ARTICLE IN PRESS

Planetary and Space Science 58 (2010) 838–846

Contents lists available at ScienceDirect

Planetary and Space Science

0032-06

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/pss

Attitude and angular rates of planetary probes during atmospheric descent:Implications for imaging

Ralph D. Lorenz n

Johns Hopkins University Applied Physics Laboratory, 11100 Johns Hopkins Road, Laurel, MD 20723, USA

a r t i c l e i n f o

Article history:

Received 22 September 2009

Received in revised form

22 December 2009

Accepted 11 January 2010Available online 25 January 2010

Keywords:

Spacecraft

Dynamics

Parachute

Imaging

Planetary entry probes

33/$ - see front matter & 2010 Elsevier Ltd. A

016/j.pss.2010.01.003

esponding author. Tel.: +1 443 778 2903; fax

ail address: Ralph.lorenz@jhuapl.edu

a b s t r a c t

Attitude dynamics data from planetary missions are reviewed to obtain a zeroth-order expectation on

the tilts and angular rates to be expected on atmospheric probes during descent: these rates are a

strong driver on descent imager design. While recent Mars missions have been equipped with capable

inertial measurements, attitude measurements for missions to other planetary bodies are rather limited

but some angular motion estimates can be derived from accelerometer, Doppler or other data. It is

found that robust camera designs should tolerate motions of the order of 20–401/s, encountered by

Mars Pathfinder, Pioneer Venus, Venera and the high speed part of the Huygens descent on Titan. Under

good conditions, parachute-stabilized probes can experience rates of 1–51/s, seen by the Mars

Exploration Rovers and Viking, Galileo at Jupiter, and the slow speed parts of the Huygens descent. In

the lowest 20 km of the descent on Titan, the Huygens probe was within 21 of vertical over 95% of the

time. Some factors influencing these motions are discussed.

& 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Early planetary probes supported only low data rates, andimaging instrumentation was resource-intensive and limited incapability. However, improvements in telemetry bandwidth andin camera technology make it now practicable to include descentimagers on probes or landers, and such an imager was a centralexperiment on the Huygens probe (a descent imager was alsoflown on the Phoenix lander, but not used) and imagers are underconsideration for many future missions, notably to Titan or Venus(e.g. Campbell and Shepard, 1997; Moroz, 2002). A given imagerdesign will require a certain exposure time in order to obtain thedesired signal-to-noise, and if the vehicle rotation during thattime approaches the scale of an image pixel, the image will beblurred. Thus design of such descent imagers requires anexpectation or specification of the angular rates of the vehicle.Further, expectations for other measurements may also beconditioned by the expected angular motions, e.g. the range ofangle of attack expected may influence the design or placement ofinlets for gas sampling, and the sampling strategy for dynamicalmeasurements such as Doppler or accelerometer data maydepend on expected rates.

While the motion of a probe, or even the multibody dynamicsof a probe–parachute system, can in principle be forward-modeled by integrating the equations of motion, the information

ll rights reserved.

: +1 443 778 8939.

to completely specify the system is rarely available. Not only isthe full set of aerodynamic coefficients not always available forthe relevant conditions (Mach and Reynolds number), but thewind environment is never going to be known exactly. In manycases even statistical descriptions of turbulence are lacking. Thus,while the response to some specified disturbance can bepredicted, as may the character of any self-excited pseudoperiodicmotions, it is generally impossible for the engineers of a probesystem to provide a quantitative description of the probability ofencountering various degrees of angular motion in a realplanetary environment.

However, one can at least provide the purely empiricalobservations of vehicle motion from previous planetary descents,ignoring for a moment the extent to which these are drivenby aerodynamic instabilities or by turbulent motions in theenvironment. Summarizing such observations is the goal of thepresent paper, to provide at least some quantitative basis forexpectations pending more detailed analysis of a specific vehicleand environment.

2. Data

We discuss below the relevant planetary missions, in chron-ological order. For each mission we note the data available, theoverall character of the attitude motion, and relevant remarkson the vehicle configuration: the interested reader will find manyadditional details of the various missions in Ball et al. (2007).

ARTICLE IN PRESS

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846 839

The motion results are summarized in Table 1. In all cases werefer to the axis of symmetry, nominally vertical, as the axial orZ-axis, with x and y to refer to transverse accelerations, nominallyin the horizontal plane. Where ‘g’ is unmodified, it refers to Earthgravitational acceleration of 9.81 ms�2; a subscript indicates localgravity on the relevant planet (i.e. gm is Mars surface gravity at3.7 ms�2, gt is Titan surface gravity �1.35 ms�2, etc.).

2.1. Viking

The Viking landers had a 1 min period of parachute descentbefore making a landing using retro-rockets. The Viking landerswere very well instrumented, with an inertial reference unit with(mechanical) gyros and accelerometers, a radar altimeter andDoppler radar, although only one report examines these datafor the descent in any detail (Seiff, 1993). Of the 1 min underparachute, about 20 s saw rapid deceleration from �100 m/sdescent speed at 6 km altitude to about 55 m/s, at which speedthe landers descended between 3.5 and 1.5 km altitude. Duringthis quasi-steady period, the inferred angle between the sensedacceleration and vertical varied between 4 and 141 for Viking 1,and 2–101 for Viking; horizontal winds of the order of 10 m/swere determined. Angular rates are not reported, but someindication of the character of the motion can be inferred from thecharacteristic variation (Seiff, 1993, figure B1) of sensed horizon-tal acceleration of the order of 0.5 ms�2 (�gm/10) in �5 s, or�11/s. One and a half cycles of a pronounced swinging motionwith an amplitude of some 101 was seen, however, during thehigher-speed decelerating part of the Viking 2 descent. The periodthis motion of 15 s is commensurate with the period of a simplependulum with a length equal to the parachute line length of�32 m—the corresponding rate is �41/s. Even considering thisepisode, the large Vikings (�660 kg, with 16.2 m Disk-Gap-Band{DGB} parachutes) saw very little attitude motion overall.

2.2. Venera 11 and 12

Venera 11 and 12 were identical vehicles flown separately toVenus, arriving in December 1978, e.g. Zaitsev et al. (1980). Afterextraction by parachute from their entry shields and briefparachute descent through the main cloud deck, they jetisonnedtheir parachutes, falling to the surface of Venus stabilized by adrag skirt at their aft end (top). Karyagin et al. (1980) report somemeasurements of the angular motion of the vehicles recorded by‘data units of angular velocity’ (presumably mechanical rategyros). A periodic motion with peak amplitudes of 101/s andperiod of 2.5 s are seen (their figure 9) for Venera 11 and 201/swith 1.5 s period for Venera 12. Their table 2 reports the angle ofattack seen at parachute jetisson as 151 and during descent as 71and 81 for Veneras 11 and 12, respectively. Zaitsev et al. (1980)report the maximum transverse rotation rates of 251/s, and thatthe rates declined steadily during descent, reaching 151 and 131/sfor the two vehicles at touchdown. (They also note that rates ofsome 45–551/s were encountered during the parachute phase—

perhaps these high rates were excited by turbulence in the cloud,since notionally one would expect that rates under a parachuteshould be low.)

2.3. Pioneer Venus

The Pioneer Venus mission saw three small probes and onelarger probe descend through the Venusian atmosphere in 1978.The small probes, roughly conical in shape, descended withoutparachutes; the more spherical large probe spent a short periodunder a parachute before free-falling to the surface from 45 km, a

descent which took some �35 min. The communications weredirect-to-Earth, and bandwidth was therefore rather small: whileaccelerometers were recorded during the descent, only summarystatistics data could be telemetered. As for Viking, the originaldata are not available in a public archive (the mission predates theNASA Planetary Data System) and the descent data are discussedonly briefly in a paper by Seiff et al. (1982).

Specifically, for the small probes, the frequency of exceedingthresholds measured by the single axial accelerometer over 32 sintervals were counted—these indicate 0.01g levels were typicallyexceeded 25 times in the interval, while 0.06g levels wereexceeded between 0 and 6 times. The former data suggest acharacteristic period of the order of 1–2 s.

On the large probe, a three-axis accelerometer did permitmeasurement of transverse accelerations, and instantaneousreadings were transmitted at 16 s intervals. These can beconverted into an equivalent angle of attack, and indicate a rangeof 0–81 throughout the descent. If these are interpreted as planarmotions with a 1.1-s period (supported by the frequency countsabove and the drop tests) then the rates corresponding to the 31typical amplitude are �201/s.

A model of the PV large probe was dropped by helicopter onEarth, and three-axis accelerometer data acquired by chartrecorder. This showed axial acceleration Az varying between 0.9and 1.1g with a period of 1 s. The axial accelerations scatter 0.1g

about a mean of 0.1g—the non-zero mean value implies that therewas a coning or spiraling motion, as confirmed by observation.

2.4. Galileo

The Galileo probe external configuration was somewhatsimilar to the Pioneer Venus large probe, but Galileo had theimportant difference that its 3.8 m conical ribbon parachuteremained attached throughout its descent. Further, since itscommunications were via a relay orbiter, the telemetry band-width was higher, enabling a more complete dynamics record tobe telemetered.

Mean, maximum and minimum normal accelerations over 16 sintervals for the �1 h descent were transmitted (Seiff et al., 1998,figure 16), although sensor readings were corrupted by lowtemperatures over the period 500–1500 s. Apart from thisanomaly, the readings are remarkably constant, with minimumof 0.02g, mean of 0.05g and maximum of 0.09g. Noting thatgravitational acceleration on Jupiter gj�2.5g, the maximum valuecorresponds to an angle of attack of �2.51. The non-zerominimum value perhaps implies a conical component (with anangle of about 0.51) to the motion.

Seiff et al. (1997) interpreted accelerations as indicating a 2.21swing with a period of 4.9 s (close to the expected simplependulum period for the parachute riser length of 13.9 m),implying rates of �2.51/s.

Fourier analysis of the probe radio signal strength recorded bythe orbiter (Folkner et al., 1998—their figure 7) show a strongvariation with a 5–6 s period early in the descent—I identify thiswith the 4.9 s simple pendulum period expected from the 13.8 mparachute riser length, possibly modified by the changing addedmass of air in the parachute as the probe descends, shifting thelocation of the system center of mass relative to the center ofgravity. A second component has an initial 2 s period, slowing to�3–4 s at the end of descent—I suggest this corresponds to thesteadily declining spin around the probe axis (Lorenz, 2006) asmeasured by the Lightning and Radio Emissions detector(Lanzerotti et al., 1998); Folkner et al. (1998) appear to confusethese attributions. Atkinson et al. (1998) analyze the Dopplerrecord of the probe radio signal and find an unexplained motion

ARTIC

LEIN

PRESS

Table 1Motions on planetary probes.

Vehicle Parachute Place Flight details Data information Summary Angular rate derived

PV large probe None Venus 0–35 km. Mach=0.026–0.17,

Re=7E6–20E6

140 samples of instantaneous

Ax at 16 s intervals converted

to equivalent alpha

0–81, 31 typical If 1.1 s period assumed, 31

corresponds to �201/s

PV small probe None Venus 0–50 km. Mach 0.02–0.18,

Re=4E6–12E6

Axial accelerometer Az

threshold crossings in 32 s

intervals

40.01g �25 times per 32 s,

40.06g, 0–4 times per 32 s

PV small probe None Venus 0–50 km. Mach 0.02–0.18,

Re=4E6–12E6

Axial accelerometer Az

averaged over 16 s periods

0.002g typ. scatter

Kansas PV large

probe model

None Earth 0.7 km. V=30 m/s, Mach 0.09,

Re=1.6E6

Axial, transverse

accelerometer data on chart

recorder

Az scatter 0.9–1.1g, period

�1 s Ax, Ay scatter 0.1g about

0.1g mean

81, 1 s, �251/s

Kansas PV large

probe model

No parachute Earth Axial, transverse

accelerometer data on chart

recorder

Ax, Ay 1–2 s 31 coning, period �2 s

Implies �101/s

Huygens Main parachute (8.3 m DGB) Titan 140–110 km Camera and sun sensor data 0–141 51/s

Huygens Stabilizer parachute Titan 30–110 km Camera and sun sensor data 0–121 131/s typical, up to 401/s

Huygens Stabilizer parachute Titan Below 30 km Camera data 0–41 51/s

Huygens Main parachute Titan 140–110 km Tilt sensors at 1 Hz during

0–900 s

Standard deviation of 91 Sensor overestimates

motion

Huygens Stabilizer parachute Titan 30–60 km, 10–15 m/s. Mach

0.05–0.08

Tilt sensors at 1 Hz during

2200–4200 s

Standard deviation of 151 Sensor overestimates

motion

Huygens Stabilizer parachute Titan 20–30 km, V=8–10 m/s Tilt sensors at 1 Hz

4200–6000 s

Standard deviation of 81 Sensor overestimates

motion

Huygens Stabilizer parachute Titan 0–25 km, V=5–8 m/s Tilt sensors at 1 Hz

6000–8800 s

Standard deviation of 3–41 Sensor overestimates

motion

Huygens Main parachute Titan 30–60 km, 10–15 m/s.

M=0.05–0.08

Radial accelerometer 0–61

Huygens Stabilizer parachute Titan 20–30 km, V=8–10 m/s Radial Accelerometer 0–81

Huygens Stabilizer parachute Titan 0–25 km, V=5–8 m/s Radial accelerometer 0–21

Huygens SM2 Main parachute Earth 35–15 km, 100–30 m/s Gyros 5–301/s, lowest at small

speed

Huygens SM2 Stabilizer parachute Earth 15–5 km, 60–35 m/s Gyros 0–501/s, random

Huygens SM2 Recovery parachute Earth 5–0 km, 10–7 m/s Gyros 0–71/s, random

Huygens SM2 Main parachute Earth 35–15 km, 100–30 m/s X, Y accelerometer 0.05g rms then �0.01g

Huygens SM2 Stabilizer parachute Earth 15–5 km, 60–35 m/s X, Y accelerometer �0.04g rms

Huygens SM2 Recovery chute Earth 5–0 km, 10–7 m/s X, Y accelerometer �0.02g rms

Huygens SM2 Main parachute Earth 35–15 km, 100–30 m/s Z accelerometer 0.2–0.01g lowest at small

speed

Huygens SM2 Stabilizer parachute Earth 15–5 km, 60–35 m/s Z accelerometer �0.1g rms

Huygens SM2 Recovery parachute Earth 5–0 km, 10–7 m/s Z accelerometer 0–0.02g rms

Galileo 3.8 conical ribbon parachute Jupiter 2–20 bars Axial accelerometer 40.58 m/

s2 (0.025gj)

40.025gj 250 times per 96s,

40.05gj 10–50 times per 96s

Galileo 3.8 conical ribbon parachute Jupiter 2–20 bars Axial accelerometer 2.21 swing with 4.9 s

period=2.21/s

Galileo 3.8 conical ribbon parachute Jupiter 2–20 bars Doppler line-of-sight (LOS)

residuals

0.4–0.9 m/s (2–4 Hz) LOS

residuals with �20 s period;

0.1–1 Hz residuals with �5 s

period

2.21 swing with 4.9 s

period=2.21/s

Pathfinder 12.7 DGB +7 kg bridle Mars 0.006 bar, 61 m/s Accelerometer data Ax 0.05–0.15 g with �2 s

period. Az 0.3–0.5 g scatter

61 amplitude, 2 s period

=181/s

MER-A Mars 0.006 bar, 61 m/s IMU—Gyro and Accelerometer 31/s

MER-B Mars 0.006 bar, 61 m/s IMU—Gyro and Accelerometer 21/s

R.D

.Lo

renz

/P

lan

etary

an

dSp

ace

Science

58

(20

10

)8

38

–8

46

84

0

ARTICLE IN PRESS

Fig. 1. Dynamics data from the MER-A (Spirit) Rover IMU (see text). Events are (A)

parachute deployment, (B) heat shield separation, (C) lander separation from

backshell and descent under bridle which becomes fully extended at (D), with

retromotor fire at (E) just before touchdown. The ‘clean’ period of descent (D–E)

lasts less than a minute. The strong axial deceleration is evident (A–C), and

changes in the dynamic character are evident at each step, especially at (C). Large

angular rates immediately after parachute deployment damp rapidly, whereas

motions (D–E) are sustained at 2–41/s. Note that axis definitions are those used

elsewhere in the present paper (Z down; X, Y transverse), not those in the PDS

dataset.

Fig. 2. As Fig. 1, but for MER-B. The attitude rates vary somewhat more slowly

during the phase (D–E).

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846 841

with a 20–25 s period, on which a periodic signal with 5 s periodis superposed: the attribution to pendulum motion is furthersupported by the fact this period signal has a Doppler amplitudeof 0.5–1.5 Hz, which is consistent with a peak swing amplitude of21 or so. (The same Doppler data also indicates much longerperiod variations, later interpreted to be gravity waves—Allisonand Atkinson, (2001) - see also Seiff et al. (1999).)

Between 2 and 20 bars pressure, an axial accelerometerturbulence threshold of 0.025gj was exceeded some 250 timesin 96 s counting windows, while a 0.05gj threshold was exceededbetween 10 and 50 times (Seiff et al., 1998).

2.5. Mars Pathfinder

The descent of Mars Pathfinder is described by Spencer et al.(1998). This �400 kg vehicle underwent a �100 s parachutedescent before an airbag landing, assisted by tractor rockets.During the parachute phase (12.7 m) at �61 m/s, transverseaccelerations of amplitude �0.05g (�0.1gm) were recorded(accelerometers comprised part of an atmospheric structureinvestigation, as well as a second set for engineering purposes).These transverse accelerations had a period of �2 s—if weinterpret the 0.1gm acceleration as indicating a 61 (�arcsin(0.1))tilt and assume a planar motion, then the resultant rate is �181/s.It may be noted that the period of this motion is too short to beexplained by the 20 m bridle length l for which the pendulumperiod would be �2p(l/gm)0.5

�12 s—it is presumably instead anoscillation of the lander about the bridle attachment points with acharacteristic dimension of �1 m.

2.6. Mars Exploration Rover

The Mars Exploration Rovers (MER-A Spirit and MER-BOpportunity—see Crisp et al., 2003) were equipped with LittonLN-2005 Inertial Measurement Units (IMUs) using accelerometersand fiber-optic gyros; each mission was equipped with one IMUmounted on the backshell and one on the rover itself. Data fromthese units are archived on the NASA Planetary Data System(Dataset MER1/MER2-M-IMU-4-EDL-V1.0), and has been used forentry profile studies (e.g. Withers and Smith, 2006), but thedynamics during descent appear not to have been studied indetail.

The descent sequence was broadly similar to that of Pathfinder,with a roughly 60-s long parachute phase during which the vehicledecelerates from supersonic speed (�430 m/s) at 6.5 km altitudeto around 75 m/s before airbag inflation and rocket motor firing afew hundred meters above the ground. The EDL sequence includedseveral events that changed the dynamical character of themotion, such as lowering the lander/rover on a bridle. Someexcerpts of the data are given in Figs. 1 and 2. Some motions of upto 51/s appear upon parachute deployment, but damp out over20 s, during which time the vehicle slows to terminal velocity andthe axial accelerometer reading Az tends towards local gravity of3.7 ms�2. However, short-term fluctuations in Az of �1 ms�2 anda period of about 0.25 s remain throughout the descent:corresponding transverse accelerations are about a factor of 2smaller. The angular rate measurements vary with a period ofabout 1.8 s and amplitude of �21/s for MER-B, with slightly higheramplitude for MER-A.

2.7. Huygens

The attitude of the Huygens probe during descent has receivedmore attention than other probe missions for a couple of reasons.First, it was the first flight with descent imaging, and thus the

attitude history was important in mosaicing images together, aswell as more importantly (but perhaps less obviously) beingrequired for interpretation of other optical measurements such asthe solar aureole. Second, it was a comparatively long descent,providing a somewhat diverse range of air density and descentspeed and meteorological regime. Third, if not well-instrumenteddynamically, the probe was at least diversely instrumented, suchthat several groups of investigators received data that was moreor less affected by the probe motion. The probe mission and someof the dynamics data are summarized in Lebreton et al. (2005).The attitude motion is described in some detail in Lorenz et al.(2007). The spin history is also discussed in Perez-Ayucar et al.(2005), Sarlette et al. (2005), Lorenz (2006), Dzierma et al. (2007)and Karkoschka et al. (2007). Because of this heavy attention, and

ARTICLE IN PRESS

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846842

the fact that some of the motions were not widely anticipated(although allegedly were consistent with some expectations—

Underwood, 2006) there is a perception that the Huygens descentwas ‘rough’. In fact, as we show here, for much of the descent –including the particularly relevant descent near the ground – themotion was rather gentle.

Fig. 4. Angular rate measurements on the Huygens SM2 test on Earth. The dark

line of coherent measurements is the Z-axis rotation (i.e. the slow intended spin

around a nominally vertical axis). The scattered points are the Y-axis instanta-

neous measurements, indicating a pitching oscillation under the main parachute

that declines in amplitude as the descent slows, then increases dramatically and is

sustained throughout the stabilizer phase, and falls to a small value under the

stabilizer. The thick grey line is the smoothed absolute value of the Y-axis

measurements.

2.7.1. Huygens SM2 parachute drop

Before reviewing the Huygens flight data, we first describe datafrom a balloon drop test on Earth. This ‘SM2’ (‘Special Model 2’)test from 40 km altitude in Kiruna, Sweden in May 1995 wasconducted to demonstrate the heat shield separation and para-chute deployment sequence (Jakel et al., 1996; Underwood, 1997).Accelerometers and gyros (gyro instrumentation was sadly notcarried to Titan) documented the vehicle dynamics. The descentprofile was chosen principally to match Mach number anddynamic pressure conditions as closely as possible—thereforecertain other parameters such as flight speed and Reynolds numberwere different from what would be encountered at Titan. Datafrom the test were not published at the time, nor analyzed in greatdetail beyond establishing some confidence that the parachutesystem would work. Some anomalously high attitude motionswere noted, but dismissed as due to wind shear during the test ordynamical conditions that would not be encountered at Titan.Sarlette et al. (2005) discusses the SM2 data briefly—the descrip-tion below is the first detailed presentation in the literature of themotions recorded in the test.

The descent profile in the test is shown in Fig. 3. The vehiclefell under the main parachute from 40 to 20 km at speedsbetween 90 and 30 m/s, and from 60 to 35 m/s under thestabilizer from 20 to 5 km. The test included a third ‘recovery’parachute to slow the descent after demonstration of thestabilizer function, to avoid damaging the test article on impactunder terrestrial gravity and atmospheric density. This recoveryparachute was not part of the system flown to Titan, and lacked aswivel in the bridle.

Fig. 4 documents the attitude motion recorded by mechanicalrate gyros. The Z-axis spin, intended to follow a profile driven bysmall vanes around the periphery of the probe (like Galileo andPioneer Venus large probe) followed approximately the expectedprofile, although with a lower magnitude than expected. This first

Fig. 3. Descent profile of the Huygens SM2 balloon drop on Earth in 1995. Note

that this test included a third ‘recovery’ parachute to avoid damaging the test

article on impact under terrestrial gravity and atmospheric density that was not

flown to Titan.

increases as aerodynamic torques progressively spin the probe upto a fixed ‘corkscrew’ rate at �1000 s, this rate decreasing as thedescent speed declines in the thicker lower atmosphere. It thenaccelerates to a new profile under the stabilizer parachute,although with more short-term variations. Finally, under thestabilizer, there is a long period winding and unwinding of theparachute lines (which were not equipped with a swivel). It wasrecognized at the time that the magnitude of the spin rate waslower than expected, and the pitch angle of the vanes wasadjusted as a result (all that could be practicably done at thatrelatively late stage in the probe development). It was not noticedat the time that the direction of spin was opposite to thatexpected—this was determined by analysis 10 years later of thearchived uplooking video tape (Sarlette et al., 2005). It is presentlythought that the torque in the opposite direction was generatedfrom asymmetric airflow over the heat shield separation fittingson the probe.

The transverse motion, had it received more attention at thetime of the test, might have been considered more alarming.Initial oscillations with peak rate of about 201/s progressivelydecrease. Since the period of the motion is �0.5 s, the declineover 400 s cannot be accurately described as damping—it seemsmore to be a continuously self-excited motion whose amplituderelates to the descent speed, which progressively declines. The10 s-averaged rates declined from �7 to �31/s.

The rate increases dramatically under the perhaps poorlynamed stabilizer parachute, when the descent speed alsomakes a step increase. The instantaneous rates in some casesexceed 401/s, although the typical rate was 201/s. These datareinforce an important point, which is that because the periodwas short (about 0.4 s), these high rates do not actually generatelarge tilts.

As shown in Fig. 5, the corresponding transverse accelerationswere of the order of 0.05g declining to 0.01g under the main, and amore or less constant 0.04g under the stabilizer. Notably, whilethe sensed axial acceleration was exactly 1g under the main, itwas a roughly constant 1.08g under the stabilizer, suggesting thata conical pendulum motion may have been occurring.

ARTICLE IN PRESS

Fig. 5. Accelerometer data (rms deviation from 10 s mean) from the Huygens SM2

test on Earth. Transverse (X, Y) accelerations of �0.04g are seen under the

stabilizer, yet there is a �0.1g offset in the axial (Z) acceleration, suggesting a

conical motion. Under the main parachute (and the recovery parachute) the

descent is fairly quiescent.

Fig. 6. Tilt angles measured for the images acquired by the Descent Imager on

Huygens (Karkoschka et al., 2007) for the three main parts of descent. The solid

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846 843

2.7.2. HASI drop test

An independent parachute drop test, with a somewhat non-flight arrangement of additional equipment, was performed in2002 in support of the Huygens Atmospheric Structure Instru-ment (HASI) investigation (Fulchignoni et al., 2005). The attitudehistory of the probe in this test is described by Gaborit et al.(2005)—in addition to accelerometers, a set of magnetometersprovided important information on the attitude of the probe withrespect to a fixed reference, namely the Earth’s magnetic fieldvector. The flight-like probe was suspended beneath a 124 kgtelemetry package, hanging on a 33.5 m bifilar line, in turnbeneath a 38.7 m-long set of parachute lines to a 24 m-diameterhemispherical parachute: the arrangement was dropped from aballoon at 32.5 km altitude, making a �50 min descent.

The magnetometer data shows the probe described aconical motion—this is supported by the fact that the axialaccelerometer measured an acceleration 20–40% higher thanlocal gravity for the first �500 s of descent. The cone angleswept by this rather violent motion was some 401 some tensof seconds after release, declining to about 71 after 1000 s.A small additional component of the field fluctuations wasidentified with a �2 s period motion associated with the shortpendulum formed by the probe hanging beneath the telemetrybox. Again, we see that large motions are associated with higherspeed descent.

The axial spin history did not resemble the Huygens flight,principally because the probe was not decoupled from theparachute by a swivel—the bifilar line coupled the parachute‘elastically’ to the probe and the lines wound up and down duringdescent.

The test also carried a set of fluid-in-vial tilt sensors, whichwere also flown on the actual Huygens probe as part of theSurface Science Package (Lorenz et al., 2007). The Gaborit et al.(2005) paper, published after the actual Huygens descent, notesthat the tilt sensors respond strongly to high frequency motions(‘wobbling’), but not to pendulum motion.

line corresponds to roll about the camera axis, while the dashed line is pitch of the

camera axis up and down. Although rates were small initially under the main

parachute, the absolute tilts were somewhat variable. A somewhat narrower

distribution, with a couple of outliers, is seen during the higher-speed part of the

stabilizer descent. Finally, in the lower troposphere, the tilt distribution is quite

narrow, with tilts of almost always 21 or less.

2.7.3. Huygens descent on Titan

Dzierma et al. (2007) use fluctuations in the radio signalstrength received by Cassini (indicated by the Automatic Gain

Control telemetry – AGC – in the receiver), which are affected bythe probe attitude via the non-uniform radiation pattern from theprobe antenna, to constrain the attitude history, finding that ingeneral tilts averaged over �30 s periods were close to zer-o—thus there was no large-scale coning or pendulum motions,although shorter-period oscillations likely occurred. An interest-ing exception is the period around 1500 s when there was strongzonal wind shear—here a parachute-suspended vehicle willnaturally be inclined, such that the tension in the parachutebridle gives a horizontal force component to accelerate the probeto keep up with the changing local wind. Tilts of around 51 wereencountered here.

Some indications of pendulum swinging are provided by theincompletely recorded Doppler history of the probe signal asreceived on Earth (Bird et al., in press; Folkner et al. (2005)).Shorter-period motions are not readily detectable in these data,however.

Karkoschka et al. (2007) analyze data from the Descent Imagerand Spectral Radiometer (DISR) and its sun sensor, as well as thespin history indicated in the AGC data (Perez-Ayucar et al., 2005).They derive a set of tip-tilt angles for each of some 225 images.These data are plotted in Fig. 6, and show that the vast majority ofimages were acquired within a few degrees of vertical. Theabsolute tilt values broadly speaking decline with altitude, and

ARTICLE IN PRESS

Fig. 7. Huygens Radial Accelerometer (RASU) readings, converted to equivalent

angle, at different portions of descent. The solid line is the raw reading, which

includes a component due to the centrifugal effect of the probe spin. The dashed

line represents the difference between successive readings, thereby eliminating

the slowly varying spin component and is a measure of angular motion.

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846844

are tightly concentrated within 21 of vertical for the lowest 20 kmof descent.

Some engineering data also provides an indication of theangular motion. A radial accelerometer unit (RASU) was installed,primarily to document the probe spin history via the centrifugalacceleration since the accelerometer was deliberately sited off thespin axis. However, a subset of raw accelerometer readings wererecorded, and are discussed briefly by Perez-Ayucar et al. (2005).The statistics of these are shown – expressed as equivalent tiltangle by dividing by local gravity, and as the change in effectivetilt between successive readings – in Fig. 7. The change insuccessive readings removes the quasi-steady spin componentand leaves what must be a dynamic tilting motion, although theoverall amplitude and distribution of the tilts is broadly similar tothe raw data.

There is an interesting discrepancy between the DISR tilt datarelative to the horizon, and the dynamically indicated tilts, duringthe first part of descent. During this part of descent, there wasstrong wind shear as zonal winds fell from about 100 m/s to nearzero and then climbed causing a tilt in equilibrium attitude asdescribed above—such a tilt is a steady offset so does not appearin the accelerometer data in Fig. 7, but is seen in the absoluteattitude indicated by the camera. In both datasets, there is a muchnarrower distribution of readings under the stabilizer chute andin particular in the lowest 20 km of descent. During this period,the fraction of measurements exceeding 21 of tilt in either dataset

is less than 5%, whereas this fraction is tens of per cent at higheraltitudes.

Although the tilt sensor data as discussed in Lorenz et al.(2007) has some utility in that its relative spectral and statisticalcharacter varies during the descent, in particular allowingthe identification of the excitation of pendulum-type motionthat may indicate freezing-cloud turbulence in the upper tropo-sphere, the indicated amplitudes are rather higher than whatactually occurred. This same effect was noted on the terrestrialdrop test by Gaborit et al. (2005) and so we do not discuss the tiltdata further here, except to note that the overall variation, namelysmall motions under the main chute, larger amplitude motionsunder the small stabilizer chute in the stratosphere and uppertroposphere, and gentle motions in the lowest 20 km, is the sameas in the DISR and accelerometer datasets.

3. Discussion

A first remark may be made on the character of periodicmotions. Although an amplitude of 21 sounds like a small number,it is important to remember that if the motion has a short period,the average angular rates can still be quite high.

For a simple pendulum motion with angle F at time t givenby F=Fm cos(ot), the angular velocity O is �Fmo sin(ot). Thetime-averaged angular position and velocity are of coursezero, but the average absolute values are 2Fm/p and 2Fmo/p,respectively. A conical pendulum motion will have non-zeroaverages. Typical attitude histories may be a combination ofplanar and conical modes, possibly combinations of severalmodes with different periods (such as the ‘scissors’ mode whereinthe probe/bridle and the parachute/lines oscillate separately andout of phase).

An obvious and intuitive point is that large parachutes tend toproduce stable conditions. Bigger is better—larger systems havelonger natural periods, and thus rates for a given amplitude arelower. Additionally, the size of a system (e.g. the probe–parachutedistance, or the parachute diameter) defines a minimum lengthscale of turbulent structure to which the system will respond—aneddy much smaller than the vehicle will not cause a netperturbation. Thus larger systems truncate the effective turbu-lence spectrum, and thereby reduce the integrated energy in thespectrum and so see less disturbance.

While the Huygens dynamics in part of the descent wereindeed less quiescent than had been expected, the relative ratesunder the main and the stabilizer were evident from theterrestrial drop test. The utility of such tests is underscored.Underwood (2006) suggests that interaction between the probewake and the parachute, possible resonance between the vortexshedding frequency of the probe and the scissors pendulum modefrequency, or atmospheric turbulence, may have been responsiblefor the excitation of the stronger motion seen during earlystabilizer descent.

Considered on a planet-by-planet basis, it is clear that Titanand Venus present certain challenges for descent imaging. Thesame factor that makes these bodies easy to enter (a thickatmosphere that ensures a vehicle decelerates to terminalvelocity) reduces the light levels available for surface imaging,and thus requires certain exposure times. Further, since a slowdescent speed will require a long descent duration (hencerequiring large battery energy budgets and communicationperiods, as well as leading to greater horizontal landing pointdispersions due to wind), there is an imperative to have eitheronly a small parachute, or none at all. A large parachute reducesthe ballistic coefficient (mass/area) of the probe–parachutesystem such that its terminal velocity is too low.

ARTICLE IN PRESS

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846 845

Both the Venera and Pioneer Venus probes, lacking parachutes,saw rather rapid motions of the order of 201/s. Since considerableeffort was devoted to improving the aerodynamic stability ofthese vehicles (Seiff et al., 1982), it is not obvious whether theseangular rates can be reduced substantially without an externalstabilizer parachute (which must of course survive the harshconditions near the Venusian surface) or some active means ofcontrol. Neither approach would be inexpensive to implementand validate.

The situation at Titan is less clear-cut. The motion under thelarge Huygens main parachute was modest and slow. During thelast 20 km of descent to the surface (quite likely the altitude rangeof most interest for descent imaging) the motion under the smallstabilizer parachute was also satisfactorily small. Some combina-tion of dynamic instability and environmental excitation duringthe 900–5500 s period of the descent under the stabilizer led torather more violent motions, which perhaps should have beenanticipated in the light of the SM2 test. Prediction of the intensityof angular motions on future Titan missions would benefit from abetter understanding of the relative contributions of intrinsic vsturbulence-driven motion. Nonetheless, descent imaging was notsignificantly degraded by motion blur (Karkoschka et al., 2007).Drop testing and/or high-fidelity dynamical simulation would bevaluable in establishing expected angular rates.

Some remarks may be made on strategies for imaging fromdescent probes. First, even the high rates indicated may betolerable if the camera is adequately optically ‘fast’ in the sense ofhaving a large light-gathering area and an efficient detector. Sincethe criterion for blur is that the angular size of the pixel is sweptduring the exposure time, then low-resolution imaging is moretolerant of angular motion of the platform (or, for that matter,vibration); similarly, if the exposure time is short enough, thenlarge rates are needed for blur. In addition to these opticalconsiderations, there are some data processing strategies that canaccommodate motion. First, metrics that can be implemented on-board, or even in real-time, can detect blur (e.g. by measuring theentropy of the image, or via Fourier transforms, etc.). It may bepossible to acquire many images, and select only the least blurredfor downlink. Another approach is to acquire very short exposureimages (which will each have low signal-to-noise), correlate themto determine any angular shift, and then stack (i.e. add) theimages after applying the relevant shift. Adding images in thisway restores the signal-to-noise, but eliminates the effect ofmotion of the scene. This approach is widely used (sometimescalled ‘lucky imaging’) in video astronomy, allowing high-resolution imaging of the planets despite image motion due toatmospheric turbulence. It may be noted too that some commer-cial digital cameras are now equipped with motion compensationin this way, or by physically shifting the detector during theexposure to cancel the scene motion.

4. Conclusions

The empirical experience of angular motion on planetaryprobes has been reviewed. Probes at Venus without parachutesmay experience rates of 201/s or more. Probes with largeparachutes, including Viking, Galileo, MER and the initial Huygensdescent, see rates that are typically below 51/s.

The reason for the apparent difference between Mars Pathfin-der whose accelerometers imply rates of 4101/s and the MarsExploration Rovers is not obvious. The MERs were more massive,and the parachute design was changed somewhat.

The irregular motion of the Huygens probe during theintermediate part of its descent is still not fully understood—partis likely to be dynamic instability, although there is also some

evidence of cloud turbulence exciting motion during part of thedescent. Descent in the lowest 20 km or so was comfortablyquiescent, however, within 21 of vertical over 95% of the time.

In general there is a robust correlation of accelerometer datawith angular data when the latter is available, so accelerometerdata is a useful proxy for attitude motion. However, the properevaluation, and causal attribution, of angular motions doesrequire additional information. Finally, terrestrial drop testsare useful in giving insight into vehicle motions, and in theresponse of instrumentation to it. Such tests should not berestricted to engineering validations, but should also be exposedto mission scientific investigators to understand the effects ofmotion on their measurements, and on the utility of their data oncharacterizing the motion. Even small-scale tests are useful in thisregard (e.g. Lorenz et al., 2005).

Acknowledgement

Some parts of this work were supported by the Cassini project.

References

Allison, M., Atkinson, D.H., 2001. Galileo Probe Doppler residuals as the wave-dynamical signature of weakly-stable, downward-increasing stratification IJupiter’s deep wind layer. Geophysical Research Letters 28, 2747–2750.

Atkinson, D.H., Pollack, J.B., Seiff, A., 1998. The Galileo Probe Doppler WindExperiment: measurement of the deep zonal winds on Jupiter. Journal ofGeophysical Research 103, 22,911–22,928.

Ball, A.J., Garry, J.R.C., Lorenz R.D., Kerzhanovich, V.V., 2007. Planetary Landers andEntry Probes, Cambridge University Press.

Bird, M.K.M. Allison, S.W. Asmar, D.H. Atkinson, I.M. Avruch, R. Dutta-Roy, Y.Dzierma, P. Edenhofer, W.M. Folkner, L.I. Gurvits, D.V. Johnston, D. Plettemeier,S.V. Pogrebenko R.A. Preston G.L. Tyler Winds on Titan: first results from theHuygens Doppler Wind Experiment Nature 438, 800–802.

Campbell, B.A., Shepard, M.K., 1997. Effect of Venus surface illumination onphotographic image texture. Geophysical Research Letters 24, 731–734.

Crisp, J.A., Adler, M., Matijevic, J.R., Squyres, S.W., Arvidson, R.E., Kass, D.M., 2003.Mars Exploration Rover mission. Journal of Geophysical Research 108, 8061,doi:10.1029/2002JE002038.

Dzierma, Y., Bird, M.K., Dutta-Roy, R., Perez-Ayucar, M., Plettemeier, D., Edenhofer,P., 2007. Huygens Probe descent dynamics inferred from Channel B signal levelmeasurements. Planetary and Space Science 55, 1886–1895.

Folkner, W.M., Asmar, S.W., Border, J.S., Franklin, G.W., Finley, S.G.,Gorelik, J., Johnston, D.V., Kerzhanovich, V.V., Lowe, S.T., Preston, R.A., Bird,M.K., Dutta-Roy, R., Allison, M., Atkinson, D.H., Edenhofer, P., Plettemeier, D.,Tyler, G.L., 2005. Winds on Titan from ground-based tracking of the Huygensprobe. Journal of Geophysical Research 111, E07S02, 2006, doi:10.1029/2005JE002649.

Folkner, W.M., Woo, R., Nandi, S., 1998. Ammonia abundance in Jupiter’satmosphere derived from the attenuation of the Galileo probe’s radio signal.Journal of Geophysical Research 103, 22847–22856.

Fulchignoni, M., Ferri, F., Angrilli, F., Ball, A.J., Bar-Nun, A., Barucci, M.A., Bianchini,G., Bettanini, C., Borucki, W., Colombatti, G., Coradini, M., Coustenis, A., Debei,S., Falkner, P., Fanti, G., Flamini, E., Gaborit, V., Hamelin, M., Grard, R., Harri,A.M., Hathi, B., Jernej, I., Leese, M.R., Lehto, A., Lopez-Moreno, J.J., Makinen, T.,McDonnell, J.A.M., McKay, C.P., Molina-Cuberos, G., Neubauer, F.M., Pirronello,V., Rodrigo, R., Saggin, B., Schwingenschuh, K., Seiff, A., Simoes, F., Svedhem, H.,Tokano, T., Towner, M.C., Trautner, R., Withers, P., Zarnecki, J.C., 2005. In situmeasurements of the physical characteristics of Titan’s environment. Nature438, 785–791.

Gaborit, V., Antonello, M., Colombatti, G., Fulchignoni, M., 2005. Huygens structureatmospheric instrument 2002 Balloon campaign: probe–parachute systemattitude analysis. Journal of Aircraft 42, 158–165.

Jakel, E., Rideau, P., Nugteren, P.R., Underwood, J., 1996. Drop testing the HuygensProbe. ESA Bulletin No. 85, February 1996.

Karkoschka, E., Tomasko, M.G., Doose, L.R., See, C., McFarlane, E., Schroder, S., Rizk,B., 2007. DISR imaging and the geometry of the descent of the Huygens probewithin Titan’s atmosphere. Planetary and Space Science 55, 1896–1935.

Karyagin, V.P., Kovtunenko, V.M., Kremnev, R.S., Kuznetsov, V.V., Pichkhadze, K.M.,Sklovskaya, I.A., Finchenko, V.S., Yaroshevskii, V.A., 1980. Aerodynamics anddynamics of the Venera 11 and Venera 12 descent vehicles. Cosmic Research17, 545–551.

Lanzerotti, L., Rinnert, J.K., Carlock, D., Sobeck, C., Dehmel, G., 1998. Spin rate ofGalileo probe during descent into the atmosphere of Jupiter. Journal ofSpacecraft and Rockets 35, 100–102.

Lebreton, J.-P., Witasse, O., Sollazzo, C., Blancquaert, T., Couzin, P., Schipper, A.-M.,Jones, J., Matson, D., Gurvits, L., Atkinson, D., Kazeminejad, B., Perez-Ayucar, M.,

ARTICLE IN PRESS

R.D. Lorenz / Planetary and Space Science 58 (2010) 838–846846

2005. An overview of the descent and landing of the Huygens probe on Titan.Nature 438, 758–764.

Lorenz, R.D., 2006. Spin of planetary probes in atmospheric flight. Journal of theBritish Interplanetary Society 59, 273–282.

Lorenz, R., Dooley, J.M., Brock,K.,, 2005. Parachute drop-tests and attitudemeasurements of a scale-model Huygens Probe from model aircraft. In:Proceedings of the International Planetary Probe Workshop, Athens, June 2005.

Lorenz, R.D., Zarnecki, John C, Towner, Martin C, Leese, Mark R, Ball, Andrew J, Hathi,Brijen, Hagermann, Axel, Ghafoor, Nadeem A L, 2007. Descent motionsof the huygens probe as measured by the Surface Science Package(SSP): turbulent evidence for a cloud layer. Planetary and Space Science 55,1936–1948.

Moroz, V.I., 2002. Estimates of visibility of the surface of Venus from descentprobes and balloons. Planetary and Space Science 50, 287–297.

Perez-Ayucar, M., Sarlette, A., Couzin, P., Blancquaert, T., Witasse, O., Lebreton, J.-P.,2005. Huygens attitude reconstruction based on flightengineering parameters.In: Paper Presented at the Third International Planetary Probe Workshop,Anavyssos, Greece.

Sarlette, A., Perez-Ayucar, M., Witasse, O., Lebreton, J.-P., 2005. Comparison of theHuygens mission and the SM2 test flight for Huygens attitude reconstruction. In:Proceedings of Third International Planetary Probe Workshop, Athens, Greece, June2005.

Seiff, A., Blanchard, R.C., Knight, T.C.D., Schubert, G., Kirk, D.B., Atkinson, D.,Mihalov, J.D., Young, R.E., 1997. Wind speeds measured in the deep jovianatmosphere by the Galileo probe accelerometers. Nature 388, 650–652.

Seiff, A., DeRose, C., Muirhead, V., 1982. Unsteady aerodynamics and motions ofthe pioneer Venus Probes. Journal of Spacecraft and Rockets 19, 404–411.

Seiff, A., Kirk, D.B., Mihalov, J., Knight, T.C.D., 1999. Further evaluation of waves andturbulence encountered by the Galileo Probe during descent in Jupiter’satmosphere. Geophysical Research Letters 26 (9), 1199–1202.

Seiff, A., 1993. Mars atmospheric winds indicated by motion of the Viking landersduring parachute descent. Journal of Geophysical Research 98, 7461–7474.

Seiff, A., Kirk, D.B., Knight, T.C.D., Young, R.E., Mihalov, J.D., Young, L.A., Milos, FrankS., Schubert, G., Blanchard, R.C., Atkinson, D., 1998. Thermal structure ofJupiter’s atmosphere near the edge of a 5-mm hot spot in the north equatorialbelt. Journal of Geophysical Research 103 (E10), 22857–22890.

Spencer, D.A., Blanchard, R.C., Braun, R.D., Kallemeyn, P.H., Thurman, S.W., 1998.Mars pathfinder entry, descent, and landing reconstruction. Journal ofSpacecraft and Rockets 36, 357–366.

Underwood, J.C. 1997. A system drop test of the Huygens Probe. AIAA 97-1429, AIAAAerodynamic Decelerators and Systems Technology Conference.

Underwood, J.C., 2006. Huygens DCSS performance reconstruction. In: Proceedingsof the Third International Planetary Probe Workshop, Anavyssos, Attiki,Greece, 27 June–1 July 2005 (ESA SP-607, April 2006).

Withers, P., Smith, M.D., 2006. Atmospheric entry profiles from the Marsexploration rovers spirit and opportunity. Icarus 185, 133–142.

Zaitsev, A.V., Kariagin, V.P., Kovtunenko, V.M., Kremnev, R.S., Fokin, V.G.,Pichkhadze, K.M., Rodin, A.L., Sukhanov, K.G., Turchaninov, V.N., Fedorov,O.S., 1980. Venera 11 and Venera 12—functioning of their descent vehicles inthe planet’s atmosphere. Cosmic Research 17 (5), 532–539.