of its true loca­tion. The test involved an aircraft-borne flas/ling light observed against a stellarbackground. Extrapolation of the test data indicates that if an active or passivesatellite at an altitude of, say, 600 nautical miles, were substituted for the air­craft, the ship could be posit·ioned to within 100 to 150 feet using more sophisti­cated equipment, assuming all other parameters are comparable to those of thereference test. These conclusions al'e borne out by a concurrent theoretical studyconducted by The Nlassachusetts Institute of Technology under contract to theU. S. Naval Oceanographic Office.

exact values for the propagation of radiowaves. The system described in this paperwas designed to complement these systemsby establishing precise geodetic positions atsea, In addition, it could be used as a stan­dard with which to calibrate and evaluateelectronic systems.

Four years ago, at which timc oceano­graphic technology was expanding rapidly,geodesists were examining optical satellitegeodesy as a method for improving geodeticnets within one land mass and for makinglong intercontinental ties with accuraciescom mensu ra te wi th geodetic work. The ou t­standing advantages of optical satellitegeodesy, as com pared wi th classical geodeticmethods, are: (1) it is independent of errorscaused by deflection of the vertical and (2),the satellite is high enough above the earth'ssurface to provide a common observationpoint from the vertices of extremely largetriangles. Both of these factors are intriguingto those interested in accurately positioninga ship at sea. A system free from deflectionof the vertical errors and capable of partici­pating in triangulation schemes with far­distant land-based positions was promising

E. A Miniaturized Test of the System AboardShip

F. A Shipboard Test tilizing the SatelliteECHO I

BACKGROUND

Ballistic cameras have been used inGermany for many years, and more recentlythe United States, to track missiles and ac­curately determine the time correlated spa­tial coordinates of selected points along themissilcs' trajectory. Basically, the techniqueuscd is simple. Ballistic cameras (cameraswith high precision optics and suitably largeapertures for recording several magnitudes ofstars) are posi tioned at known geodetic posi­tions in such a manner as to provide goodgeometry for triangulation of the anticipatedpath of the missile. A series of short, timecorrelated exposures is taken prior to andafter the missile tracking event to provide areference star field on the photographic plate.

As the camera remains in a fixed position,images are formed by the apparent motionof the stars across the plate. The referencestar field serves as the basis for the deter­mination, for each camera, of the absolute

662 PHOTOGRAMMETRIC ENGINEERING

orientation with respect to any chosencelestial coordinate system. During the track­ing event, all camera shutters are open andsimultaneously record the images of flaresejected from the missile in flight. Measure­men t of the coordinates of the flare images oneach plate provide data to yield direction co­sines from each station to each flare and pro­\·ide a means of triangulating the spatial co­ordinates of each flare. It follows that if oneof the several camera stations be treated asan unknown location. it is possible to deter­mine the direction from the flares back tothe earth's surface to locate this station. Thus,\I·e have the basis for photogrammetric flaretriangulation or optical satellite geodesy. Inrecent years, however, geodesists have beenworking more with passi\-e satellites thanwi th active satelli tes and missiles. In observ­ing passi ve sa telli tes the flares or flashes arereplaced by a series of short, time-correlatedimages of the satellite in reflected sunlight.The images are produced by synchronouslyopening and closing the shutters on allcameras.

A. PRELIMINARY PLANNING

v\'hen the U. S. Naval OceanographicOffice examined the possibility of locating aship by means of observing satellites againsta stellar background, it was fortunate thatmethods and techniques-established for landbased cameras could be drawn upon. Asmentioned previously, cameras used in ter­restrial observations are normally stationary(i.e., without sidereal mounting). Pre- andpost-event stellar exposures provide a meansof determining camera orientation and stabil­ity, from which directions to the flares, flash­ing light, or shutter chopped portions of apassive satellite orbit can be computed toprovide secondary con tml poi nts II' hose spa­tial coordinates can finally be determined.

If one considers utilizing the terrestrialmethod aboard ship, he notes that the cameramust be presumed to have remained in afixed orientation between the pre- and post­event star calibration exposures. That is, itwould require that the camera be stabilizedto a second of arc over a period of, say, 10minutes. This was not considered practical.Very early in the development program, itwas indica ted that the camera would notonly have to be somewhat stabilized, but alsothat it would be necessary to develop anobserving technique that would provide forcomputation of an instantaneous orientationof the camera at the time of each flash ratherthan to employ the pre- and post-event cal-

ibration exposures normally used for stillcameras.

One method of accomplishing this instan­taneous orientation was suggested by DuaneC. Brown while he was with RCA at theEastern Test Range. Basically, the idea wasto rotate the camera at a constant angularrate during exposure. The rate would berapid as compared to the sidereal rotationbut slow enough to permit the stars to createa good quality photographic image on theplate. By introducing such a motion, smallirregularities in the star trails, resulting fromimperfect stabilization, could be elongated inthe direction of rotation and the breaks inthe images programmed from the closing andopening of the shutter could be measured.During the data reduction process, it wouldbe possible to use the coordinates of thesetime correlated breaks in each star trail tofit a mathematical description of the curveformed by the star's path across the plate.The problem would then be reduced to oneof interpolation along the curve to find thetime correlated coordinates of the stars at thecorresponding times of each flash.

As will be discussed later, this concept ofobservation was eventually abandoned. How­ever, it was a good beginning and was aban­doned because the availability of improvedequipment in the field of timing permittedthe use of a system more sui table for datareduction.

After formulating these preliminary sys­tem concepts, it appeared that the most prac­tical means of evaluation would be to assem­ble a prototype system and examine its per­formance at sea.

B. DEVELOnIEl\T OF HARDI\"ARE

The preliminary approach was to assembleand test a prototype system in order to verifyits feasibility and investigate design param­eters for an optimum system. Accordingly,available components were used when pos­sible. A two axis (pi tch and roll) gyrostabi­lized aerial camera mount and an aerialcamera with an j2.6 aperture and 300 mm.focal-length lens were obtained with verylittle effort. The particular type of camerathat was obtained had been used during theearly days of the missile ranges as a ballisticcamera with some degree of success. Cameramodification was relatively simple. The shu t­ter mechanism was replaced wi th a solenoidarrangement to provide for opening and clos­ing the lens according to any desired program.In addition, an adapter was made for thecamera back which would accommodate plate

A PHOTO-ASTRO SHIP LOCATING SYSTEM 663

holders designed to accept standard glassballistic camera plates.

The gyrostabilized mount, on the otherhand, required more modification. It hadbeen designed for use in an aircraft to sup­port an aerial camera in a vertical position.Consequently, adaptation was necessary tomake it suitable for shipboard application instabilizing a stellar camera. Figure 1 (sho\\"non first page of article), which shows theequipment after assembly, will be a conve­nien t reference in describing the yarious com­ponents.

The stand was constructed to support thecamera and mount at a convenient workinglevel above the deck. The shock mounts,which can be seen between the stand and thegyrostabilized mount, were installed toeliminate vibration from the ship's enginesand screws. These particular shock moun ts\\'ere removed later and replaced by shockmoun ts between the deck and the bottom ofthe stand. The reason for this change wasthat, al though the original shock mountsabsorbed the vibrations, they allowed themount to swing in azimuth as the ship rolledand pitched.

The stabilized mount, as it appears inFigure 1, required no physical changes to itsconfiguration other than the addition ofweights to balance the completed system. Anadapter was designed and constructed, how­ever, to support the camera in the mount.The top portion of this adapter can be seensupporting the camera trunnions. Designed toattach to an existing gimbal ring, the adapter,in addition to supporting the camera, pro­vides for azimuth settings through 360° andelevation settings from 0° to the zenith. Ascan be seen, the camera was nestled as lowas possible in the mount in order to keep thecenter of gravity of the stabilized weightclose to the intersection of the gimbal axes.This limited the minimum elevation angle tobetween 20° and 40°, depending on azimuthsetting, but reduced the size of the massivecounter balance which was lat.er installedbelow the gimbals. Perfect balance, regard­less of azimuth and elevation setting, is essen­tial for this system as the stabilizing "tor­quers" produce very little torque. The centerof gravity of all stabilized weight must be atthe intersection of the gimbal axes. Below theca mera and moun t, on the left, can be seenthe power supply. On the right is the controlpanel and electronic equipment. which wasadded to provide for an induced angular rota­tion of the camera through either the pitch orroll "torq uer."

C. PRELIMINARY TESTING

After the system was assembled and operat­ing, it was taken to sea on the USNS Rich­field, a Pacific Missile Range tracking ship, forpreliminary testing. The purpose of thesetests was to obtain stellar photography toascertain the quality of star images whichcould be obtained, and at the same ti me toevaluate the performance of this particulargyrostabil ized mou n t.

These tests showed that use of this equip­ment for test purposes was feasible if it wereused aboard a large ship in a fairly calm sea.This \\'as satisfactory, for it \\'as believed thatgood test results under these conditions wouldprovide an indication of what could be ex­pected of a more sophisticated system undernormal operating conditions.

D. A MINIATURIZED TEST OF THE SYSTEM

UTILIZING A SHIP MOTION SIM LATOR

Starting in 1962, a series of tests was con­ducted at the Atlantic Missile Range. Thesetests, a joint effort by the Atlantic MissileRange and the U. S. Naval OceanographicOffice, were performed utilizing Navy equip­ment with the exception of the camera. Theoriginal camera was removed and replaced\\,ith a \Vild B-C4, 210 mm.-fl. camera fur­nished by Atlantic Missile Range. This wasconsidered desirable for several reasons: first,the calibration constants for the lens \\'ereknown; second, the shutter was compatiblewith range timing and sequencing; third,many plate holders were already availablefor use with this camera; and fourth, themissile range ballistic camera operators werefamiliar wi th its operation. Necessary modifi­cations for the camera change-over were doneat the AMR.

Although the primary objective was to testthe system aboard ship, it developed thatmuch could be learned by testing it on a shipmotion simulator first.

The first tests aboard the simulator, Figure2, were designed to determine an optimumshutter sequence and induced angular rota­tion rate,

While modifications to camera and mountwere being made, plans were formulated forthe simulator and shipboard tests. One of themost important problem areas that requireda decision was that of timing and sequencingthe camera. It was decided that the curvefitting technique previously planned for datareduction of stellar images would be ratherdifficult and that it would be better tosynchronize the shutter with the flashing

664 PHOTOGRAMMETRIC ENGINEERING

FIG. 2. Gyrostabilized camera mount with WildBC-4 camera on ship motion simulator.

light to be carried by a high altitude aircraftto su pply secondary con trol pain ts for thesetests. This would have the advantage of pro­viding a short, discrete image of each starfor each flash and eliminate the need forinterpolation in data reduction. However,the camera would still be driven through anangular displacement in order to stretch outthe star trails and smooth ou t some of thestaLilization errors.

For the actual test, the cameras were de­ployed as depicted in Figure 3. The flashinglight was carried by an aircraft at an altitudeof 20,000 feet. Light flashes were commandedfrom the ground and synchronized to themean time of the shutter opening for thegyrostabilized camera. Supporting camerasoperated in the normal mode, i.e., a pre- andpost-event star calibration and an openshutter to record the event.

Results of this test, according to a reportby H. L. Jury of Pan American World Air­ways at AMR, showed that the stabilizedcamera was positioned to 1.22 feet CircularProbable Error with respect to the knowngeodetic position of the station. Had a satel­lite been observed at a slant range of 1,000nautical miles, the error probably would havebeen about 300 feet.

E. A MINIATURIZED TEST OF THE

SYSTEM ABOARD SHIP

The results of the motion simulator testwere extremely encouraging and plans weremade to proceed with the shipboard test. Aswill be noticed by comparing Figures 3 and4, the geometry of the two tests was quitesimilar. Two additional problems, however,needed to be considered in conducting theshipboard test which should be noted. First,as the ship could not sail any closer thanthree miles off shore, the distances betweencameras were greater and it was necessaryto fly the aircraft carrying the flashing lightat an alti tude of 40,000 feet in order to ob­tain orientation conducive to reliable geo-

I

~

.... /..Be \\'\'\\\

\ \, \

\

\\~ ~ '- :YflOS'iAB

I

I* !\ADIR POI!rr!!£ = MU.JS'!'IC CAfo'.F..RA

FIG. 3. Ship motion simulator test.

1\0; _ PJ..U.I"TlC CAMrJM9: .. Cli8':'HEODOu'rr.:

A 1'IIOTO-ASTRO SHIP LOC.\TING SYSTEM

_""':

I AlflCHAIT B01Hf~ FI.ASHt!l~ Lx a-r/1 , / -

I~I "ll

"~II /::11II I " I: ~

II 1/ " I, II I II

,I,' "':I I I I I

I I I I

~- - ,r-----r--:?i'--J;f.H..---------------I

I Ill;/ -

,I ,,/ I

/1 ,./!!':."

\ /"r.A.DIRP('tlU\ /,,-- ~ /

" '" ~,/-~ ",

- - _ ~ rRu$T...~IL1tJ MI.iI- Ir CAfl.:."'l>,A_ 1.1 lIT Oil t1<l3---------- ---

FIG. 4. Shipboard test.

665

metrical solution. Second, the ship was mov­ing slowly in order to maintain a heading (thegyrostabilized mount did not stabilize inyaw) and a method for accurately locatingthe ship was required in order to have a basisfor evaluating results obtained by the systembeing tested. rt was deduced that time-cor­related positions to within ten feet could beacquired by tracking a light on the ship'smast with cine theolites positioned at first­order locations along the shore.

Fortunately, during the period of the testthe sea was very calm. As mentioned earlier,the capabili ties of the particular gyrostabi­lized mount were limited and a relativelycalm sea was needed for a successful test.Everything went smoothly and only onechange was made in the test plan. It hadbeen planned that the ship would sail twoconsecutive courses (one normal to theother) and observe three passes of the air­craft on each course. This plan was aban­doned in favor of sailing one course twice be­cause the heading was such that naturalsheltering of the camera from the wind bythe ship's superstructure was extremely goodon this course.

As in the previous test, all data were re­duced at AMR. Results were very encourag­ing. According to a report by Messrs. ]. E.French and C. H. Rosenfield of RCA, theship was positioned with a CPE of 2.06 feet.

F. A SHIPIJOARD TEST UTlLIZI:\G

THE SATELLITE ECHO I

From favorable results on all previoustests, it was concluded only one additionaltest would be necessary to demonstrate con­clusively the feasibility of the system. Earlyin 1964, the equipment was taken to sea againfor a test using ECHO 1. Cameras capable ofsynchronous shutter operation were spreadout along the Atlantic Missile Range atknown first-order positions which "'ouldprovide good geometry to satellite orbitswhich passed over the ship. The ship wassimultaneously located with respect to trans­ponders which had been previously locatedby another method. Although the data havenot yet been reduced, they have been ex­amined, and it is believed that this test willfurther verify previous convictions.

CONCURRENT STUDIES

Early in 1963, a detailed theoretical feasi­bility study of the system was independentlyconducted under contract to the U. S. NavalOceanographic Office by the ExperimentalAstronomy Laboratory of the Mas£achusettsInstitute of Technology. The study was con­ducted by A. C. Conrod undel the ableguidance of the Laboratory Director, Profes­sor Winston Markey. As a result of the study,the U. S. Naval Oceanographic Office wasfurnished with a complete report stating that

666 PHOTOGRAMMETRIC ENGINEERING

FIG. 5. Wooden mockup model of a shipboardgyrostabilized mount for 300 mm focal-lengthballistic camera.

the system is feasible and detailing the designof an optimum system. The study concludedthat a well designed and properly operatedsystem should be capable of positioning a shipwith accuracies approaching 100 feet.

Recommendations made in the reportstated that an optimum system would con­sist of:

1. A ballistic camera with distortion uncertain­ties of less than 3 microns.

2. A timing system capable of time determina­tion for shutter functions to 0.1 millisecond.

3. A stabilization system capable of stabilizingthe camera to 2 arc seconds over a 5 secondperiod of time.

Figure 5 shows a wooden model of the sug­gested system. The train (azimu th) stabiliza­tion torquers will be con tained in the baseof the pedestal, pitch torquers will bemounted on either side of the fork, and rolltorquers will be mounted on the inner gimbal.The inner gimbal will support the camera atone end, counter balanced by the cylindricalhousing at the other end which will house thethree gyros. In actual use pointing angles fordata acquisition would be established byrotating the camera in elevation about the

roll axis and in azimuth about the train axis.As there is no need for an accurate verticalreference, the gyros will not be controlledby accelerometers, and consequently thecamera will be standing still in inertial(stellar) space. In order to stretch out thestar trails and produce measurable time­correlated star images, each of which can beidentified with one of the satellite images, thecamera will rotate through an angular dis­placemen t.

The system will be capable of use wi theither passive or active satellites and 111

either the intervisible or orbital mode.

ACKNOWLEDGMENTS

The U. S. Naval Oceanographic Officeacknowledges the work done by all who par­ticipated in the development and testing ofthe system. The Office is particularly appre­ciative of support provided by the U. S. AirForce through the work done by H. L. Jury,Pan American World Airways, whose untir­ing efforts at the Atlantic Missile Range werelargely responsible for the accomplishment ofsuccessful development and testing of theprototype system. In addition, the efforts ofG. H. Rosenfield of the Radio Corporationof America in the field of data reduction,which involved the formulation of some newcomputer programs, were equally important.

BIBLIOGRAPHY

Brown, D. C. "The Precise Determination of theLocation of Bermuda from PhotogrammetricObservations of Flares Ejected from JUNO If,"PHOTOGRAMMETRIC ENGINEERING, Vol. XXV!.No.3. June 1960.

Brown. D. c.. "Photogrammetric Flare Triangu­lation," RCA Data Reduction Technical ReportNo. 46, December 1958.

Schmid, H. H., "A General Solution to the Prob­lem of Photogrammetry." Ballistic ResearchLaboratories Report No. 1065. July 1959.

French, ]. E. and Rosenfield. G. I-I.. "ProjectEGYPT, Data Reduction and Analysis Report(Evaluation of Gyrostabilized PhotogrammetricTechniques)," Range Geodetic Committee ReportNo.1.

Jury, H. L., "Photogrammetric Ocean SurveyEquipment (POSE) Program," PreliminaryReport, Range Planning Department.