1010 woolsey[1]

A self-sustaining, boundary-layer-adapted system for

terrain exploration and environmental sampling

Phase I Final Report

C. Woolsey, Ph.D.

Aerospace & Ocean Engineering

Virginia Tech

Randolph Hall, Room 217-D

Blacksburg, VA 24061

[email protected]

G. Hagerman

Alexandria Research Institute

Virginia Tech

206 North Washington Street, Suite 400

Alexandria, VA 22314

[email protected]

M. Morrow

Aerospace & Ocean Engineering

Virginia Tech

Randolph Hall, Room 1

Blacksburg, VA 24061

[email protected]

May 2, 2005

Abstract

This report describes Phase I progress in conceptual design of a revolutionary system forremote terrain exploration and environmental sampling on worlds with dense atmospheres, suchas Titan and Venus. This system addresses several visionary challenges enumerated by NASA’sOffice of Space Flight in the broad category of Surface Exploration and Expeditions. The pro-posed system is entirely self-sustaining, extracting energy from the planetary boundary layer forboth energy renewal and efficient locomotion. The system consists of three major components:a fleet of rechargeable, internally actuated, buoyancy driven gliders which are programmed tosoar at low altitude; a tethered, high-altitude, oscillating wing whose motion is tuned to extractmaximum energy from the ambient wind; and an attached anchor and base station to inductivelyrecharge the gliders, upload science data, and download revised mission commands. While theproposed system concept is novel in both form and mode of operation, the enabling technolo-gies already exist or are current topics of applied research. Thus, while the proposed systemis innovative and ambitious, it also appears quite feasible. If so, the proposed system couldrevolutionize the way other worlds are explored. Rather than deploy one or a few independentvehicles with short lives, future robotic exploration missions will deploy fleets of rechargeableexplorers along with a distribution of power and communication nodes. The long-term impactwill be that foreign environments, including the Venusian atmosphere, the oceans of Earth, theatmosphere and surface of Titan, and perhaps even the ice-covered oceans of Europa may bethoroughly instrumented and studied by future scientists.

Phase I accomplishments include preliminary sizing and aerodynamic modeling of the low-altitude buoyancy driven glider. Sufficient conditions for dynamic soaring have been identifiedand will be checked against wind data derived from the Cassini-Huygens mission as soon as thosedata become available. A nine degree of freedom vehicle dynamic model has been developedto validate design choices and illustrate feasibility of the concept. The model can also be usedwith numerical trajectory optimization routines to identify efficient dynamic soaring behaviors.Primary technological hurdles include global localization and mapping without a positioningsatellite constellation, packaging and delivery for a fleet of gliders, and materials for cryogenicenvironments.

The current concept for the high-altitude energy-harvester involves a two-stage system in whichthe energy harvester trails behind a highly buoyant float that is attached to a ground-basedanchor and docking station. A dynamic model has also been developed for this component ofthe system. This model can be used to tune vehicle design parameters so that the harvester’snatural oscillations in ambient wind provide forcing for an array of internal inductive chargingdevices. Primary technological hurdles include light-weight conducting tethers, packaging anddelivery, and materials for cryogenic environments.

The anchor and docking station concept is based primarily on the Rosetta Lander, which isscheduled to make contact with Comet 67P/Churyumov-Gerasimenko in November 2014. Thedevice will store energy collected by the harvester and will communicate with Earth scientists,either directly or via an orbiting communications satellite. When docked, vehicles will commu-nicate and be recharged through an inductive coupling device; such devices are already availablecommercially. The docking station is perhaps the least technologically challenging; there are,however, some conventional technological challenges such as light-weight components and high-density energy storage. Although the docking station itself does not appear to present majortechnological hurdles, vehicle guidance and control during the docking process will likely requireactive vision based control, a technique that is currently immature.

The report concludes with recommendations for a follow-on study that would provide a detailedtechnology roadmap.

ii

Contents

1 Introduction and Historical Perspective 1

1.1 Planetary Exploration: Historical Highlights . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.1 Venus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.2 Titan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1.3 Mars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 Visionary Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Mission Concept: Exploring Titan 5

2.1 Atmospheric Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1.1 Atmospheric Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1.2 Meteorology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Surface and Subsurface Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2.1 Surface Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.2.2 Magnetometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.2.3 Gravity Gradiometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.2.4 Surface and Subsurface Chemistry . . . . . . . . . . . . . . . . . . . . . . . . 10

2.3 Special Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.3.1 Methane icing on lifting surfaces . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.3.2 Materials for cryogenic environments . . . . . . . . . . . . . . . . . . . . . . . 11

3 System Description 12

3.1 Overview of the System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.2 Expected Significance and Performance Metrics . . . . . . . . . . . . . . . . . . . . . 14

4 Low Altitude Robotic Soarer 16

4.1 Technology Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

4.2 Concept Development and Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . 20

4.3 Technology Hurdles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

5 High Altitude Wind Energy Absorber 34


5.2 Concept Development and Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . 39

iii

5.2.1 Physics of Oscillating-Wing Energy Absorption . . . . . . . . . . . . . . . . . 39

5.2.2 An “Inverted Two-Stage Tow” . . . . . . . . . . . . . . . . . . . . . . . . . . 42


6 Docking Station 44



7 Extendability to Other Operating Environments 47

7.1 Venus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

7.2 Earth Atmosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

7.3 Earth Hydrosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

7.4 Europa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

8 Conclusions and Recommendations 51

8.1 Phase I Accomplishments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

8.2 Recommendations for Future Concept Development . . . . . . . . . . . . . . . . . . 52

8.2.1 SCALARS Vehicle Concept Development . . . . . . . . . . . . . . . . . . . . 52

8.2.2 Wind Energy Conversion Concept Development . . . . . . . . . . . . . . . . . 52

8.2.3 Docking Station Concept Development . . . . . . . . . . . . . . . . . . . . . . 53

A Operating Environment on Titan 54

iv

List of Figures

1 Photograph of Venus’ surface by Venera 13. (Image credit: NASA/JPL.) . . . . . . . . . 1

2 Photographs of Titan’s surface from the Huygens probe. (Image credit: ESA/NASA/U.Az.) 2

3 The Mars rovers Sojourner and Spirit/Opportunity. (Image credit: NASA/JPL) . . . . . 3

4 Typical structure of a Planetary Boundary Layer, with representative elevations forgiven layers in the atmospheres of Earth and Titan. . . . . . . . . . . . . . . . . . . 6

5 Gradient winds in the upper troposphere of the southern hemisphere of Earth duringwinter. (Image available at http://skywindpower.com/.) . . . . . . . . . . . . . . . . . . . . 7

6 SCALARS following the shore of a fictitious methane lake. (Background by M. Robertson-

Tessi and R. Lorenz [53].) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

7 False color image of Titan from Cassini. (Image credit: NASA/JPL) . . . . . . . . . . . 9

8 Titan’s surface viewed from Huygens. (Image credit: ESA/NASA/U.Az.) . . . . . . . . . 10

9 SCALARS vehicle concept. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

10 Conceptual illustration of the SCALARS ballast actuation system. . . . . . . . . . . 13

11 System components and the PBL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

12 Proposed system architecture, superimposed on artist’s vision of Titan. (Background

by M. Messerotti.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

13 Cartoon representing buoyancy driven gliding flight. . . . . . . . . . . . . . . . . . . 16

14 Three buoyancy driven underwater gliders. . . . . . . . . . . . . . . . . . . . . . . . 17

15 An albatross in flight. (Image by T. Palliser.) . . . . . . . . . . . . . . . . . . . . . . . . 18

16 Frame from a dynamic soaring animation by G. Sachs. . . . . . . . . . . . . . . . . . 19

17 Hull length and diameter versus fineness ratio for m = 10 kg. . . . . . . . . . . . . . 21

18 Vehicle lift coefficient versus angle of attack. . . . . . . . . . . . . . . . . . . . . . . . 24

19 Vehicle drag coefficient versus angle of attack. . . . . . . . . . . . . . . . . . . . . . . 25

20 Glide angle versus angle of attack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

21 Glide angle versus angle of attack: AR = 2 and b = 34L. . . . . . . . . . . . . . . . . . 27

22 Illustration of reference frames and notation for SCALARS vehicle model. . . . . . . 27

23 Angles in the plane of symmetry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

24 Simulated flight path of a maneuvering SCALARS vehicle. . . . . . . . . . . . . . . . 30

25 Simulated flight path of a maneuvering SCALARS vehicle. . . . . . . . . . . . . . . . 31

26 Vehicle motion in a steady laminar shear layer. . . . . . . . . . . . . . . . . . . . . . 32

27 Deployment of Sky WindPower autogiro prototype in Australia. . . . . . . . . . . . . 35

v

28 Stingray oscillating-wing tidal stream energy prototype in the United Kingdom. . . . 36

29 Two variable-lift aerostats: Sky-Doc (left) and Helikite (right). . . . . . . . . . . . . 37

30 Field tests of Sky-Doc (left photo, upper aerostat) and Helikite (left photo, loweraerostat) in towing trials behind a small boat on the Columbia River (right photo). . 38

31 TCOM, LP aerostat family (upper left photo), 15-m long aerial photo and advertisingplatform for sporting event (lower left photo), and 17-m tactical surveillance platformused by military ground forces (right photo). . . . . . . . . . . . . . . . . . . . . . . 38

32 Aerodynamics of an oscillating-airfoil, fixed in surge, but freely moving in pitch andplunge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

33 Time-domain simulation of “Stingray” oscillating-wing. . . . . . . . . . . . . . . . . 41

34 A two-stage towing system for ocean science. . . . . . . . . . . . . . . . . . . . . . . 42

35 A limit cycle oscillation in an unstable two-stage towing system. . . . . . . . . . . . 43

36 The Rosetta Lander. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

37 Venus, as imaged by Galileo. (Image credit: NASA) . . . . . . . . . . . . . . . . . . . . 48

38 Earth, as imaged by Galileo. (Image credit: NASA) . . . . . . . . . . . . . . . . . . . . 49

39 Europa, as imaged by Galileo. (Image credit: NASA) . . . . . . . . . . . . . . . . . . . . 50

40 A deep-sea hydrothermal vent (left) and a colony of tubeworms (right). . . . . . . . 50

41 Atmospheric properties of Titan below 10 kilometers. . . . . . . . . . . . . . . . . . . 55

vi

List of Tables

1 Comparison of specific power required by various flyers (on Earth). . . . . . . . . . . 19

2 Wing parasite drag coefficient CDwpversus Reynolds number and base area. . . . . . 23

3 Planetary Atmospheric Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

vii

Nomenclature

· = cross-product equivalent matrix

0 = column vector of 3 zeros�

= 3 × 3 identity matrix

AR = inboard wing aspect ratio: AR = b2

S

b = inboard wing span (m)

bi = ith basis vector for body frame

c = inboard wing chord length (m)

Cf = hydrodynamic coupling matrix (kg-m)

D = outboard hull diameter (m)

D = drag force (N)

ei = ith basis vector for �3

f = outboard hull fineness ratio: f = DL

f = external force (N)

g = local acceleration due to gravity (m/s2)

h = body angular momentum (kg-m2/s)�= generalized inertia

Ii = ith basis vector for inertial frame

Jb = rigid body inertia (kg-m2)

J f = added inertia (kg-m2)

L = outboard hull length (m)

L = lift force (N)

m = total vehicle mass (kg)

mb = mass of body (kg)

mp = mass of particle (kg)

m = external moment (N m)

M f = added mass matrix (kg)

p = translational momentum (kg-m/s)

rp = relative particle position (m)

rp1= e1 · rp

rcg = body center of gravity (m)

R = body to inertial rotation matrix

S = inboard wing area (m2)

SH = horizontal stabilizer area (m2)

SV = vertical stabilizer area (m2)

T = kinetic energy (J)

vp = particle velocity (m/s)

vp1= e1 · vp

v = body translational velocity (m/s)

V− = total, nominal displaced volume (m3)

X = inertial position of body origin (m)

Xp = inertial position of particle (m)

η = ratio of buoyancy lung capacity to V−

η = generalized velocity

ν = generalized momentum

ω = body angular velocity (rad/s)

Subscripts

b = rigid body

b/f = rigid body/fluid system

b/p = rigid body/mass particle system

f = fluid

p = mass particle

sys = rigid body/fluid/mass particle system

viii

1 Introduction and Historical Perspective

Celestial mechanics inspired Copernicus, Galileo, Newton and a host of other progenitors of modernmathematical physics. And the same planetary bodies that were so intriguing more than fivecenturies ago are even more so today. It seems that with every telescopic image or planetary probe,with every step toward greater understanding, comes a mystery even more profound. Among thegreatest enigmas of our solar system are the bodies with dense atmospheres, such as the planetVenus and Saturn’s moon Titan. With thick atmospheres, these bodies exhibit meteorological andother processes that could inform our understanding of Earth’s own atmospheric science. Moreover,the chemistry of Titan’s atmosphere, surface, and subsurface may provide clues to the pre-bioticprocesses which may have led to the emergence of life on Earth. The same atmospheres whichmake these planets so interesting also obscure the underlying surfaces from high-resolution imagingsystems, making it difficult to target probes for safe or interesting landing sites.

As NASA increases its focus on space exploration, it seems appropriate to reconsider the variousapproaches which have been used or proposed for planetary exploration. We begin by revisitingsome astounding successes, and a few shortcomings, of planetary exploration missions which haveflown in the past.

1.1 Planetary Exploration: Historical Highlights

Rather than consider the complete history of planetary exploration, we focus on a few representativecases that are particularly relevant to the proposed exploration system architecture.1

1.1.1 Venus

Venus was first examined at close range by the American Mariner 2 mission in 1962, a missionwhich verified that the planet is very hot with a cloudy, carbon dioxide atmosphere. The PioneerVenus mission in 1972 provided a map of the surface of Venus and deposited four atmosphericprobes. More recently, the 1989 Magellan orbiter obtained a detailed (300 m resolution) globalmap of Venus using radar, as well as a global map of its gravitational field.

Figure 1: Photograph of Venus’ surface by Venera 13. (Image credit: NASA/JPL.)

In fact, Venus has been a final or intermediate destination for more than twenty spacecraft. Oneof these was the 1970 Soviet probe Venera 7, which became the first spacecraft to land on anotherplanet. A later probe, Venera 9, became the first probe to return an image of another planet.The series of Soviet Venera probes continued to land and send back intriguing images of this alien

1A principal source of information for this section was the web site www.nineplanets.org, an excellent repositoryof information on the planets and their satellites.

1

world, although very limited in scope; see Figure 1. Even by today’s standards, it is an impressivebit of engineering that led to a probe capable of surviving entry and returning an image from thesurface, at 457◦ Celsius and 94 Earth atmospheres.

Despite the flurry of investigative missions over the past four decades, Venus still intrigues scientistson a number of levels. Like Earth, the planet features an iron core and a molten mantle. Magellanidentified volcanic activity, some of it occurring in patterns quite distinct from those seen on earth.While Venus is not believed to suffer the same plate tectonic processes that affect Earth, otherinteresting mechanisms exist to relieve stresses induced in floating crust [2]. To explore disparateand geographically sparse features, such as volcanoes, requires either a shower of Venera-like probesor a self-sustaining system of suitably equipped atmospheric flight vehicles.

1.1.2 Titan

Titan, the largest moon of Saturn, has long been an enigma. The 200 mile thick, smoggy atmospherewhich enshrouds the planet made even the simplest characterization impossible: obtaining anaccurate measurement of its diameter. Indeed, it was not until Voyager 1 imaged the planetin 1981 that this and other basic questions were resolved to the general satisfaction of planetaryscientists. More recently, the scientifically fascinating and visually stunning results from the Cassiniorbiter and the highly successful Huygens probe have made Titan an irresistible target for furtherexploration. As pointed out by R. D. Lorenz in [35], Titan is a cryogenically preserved organicchemistry lab. The surface and subsurface could well host a variety of prebiotic or protobioticcompounds that could inform ongoing scientific debates concerning the emergence of life on Earth.Titan also provides a unique opportunity to improve our understanding of atmospheric mechanicsand geophysics. The major atmospheric component is nitrogen, but with a substantial componentof methane, as well. Interestingly, the temperature near Titan’s surface (94 K) is very near thetriple point for methane, so that this component exists in all three phases. Thus, there are methaneclouds, methane ice and, as supported by recent images from the Huygens probe, methane liquid.Methane plays a similar role in Titan’s atmospheric mechanics and geophysics to the role of wateron Earth.

Figure 2: Photographs of Titan’s surface from the Huygens probe. (Image credit: ESA/NASA/U.Az.)

The left image in Figure 2 shows what appears to be a floodplain, which the Huygens probe revealedto have the consistency of wet clay or sand, with a thin, icy crust. The two cobbles just below themiddle of the image have an estimated width of 15 centimeters (flat one at left) and 4 centimeters(rounded one at center, with deep scour around base), at a viewing distance of about 85 centimeters.

2

The aerial view in the right image shows what appears to be a drainage field leading to a shore line.A major advantage of a buoyancy driven glider over a wheeled or tracked rover will be its ability tosurvey terrain such as that shown in Figure 2 and vertically take off or land to sample the surfacewithout otherwise disturbing this environment, and without getting “mired in the mud” or stuckamong the cobbles. The Huygens landing site is darker than the higher terrain nearby, which iscut by dark, river-like channels. Although water is rock-solid at Titan’s cryogenic temperatures,methane can exist in all three phases and appears to play a similar role in Titan’s weather as waterdoes on Earth. Scientists speculate that photochemical reactions high in Titan’s thick atmosphereproduce a slow drizzle of complex organic “crud” onto the surface, which is periodically washedinto these channels and onto the adjacent floodplain by methane rain. The left image in Figure 2indicates scour around the base of the cobbles, and the cobbles themselves appear to be roundedby erosion, suggesting a highly dynamic liquid environment. As with Venus, it would seem thatTitan’s remaining secrets could best be exposed by a self-sustaining system of suitably equippedatmospheric flight vehicles.

1.1.3 Mars

During the past decade, the standard for planetary exploration appears to have shifted to wheeledrobots. In 1997, the Sojourner rover became the first self-locomoting robot to explore anotherplanet. In three months, the vehicle travelled roughly 100 meters, taking samples of the groundand rock surrounding its immediate landing site. More recently, the Spirit and Opportunity rovershave made continual headlines, exploring a combined 7 kilometers since January 2004, and providingspectacular imagery, including veritable proof that water has existed on the surface of Mars at sometime.

Figure 3: The Mars rovers Sojourner and Spirit/Opportunity. (Image credit: NASA/JPL)

The successes of the Mars rovers are almost too numerous to list, but ground vehicles have lim-itations which are particularly relevant when considering preliminary exploration of planets withdense atmospheres. The most obvious limitations are in range of operation; while 7 kilometers isimpressive, it is small compared with circumference of Mars. Moreover, the rovers were intention-ally targeted toward relatively benign features of the landscape which would pose the least challengefor wheeled locomotion. Tasked with exploring Titan, it is not obvious that one could intentionallytarget such benign terrain, or that one would want to, for that matter. Even on Mars, it is possible,even likely, that features of great scientific value exist in regions inaccessible to wheeled robots.

3

1.2 Visionary Challenges

Recently, NASA outlined a number of visionary challenges in the enterprise strategies of its variousdivisions. Among the near-term challenges (sooner than 2025) pertaining to “Surface Explorationand Expeditions” are (quoting from [1]):

• Mobile Surface Systems. Highly robust, intelligent and long-range Mobile Systems to enablesafe/reliable, affordable and effective human and robotic research, discovery and explorationin lunar, planetary and other venues.

• Flying and Swimming Systems. Highly effective and affordable Flying and Swimming sys-tems to enable ambitious scientific (e.g., remotely operated sub-surface swimmers) and op-erational (e.g., regional overflight) goals to be realized by future human/robotic missions inlunar, planetary and other venues.

• Sustained Surface Exploration & Expeditions Campaign Architectures. Novel and robust ar-chitectures that best enable Sustained Surface Expeditions and Exploration Campaigns tobe undertaken to enable ambitious goals for future human/robotic research, discovery andexploration.

Dramatically improved “Surface Mobility and Access” is a longer-term challenge (after 2025), whichincludes “regional and global mobility in accessible planetary venues, with access at various depthsbelow – and altitudes above – planetary surfaces.” Taken together, these challenges seem to suggestthe development of complete system architectures for autonomous or semi-autonomous exploration,architectures which include high-mobility agents and energy renewal mechanisms. The Office ofSpace Science and the Office of Earth Science also stated visionary challenges which can be ad-dressed, in part, by developing novel systems for autonomous exploration.

4

2 Mission Concept: Exploring Titan

The choice of Titan’s atmosphere as a representative environment for the proposed system waswell-timed. A wealth of information is becoming available from astoundingly successful Huygensdescent probe which landed on Titan’s surface on January 14, 2005. In addition to importantplanetary science data, which are still being prepared for dissemination, there is a great deal ofvaluable engineering design information concerning the Huygens probe.2

In [35], R. D. Lorenz quotes a number of post-Cassini/Huygens exploration imperatives for Titanfrom a report by a NASA-sponsored panel tasked with advising the Solar System ExplorationSubcommittee [10]. The report recommends . . .

. . . the following prioritized order for immediate post-Cassini/Huygens exploration ofTitan: Surface, subsurface, and atmosphere. With this in mind, our suggested prioritiesare to understand the following aspects of Titan:

1. Distribution and composition of organics;

2. Organic chemical processes, their chemical context and energy sources;

3. Prebiological or protobiological chemistry;

4. Geological and geophysical processes and evolution;

5. Atmospheric dynamics and meteorology;

6. Seasonal variations and interactions of the atmosphere and surface.

Each of these tasks would require or greatly benefit from in situ measurements. Moreover, manyof them require some type of locomotion. Rovers can certainly be effective tools for subsurfaceand surface sampling, but the utility of the science data will depend on the sample site, whichmust be selected with little a priori information. The topography revealed by Huygens wouldpose a significant challenge in robot locomotion, making it difficult to explore around the landingsite. A preferable approach would be to map the surface of Titan at low altitude in order to makebetter informed decisions about where to sample at and below the surface. With clever design,the mapping vehicle itself could perform some of the more basic sampling tasks in addition to anyatmospheric sampling tasks of interest.

To image and sample Titan on a large spatial and temporal scale requires a more sophisticatedexploration system than has yet been demonstrated. Ironically, the same feature which protectsTitan’s secrets from external observers, its atmosphere, may also provide the most efficient meansof revealing those secrets using in situ robotic explorers.

The system architecture to be introduced in Section 3 is uniquely adapted to the vertical gradientof dynamic energy in an atmosphere. As an integrated whole, the system harvests wind energy athigh altitudes in the upper region of the Planetary Boundary Layer (PBL), where energy content isgreatest, and expends it at low altitudes, in the Surface Layer of the PBL, where conditions are morequiescent and consequently more favorable for aerial exploration; see Figure 4. Moreover, givensufficient wind-shear strength, the aerial vehicles themselves may repeat this pattern of transferringenergy downward by exploiting the steeper wind speed gradient in the Surface Layer, possibly usingdynamic soaring for efficient low-altitude locomotion.

2See, for example, NASA JPL’s Cassini-Huygens web page (http://saturn.jpl.nasa.gov/home/index.cfm) orESA’s Cassini-Huygens web page (http://www.esa.int/SPECIALS/Cassini-Huygens/).

5

Figure 4: Typical structure of a Planetary Boundary Layer, with representative elevations for givenlayers in the atmospheres of Earth and Titan.

On Earth, the “gradient winds” of the upper troposphere (so called because they are driven byheat fluxes and thermal gradients) have been relatively well mapped for potential use in utility-scaleelectric power generation; see Figure 4. As discussed in Section 5, this has led to the developmentof a variety of technologies for extracting energy from these energy-rich, high-altitude winds.

By comparison, our knowledge of the gradient wind structure on Titan is severely limited. At thetime of this report, researchers were working to determine the wind profile measurements of theDoppler shift in radio signals transmitted from the Huygens descent probe on January 14, 2005;the data have not yet been published. In March 2005, wind speeds were published as measuredfrom Cassini fly-by tracking of discrete mid-latitude clouds, indicating super-rotation of Titan’satmosphere at an estimated rate of 19 to 22 m/s at the equator [50]. Because of the no-slipcondition at Titan’s surface, there will certainly be wind gradients there, although the strength ofthese gradients is still uncertain.

The following sections identify representative science tasks which could or should be incorporatedin a follow-on mission to Titan. Many of these tasks were suggested by R. D. Lorenz in [35].The science tasks define the scientific payload size and power requirements for a future expedition.These requirements, in turn, provide quantifiable boundaries for specific system design parameters,including power generation and storage requirements for OAWEA and speed and maneuverabil-ity requirements for SCALARS. Once these requirements are fixed, specific design optimizationdecisions can be made regarding the size and shape of the vehicle hull, the inboard wing and em-pennage, the variable buoyancy actuator and the moving mass actuators. This Phase I effort doesnot involve detailed system design, however it does identify key design freedoms and constraintsimposed by the type of science missions that are of most interest.

6

Figure 5: Gradient winds in the upper troposphere of the southern hemisphere of Earth duringwinter. (Image available at http://skywindpower.com/.)

2.1 Atmospheric Processes

A fleet of SCALARS vehicles could provide an enormous amount of data regarding atmosphericprocesses, data obtained over very long time scales and very large, even global spatial scales.Two areas which would be of great interest to planetary atmospheric scientists would be Titan’satmospheric chemistry and its meteorology.

2.1.1 Atmospheric Chemistry

One of the more intriguing aspects of Titan’s atmosphere is the fact that methane decomposes inultraviolet light to form a variety of other organic compounds commonly referred to as “smog.”This process takes place on a relatively short time scale and yet Titan’s atmosphere appears to bein a state of equilibrium. Thus, it has been conjectured that there must be a source of methane onor underneath the surface which buffers the atmospheric component [34].

Instruments which could provide insight into Titan’s atmospheric chemistry, and whose size wouldbe compatible with the proposed architecture, might include a miniature mass spectrometer and/orspecialized transducers for detecting the presence and concentration of specific chemical con-stituents. We note that, by using several robotic explorers, mission planners may distribute the

7

Figure 6: SCALARS following the shore of a fictitious methane lake. (Background by M. Robertson-Tessi

and R. Lorenz [53].)

science payload among a number of vehicles in order to satisfy per-vehicle payload limitations.

2.1.2 Meteorology

While the atmospheric chemistry holds no promise for understanding the origins of life (becausethere can be no water in the atmosphere at 94 K), it provides an dramatically different naturallaboratory for testing theories about atmospheric mechanics. On Titan, for example, solar effectson the atmosphere are dominated by tidal effects from the gravitational interaction with Saturn[37]. Local gravity is one-seventh of Earth’s and density is about four times that of Earth’s,at the respective surfaces. Meteorological activity on Titan is expected to be quite “gentle,”with prevailing east-to-west winds at low altitude and occasional, time-varying methane cloudphenomena [35]. The methane clouds may even precipitate, in certain regions and at certain timesof (Saturn) year [34], possibly leading to the fluvial activity that was indicated in photographs fromHuygens; see Figure 2.

Instruments which could provide insight into Titan’s meteorology, and whose size would be com-patible with the proposed architecture, might include a thermistor, a pressure gauge, a methanehumidity sensor, and Doppler velocity profiler. These sensors would also be quite useful for vehiclecontrol.

2.2 Surface and Subsurface Features

As described at the beginning of this section, the topics of most current interest to planetaryscientists interest in Titan have to do with its surface and subsurface. Of course, one could generatea litany of experiments of interest, ranging from light-weight, non-contact experiments (such asvideo imaging) to heavy, deep penetration experiments (such as chemical analysis of core samples).

8

Figure 7: False color image of Titan from Cassini. (Image credit: NASA/JPL)

Science objectives that can be addressed, to some extent, by the proposed architecture includesurface imaging, magnetometry, gravity gradiometry, and surface and subsurface chemistry.

2.2.1 Surface Imaging

The product of planetary exploration missions that is of the most popular interest is undoubtedlyvideo and still imagery. Besides providing accessible and immediate insights into alien worlds,imagery is a invaluable tool for mission planning and execution. The number of fascinating and rarefeatures on our own planet (such as lakes and rivers, volcanoes, and deep-sea hydrothermal vents)underscores the importance of “knowing where to look” when planning science missions. While thisis not such a problem for planets with thin atmospheres, such as Mars, it is an enormous problem forbodies with thick, opaque atmospheres like Titan and Venus. A high-resolution image-based map ofTitan, obtained from a fleet of low-altitude gliders, would be an invaluable resource for determiningand executing specific surface and subsurface sampling tasks in the continuing exploration missionand in any future missions. Other types of imaging, such as radar or spectroscopic imaging, mightcomplement the visual sensors, depending on the camera resolution and lighting.

As the state of the art for terrestrial unmanned vehicles progresses, increasing attention is beinggiven to vision-based vehicle control. It is reasonable to expect that any scientific imaging equipmentwould also be used directly by the SCALARS vehicles while in flight, while landing on the surface,or while docking with the base station.

2.2.2 Magnetometry

Besides providing a heading reference for the SCALARS vehicles, a small and lightweight magne-tometer mounted on each of the SCALARS vehicles would allow an accurate, global map of Titan’s

9

Figure 8: Titan’s surface viewed from Huygens. (Image credit: ESA/NASA/U.Az.)

magnetic field. As Lorenz points out, the 200 mile thickness of Titan’s atmosphere prevents thedevelopment of such a map from space; to remain aloft for a reasonable time, an orbiter wouldhave to be so distant from the planet that small-scale variations in the magnetic field could not bedetected. Such measurements must be made in situ by some type of aerial platform. Moreover,small variations in the magnetic field as Titan orbits Saturn can be used detect subsurface features.For example, it has been speculated that, similar to Earth, Titan’s surface is a layer of crust floatingon a molten layer. Magnetometric fluctuations might support or oppose this hypothesis.

2.2.3 Gravity Gradiometry

A gravity gradiometer would provide another effective tool for detecting subsurface inhomogeneities,such as fluid filled cavities. Again, the thick atmosphere precludes making such measurements fromspace.

2.2.4 Surface and Subsurface Chemistry

According to Lorenz [35], W. R. Thompson and C. Sagan argued that the most interesting chemistryon Titan would occur at sporadic times and locations on or below the surface [66]. Here, “interestingchemistry” means pre- or protobiotic chemistry. Liquid water, a prerequisite for such chemicalreactions, can not exist anywhere near the surface of Titan. Random events such as meteor strikesor “cryovolcanic eruptions,” however, could raise local temperatures enough to allow processes thatultimately lead to the formation of amino acids. Ultimately, the local temperature would return tonormal and the results of this chemistry experiment would be cryogenically preserved.

Instruments for investigating Titan’s surface and subsurface chemistry would be similar to thoseused to investigate its atmospheric chemistry, with the important additional requirement of sampleacquisition system. While a buoyancy driven glider could land and take off from the surface ofTitan, it could only perform simple sample acquisition tasks such as sampling surface liquids orperforming “light duty” drilling analyze solid surface components. On the other hand, the imageryobtained might provide important clues about where to look for these rare sites of interest to

10

biologists.

2.3 Special Challenges

It may seem wry to title a section “Special Challenges” when discussing buoyancy driven glidingflight on Titan, however many aspects of the proposed vehicle design are quite conventional. Twoof the distinguishing issues are the possibility of methane icing on the wings of an aerial vehicleand the the problem of finding materials that function well in cryogenic environments.

2.3.1 Methane icing on lifting surfaces

The possibility of methane icing has been considered in [35]. It was suggested that, according to thenominal temperature profile, icing could be a problem above 14 km altitude. Seasonal variationsin temperature and methane humidity could lower the threshold to as little as 10 km, placing oneupper bound on the operational altitude of the SCALARS and OAWEA vehicles. Currently, nocomponent of this system is expected to operate near or above that altitude except possibly duringentry.

2.3.2 Materials for cryogenic environments

The success of the Huygens probe illustrated that challenges such as choice of materials can bemet, at least for short durations. There are, however, peculiar aspects of the proposed systemwhich suggest special challenges. First and most obvious is the intended duration of the explorationsystem. Common engineering materials become brittle at low temperature, potentially exacerbatingany tendency toward cyclic fatigue failure.

In addition, the current SCALARS and OAWEA vehicle designs call for bladders that can beinflated or deflated from an internal assembly of pressure vessels and pumps. Indeed, packagingand delivery issues may indicate that the vehicle itself should be inflatable on arrival. Both thechoice of working fluid and the bladder material are questions that should be resolved sooner ratherthan later. From Archimedes’ principle, the buoyant force is the product of displaced volume andthe ambient fluid density. For underwater gliders, the buoyancy fluid is typically incompressible(liquid), however operation on Titan would probably require a gaseous buoyancy fluid. Obviouschoices of lifting gas are the typical ones: hydrogen or helium. Related to the choice of lifting gasis the design of the pump which will expel gas from a reservoir at lower-than-ambient pressure orreturn gas to a reservoir at higher-than-ambient pressure (depending on which mode offers betterefficiency). Questions regarding the choice and design of ballast pumps have been considered bydesigners of underwater gliders, although only with regard to liquid buoyancy fluids and hightemperature operation (relative to Titan, at least). As for the choice of materials appropriate forcryogenic environments, NASA and supporting industries have a great deal of experience in thisarea. For example, flexible materials for sealing cryogenic pressure vessels are discussed in [18].Another potentially useful material is microwave radome fabric, which is rated for use at cryogenictemperatures.3

3See http://www.gore.com, for example.

11

3 System Description

This report describes conceptual design of a revolutionary system for remote terrain explorationand environmental sampling on worlds with dense atmospheres, such as Titan and Venus. Theproposed system addresses several visionary challenges enumerated by NASA’s Office of SpaceFlight, as described in Section 1.2.

While recent successes in the Mars exploration program have illustrated the value of autonomousland vehicles for planetary exploration, they have also shown the need for more advanced au-tonomous sensor platforms with longer range and safe access to steep or rugged terrain. The 1997Sojourner rover travelled roughly 100 meters over a three month period. Combined, Spirit andOpportunity rovers have travelled almost 7 kilometers over a fifteen month period. (To be fair, theyare still going strong at the time of this report!) Atmospheric flight vehicles can provide greaterarea coverage than ground vehicles, but the benefits of airborne sensors diminish with altitude andspeed. We propose a self-sustaining system of low speed, low altitude flight vehicles which couldbe used to identify and visually characterize features of interest to scientists.

Limitations on energy storage restrict the range of conventionally powered vehicles, but self-contained energy harvesting devices (such as solar cells) increase the cost, weight, and complexityof individual vehicles. Added weight is a particularly serious impediment for atmospheric flightvehicles and added complexity generally reduces vehicle reliability. Moreover, in many of the mostinteresting environments, conventional energy regeneration technologies like solar cells are of ques-tionable use. Titan, for example, is more than nine astronomical units from the sun. The relativelysmall amount of radiant solar energy that reaches Titan is further diminished by the dense, smoggyatmosphere.

3.1 Overview of the System Architecture

The proposed system for planetary exploration is entirely self-sustaining, extracting energy fromthe planetary boundary layer for both energy renewal and efficient locomotion. The system consistsof three major components:

1. A fleet of rechargeable, autonomous, buoyancy driven gliders, actuated solely through internalshape control. These twin-hull, inboard wing gliders will exploit the environmental dynamicsby performing static and dynamic soaring, as appropriate. We refer to the proposed vehiclesas Shape Change Actuated, Low Altitude Robotic Soarers (SCALARS).

2. A tethered, high-altitude, Oscillating-Aerofoil Wind Energy Absorber (OAWEA), similar inshape and structure to the SCALARS vehicles, but de-tuned to oscillate in the ambient wind.The vehicle’s natural oscillation drives a bank of on-board linear inductive generators. Aswith the SCALARS vehicle, the OAWEA carries no external moving parts.

3. A fixed anchor and docking station to which the OAWEA is tethered. The docking stationstores energy generated by the OAWEA for use by the SCALARS vehicles. It also servesas a communication node, via an orbiting satellite, between the SCALARS vehicles and anEarth-based science station.

The proposed system is uniquely adapted to exploit the vertical gradient in atmospheric energywithin a PBL. Wind energy is harvested at high altitudes, in the mixed region of the PBL, where

12

the energy content is greatest. The harvested energy is transmitted to a ground-based storage nodefor transfer to vehicles operating in the low-altitude surface layer of the PBL, where conditions aremore quiescent. Further, the SCALARS vehicles exploit low-altitude gradients in horizontal windspeed through dynamic soaring, thus improving locomotive efficiency.

Figure 9: SCALARS vehicle concept.

The SCALARS vehicle features an inboard-wing, twin-hull configuration. Each outboard hullcontains a “free-flooded” compartment in which one or more buoyancy bladders may be inflatedusing a liquid or gas that is otherwise stored within a rigid vessel. Each hull also contains a linearmoving mass actuator aligned with its longitudinal axis. By inflating the bladders and moving thetwo masses aft in a symmetric way, the vehicle will become more buoyant and pitch up. If properlytrimmed, it will rise and glide forward. Deflating the bladders and moving the masses forwardcauses the vehicle to sink while continuing to glide forward. The inboard-wing is elastic so thatasymmetric movements of the mass actuators will induce twist, providing a passive morphing wingfor roll control.

Central pressure housing(actuators, sensors, computer)

Buoyancy ballonets(in flooded section)

Ballast reservoirand pump

Nose cone(sensors)

Tail cone(sensors)

Figure 10: Conceptual illustration of the SCALARS ballast actuation system.

The OAWEA structure is similar to that of the SCALARS vehicles, however the OAWEA is tetheredto a highly buoyant float. The float, in turn, is tethered to a fixed anchor which serves as a dockingstation for the SCALARS vehicles. The light-weight conducting tether passes energy generated bythe OAWEA to a bank of storage cells within the docking station.

The docking station communicates with docked SCALARS vehicles through an inductive couplingdevice, which also serves to recharge the gliders. The station relays science data collected by thegliders to Earth and relays new mission commands from Earth to the gliders. The inductive couplingdevice maintains a consistent theme in which there are no external moving parts or electricalconnections.

13

Figure 11: System components and the PBL.

This report outlines the basic physics of the proposed concept and identifies the major engineeringfeasibility issues, using the atmosphere of Titan as a representative operating environment. Thereport also discusses the adaptability of this system to denser environments such as Titan’s liquidhydrocarbon lakes, the Venusian atmosphere, the atmosphere and oceans of Earth and, potentially,the ice-covered waters of Europa. It has been suggested that buoyancy driven gliders might alsobe an effective means for exploring the interior of the gas giants, although we do not consider thatscenario here. While the proposed system concept is novel in both form and mode of operation, theenabling technologies already exist or are current topics of applied research. This report provides aliterature survey representing the current state of the art in several critical enabling technologies.

3.2 Expected Significance and Performance Metrics

The goal of this Phase I effort is to provide a preliminary evaluation of an innovative autonomoussystem for exploring other worlds. The evaluation focuses on the viability of three major com-ponents: the SCALARS vehicle concept, the OAWEA wind-energy harvesting concept, and thedocking mechanism for recharging and exchanging information.

The concepts of buoyancy driven, gliding flight, internal shape control, and autonomous dynamicsoaring are current topics of applied research. Combining them within a twin-hull, inboard wingairframe, however, provides a truly revolutionary design for a terrain-following, low-speed vehicle,propelled entirely by internal actuation and the natural atmospheric gradients of the lower PBL.Likewise, using a similar airframe and internal actuation technologies in combination with linearinduction generators enables a revolutionary device for harvesting wind energy from the upperPBL. Taken together, these architectural elements could shift the paradigm by which other worldsare explored and our own world is monitored. Rather than deploy one or a few highly capable butshort-lived vehicles, operating independently of one another over a short range of time and space,future planetary exploration and Earth-observing missions will deploy fleets of low-cost, recharge-able explorers along with a distributed network of power and communication nodes energized bythe surrounding environment. Besides relieving the individual vehicles of cumbersome power andcommunication hardware, and thus expanding their range, such a system architecture has the addi-

14

tional advantage of being scalable and extendable. Future robotic and human exploration missionscan make use of an existing infrastructure for communication and power as well as for localizationand mapping.

Figure 12: Proposed system architecture, superimposed on artist’s vision of Titan. (Background by

M. Messerotti.)

The impact of this paradigm shift will be that remote environments, including the surface andlower atmosphere of Venus, the oceans of Earth, and the lower atmosphere, continental terrain andhydrocarbon lakes of Titan can be thoroughly instrumented and studied. Moreover, each missionwill truly build on previous missions, rather than start from scratch at each mission landing site oroceanographic sampling station. Such a sustainably energized system provides for persistent, evenperpetual, scientific presence between missions.

15

4 Low Altitude Robotic Soarer

SCALARS have no external moving parts, other than those required by sensors in the sciencepayload. The vehicles are thus impervious to corrosion, dust, and other processes or elementswhich might otherwise impair their operation. The SCALARS vehicle is configured as a twin-hulled, buoyancy driven glider with an elastic inboard wing. The vehicle changes its net weight(i.e., weight minus buoyancy) by inflating or deflating a “lung” which is exposed to the ambientfluid. This buoyancy lung may be an elastic bladder contained in a flooded volume within one ofthe outboard hulls or it may be a piston-cylinder mechanism, with an orifice to the ambient fluid.More ambitiously, the lung may be the inboard wing itself, which may need to be inflatable forpackaging and deployment reasons.

®

°

-µV

W

B

M

+

L

D

Figure 13: Cartoon representing buoyancy driven gliding flight.

Longitudinal control is provided by two moving mass actuators, each aligned with the long axis ofan outboard hull. For pure longitudinal flight, the vehicle would deflate its lung, thus becomingheavier than the ambient fluid. By displacing the two masses forward (symmetrically) from theirnominal positions, the vehicle would nose down as it descends. The lift generated over the wingwould propel the vehicle forward as it glides down. At a given altitude, the lung would again beinflated, making the vehicle lighter than the ambient fluid. By displacing the two masses backward,the vehicle would nose up as it ascends and the lift generated by the inboard wing would continueto propel the vehicle forward as it glides up. In this way, vertical motion due to alternate deflationand inflation of the buoyancy lung is rectified into horizontal translation at low energetic cost.

Lateral-directional control is provided by twist in the inboard wing, which is induced by asymmet-ric displacements of the moving mass actuators. To perform a downward turn to starboard, forexample, the vehicle would deflate its ballonets to become heavy, move its internal masses forwardto pitch the nose down, and move the starboard mass slightly farther forward to twist the wing sothat port side generates more lift. An analogous maneuver could be performed while climbing.

The state of component technologies contributing to the SCALARS vehicles is reviewed in Sec-tion 4.1. A nine degree of freedom dynamic model is developed in Section 4.2. Preliminary sim-ulations demonstrate stability in steady flight and basic controllability; a formal stability andcontrollability analysis would provide considerably more insight with regard to design, howeversuch detailed analysis was beyond the scope of the Phase I effort. Potential technology hurdles areidentified in Section 4.3.

4.1 Technology Review

Buoyancy-driven gliding flight. Because battery power is the essential limitation for au-tonomous underwater vehicles (AUVs), and because conventional marine propulsion is sorely inef-

16

ficient, underwater gliders have begun to play a crucial role in long-term oceanographic monitoringon Earth [13]. Conventional propeller-driven vehicles can operate on the order of one or a fewdays before their power is depleted. Underwater gliders, on the other hand, have proven to beeffective for long-range, long-term oceanographic sampling. In autumn 2004, for example, a Sprayunderwater glider [63] travelled autonomously from just south of Nantucket to Bermuda becoming,in the process, the first autonomous vehicle to cross the Gulf Stream [68]. Almost simultaneously,a Seaglider crossed the Kurohio current in the East China Sea, a highly energetic current much likethe Gulf Stream. It also survived two typhoons that passed directly overhead during its deploy-ment, maintaining Iridium communications throughout the exercise [38]. The underwater gliderSlocum, developed by Webb Research Corp., is designed to travel 40,000 km over a period of 5 yearswithout service [70]; if realized, this technology would provide a several order of magnitude improve-ment over conventional autonomous underwater vehicles. A thorough review of existing buoyancydriven glider technology is given in[24], along with design guidelines and useful comparisons toother engineered flight vehicles and biological fliers.

Seaglider (UW-APL)

Spray (Scripps Institute)

Slocum (Webb Research)

Figure 14: Three buoyancy driven underwater gliders.

Key to the long endurance of underwater gliders is their reliance on internal actuators (buoyancybladders and moving masses) for propulsion and control. With no exposed moving parts, thesevehicles have a somewhat magical allure. While it is certainly not magic, control of these systemsis challenging. Classical input/output-based control design is of limited use because low-speedhydrodynamic effects make the vehicle dynamics nonlinear, at least in some portions of the per-formance envelope. Model-based control design is also challenging because the dynamic modelsfor underwater gliders are high-dimensional and the vehicles are intrinsically under-actuated (i.e.,have fewer inputs than degrees of freedom). While considerable effort has been spent developingfieldable underwater gliders for oceanographic sampling missions, comparatively little effort hasgone into fundamental analysis of the dynamics and control of buoyancy driven gliders; see [20]and [32] for some preliminary efforts. If revolutionary operational concepts, such as autonomousdynamic soaring on Titan, are ever to be realized, more effort must be focused on dynamic analysisand control design for buoyancy driven gliders.

17

Buoyancy driven gliding flight has also been proposed for use in air, for efficient transportation ofpeople and cargo.4 Of course, questions about vehicle scaling arise, but the basic principle (i.e.,Archimedes’ principle) is quite sound.

Aerodynamic shape control. The SCALARS vehicle features an inboard-wing, twin-hull con-figuration. The arrangement protects the wing and reduces the length of the vehicle for a givenbuoyancy requirement [64]. The elastic inboard wing, which twists under asymmetric displace-ments of the moving mass actuators, allows roll control without external moving parts that couldbe fouled by corrosion, ice, dust, or other elements. Internal actuation also provides a sterile meansof investigating an environment, leaving little or no environmental footprint.

The use of passive wing deformations for control places SCALARS in the category of morphingwing aircraft. Morphing wing aircraft have attracted considerable attention in the last decade.The advent of “smart materials” heralded a new age of aircraft actuator design in which the liftingsurface is continuously deformed to provide optimum performance. While obstacles to this grandvision remain, the feasibility of the concept was famously demonstrated by the Wright brothersover 100 years ago. Similar to the Wrights’ “wing warping” mechanism for roll control, SCALARS’wing is twisted to provide a roll moment in a uniform flow. Wing morphing technologies canbe categorized as “discrete” and “continuous,” depending on whether the number of parametersrequired to describe the shape change is finite or infinite. Discrete wing morphing approachesinclude conventional hinged surface actuators as well as variable sweep or variable span wings.Continuous wing morphing approaches include conformal control surfaces. See [14] and [58] andreferences therein.

Autonomous Dynamic Soaring. The ability of birds to locomote efficiently by exploiting anambient flow has long been admired by people. Sailplane pilots compete with one another to soarlonger and farther without propulsion, but their abilities pale compared to those of the wanderingalbatross [62] and other soaring birds. We propose to make similar use of the environmentaldynamics on Titan to extend the range of SCALARS.

Figure 15: An albatross in flight. (Image by T. Palliser.)

Soaring flight generally falls into two categories: static and dynamic soaring [69]. In static soaring,a bird or a vehicle exploits a local upward movement of air, for example, wind being deflectedupward by a geographic feature. This category includes “slope soaring” (as just described) and“thermal soaring,” in which a warm column of air rises through cooler surrounding air. See [69] foran overview of different soaring behaviors.

4See http://www.fuellessflight.com/.

18

In dynamic soaring, a bird or vehicle exploits a vertical gradient in a horizontal flow field, such asthe boundary layer between the ground or water and a moving body of air. Briefly, a bird or vehiclestarts from a higher elevation and glides downward with the wind. As its elevation decreases, thebird’s air-relative speed increases because the lower air moves more slowly. Turning into the wind,the bird glides upward, trading its excess kinetic energy for altitude, and the cycle repeats. SeeFigure 16.5

Table 1: Comparison of specific power required by various flyers (on Earth).

Type of Flight Vehicle Specific Power(Watts per kg)

Wandering albatross (with a mass of 10 kg at a ground speed of 11.7 m/s) 5.1Lighter-than-air vehicle (at albatross mass and speed) 112Fixed wing aircraft (at albatross mass and speed) 63Coaxial helicopter (at albatross mass and speed); see [73] 75

This efficient use of wind energy gradients near the sea surface enables the wandering albatrossto travel vast distances at a fraction of the power cost that would be incurred by similarly sizedman-made vehicles traveling at the same speed. Table 1 compares the power cost of dynamicsoaring by the wandering albatross with the estimated power cost for conventional flight vehiclesof the same mass travelling at the same speed. These estimates are based on scaling relationshipsdeveloped in [36] for heavier-than-air (HTA) fixed-wing airplanes, lighter-than-air (LTA) vehicles,and HTA rotary-wing vehicles. If SCALARS can perform with an energy efficiency comparable tothe wandering albatross, its power needs should be one to two orders of magnitude less than anyconventional micro-air vehicles with a comparable sensor suite (mass) and survey (speed) capability.

Figure 16: Frame from a dynamic soaring animation by G. Sachs.

Dynamic soaring has been demonstrated using a remotely piloted radio control aircraft, as describedin [6]. Making this behavior autonomous, however, requires further progress in sensor technology

5The animation is available at Prof. G. Sachs’ web site: http://www.lfm.mw.tum.de/lfm sources/albatros.html.

19

and control theory. A vehicle must be capable first of sensing a wind gradient either directly, usinga Doppler velocity sensor, for example, or indirectly, using a model-based state observer. Next, thevehicle must be capable of responding adequately to the sensed gradient, that is, of performing thenecessary maneuvers to accomplish dynamic soaring.

While we do envision the use of dynamic soaring by SCALARS, appropriately adapted to the lowerPBL profile of the given environment, the vehicles would not be handicapped by an inability tosoar. Like soaring birds, SCALARS have an alternative propulsion mechanism; birds may flap theirwings, if necessary, and SCALARS may adjust their buoyancy. While the environmental dynamicsgenerally dictate the direction of locomotion for a vehicle which can only soar, SCALARS can movein any direction, provided the buoyant lift capability is sufficient.

4.2 Concept Development and Preliminary Analysis

This section describes preliminary development of the SCALARS vehicle concept, including prelim-inary sizing and vehicle dynamic modeling. The methods used here are intentionally simplistic andsome design assumptions are based solely on physical intuition. The goal, at this point, is to obtaina general sense of the vehicle size and performance capabilities and to provide a starting point forfuture design iterations. The accuracy of the aerodynamic property estimates, in particular, can beimproved significantly by using a suitable vortex lattice method (VLM) code. Development of sucha tool, however, is beyond the scope of this Phase I effort. (Commonly available VLM codes arenot applicable to the twin-hulled, inboard wing configuration.) In any case, the gains in accuracyfrom using such a tool would not dramatically affect the order-of-magnitude size estimates we areseeking at this stage.

In [36], R. D. Lorenz argues for a “small (20-100 kg) unmanned aerial vehicle (UAV)” as a follow-onto the Huygens mission. He goes on to claim that “A reasonable flight speed requirement wouldbe 1 m/s, giving the ability to traverse pole to pole twice in one year (and thus, because thereare east-west winds that are strong at altitude, access to anywhere on the surface).” Rather thandeploy a single, highly capable UAV, our concept is to deploy a fleet of autonomous vehicles which,taken individually, may be less capable but which provide even greater access and science capabilityas a group.

As a starting point for sizing calculations, we assume that each SCALARS vehicle has a massof m = 10 kg. Such a small mass certainly presumes a number of advancements in materials,computation, and sensing. Moreover, because the science payload per vehicle would be so limited,it may be necessary to distribute various sensors among the fleet, with some redundancy to ensurethat loss of a single vehicle does not end a particular sampling capability.

We will assume that the outboard hull is a prolate spheroid. A more aerodynamic shape wouldundoubtedly improve efficiency. As far as sizing is concerned, however, the geometric properties ofsuch a shape would scale in direct proportion to those of a prolate spheroid.

To displace m = 10 kg of Titan’s atmosphere at a density ρ = 5.26 kg/m3, a vehicle must displace

V− =m

ρ= 1.9 m3.

Given two prolate spheroidal outboard hulls, each of fineness ratio f , and neglecting the wing’s

20

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

1

2

3

4

5

6

Fineness Ratio

L (m)D (m)

Figure 17: Hull length and diameter versus fineness ratio for m = 10 kg.

contribution to the total volume, one finds that

V− = 2

(

4

3π

(

L

2

) (

fL

2

)2)

=π

3f2L3.

Thus, the required hull length for a neutrally buoyant 10 kg SCALARS vehicle is

L = 3

√

3m

πρf2.

Figure 17 shows the hull length and diameter necessary, over a range of fineness ratios. A 4 : 1prolate spheroidal hull (f = 1

4), which provides a moderately streamlined shape, would be just over

3 meters in length with a maximum hull diameter of about 34

meters. The same vehicle in Earth’satmosphere would be almost twice as long.

The displaced volume required for buoyant flight on Titan may seem large, even in comparisonto airships on Earth. Issues of structure, materials, and packaging and delivery certainly requirecareful consideration. It is likely, for example, that inflatable structures will play an importantrole in the proposed architecture. With regard to the flight characteristics, however, there are twofactors which make buoyant flight on Titan quite appealing. First, the winds at low altitude areexpected to be small, on the order of one or two tenths of a meter per second [35]. Because dynamicpressure scales quadratically with speed, but only linearly with density, the effects of low altitudewind disturbances would be quite mild. Second, the gravitational acceleration on Titan’s surfaceis one-seventh that on Earth. The structural loads on the aircraft would therefore be much lowerthan they would be on Earth.

21

A parameter of primary importance in sizing the vehicle is the buoyancy lung capacity η. Thebuoyancy lung capacity determines the allowable change in net weight (positive or negative) andis primarily driven by the required nominal speed. A nominal speed of U = 0.5 m/s would enablethe SCALARS vehicle to make upwind progress against expected headwinds, based on estimatesof wind speeds at low altitudes [35]. Higher nominal speeds require larger buoyancy lungs; thepreliminary sizing analysis described below suggested that Lorenz’ original speed goal of 1 m/swould require a prohibitively large lung.

We note that Lorenz’ choice of nominal speed was based on the assumption that a single vehiclewould perform an entire mission. Given that the proposed architecture calls for multiple vehicles,and that the gust disturbances are expected to be small, it seems reasonable to relax the targetflight speed to 0.5 m/s. To determine the buoyancy lung capacity required to achieve an averagespeed of 0.5 m/s, one must investigate the vehicle aerodynamics.

The nature of the flow over the twin hulls and inboard wing is determined by the Reynolds number

Re =ρUl

µ

where l is a characteristic length. To ensure that the flow over the inboard wing remains fullyturbulent, we must choose the chord length c such that, with l = c, we have Re > Rec = 105.Substituting given values, we require

c >µRec

ρU=

(8.051 × 10−6 Pas)(105)

(5.26 kg/m3)(0.5 m/s)= 0.30 m. (1)

The chord length is also influenced by other concerns, such as the desired torsional stiffness; equa-tion (1) simply provides a lower bound.

Wing contributions. The aerodynamic terms affecting longitudinal motion are lift, drag, andpitching moment. We first consider the effect of the inboard wing, which is the primary generatorof aerodynamic lift. Because the vehicle is designed for both downward and upward gliding, itis reasonable to assume a symmetric airfoil.6 A symmetric wing generates no pitching momentwhen generating no lift. Thus, there will be no pitching moment about the aerodynamic center, animaginary point (about one-quarter chord length aft of the leading edge) about which the pitchingmoment does not vary with angle of attack. For simplicity, we assume that the lift and drag forcesact at this point.

The lift generated by the wing is

Lw = CLw

(

1

2ρV 2

)

S

where ρ is the local fluid density, V is the fluid-relative speed, and S is the wing area. For smallangles of attack, the lift coefficient takes the form

CLw= CLαw

α.

The lift-curve slope can be roughly approximated using the Helmbold equation [65]

CLαw=

πAR

1 +

√

1 +(

AR2

)2. (2)

6This assumption may be revisited, if later mission plans suggest that the vehicle should be more efficient indownward glides than in upward ones.

22

In the limit AR → ∞, the slope of the lift curve approaches 2π, as predicted by aerodynamic theory.

At large angles of attack, one would expect the lift to reach some maximum value and to thendecrease toward zero. A crude approximation which captures both the small and large anglefeatures is

CLw=

1

2CLαw

sin 2α.

This approximation gives a stall lift coefficient of 12CLαw

.

The drag force acting on the wing is

Dw = CDw

(

1

2ρV 2

)

S

where the drag coefficient CDwdecomposes into the so-called parasite and induced drag coefficients:

CDw= CDwp

+ CDwi

The term CDwpaccounts for skin friction and is largely a function of Reynolds number and wetted

area. An empirical approximation is [65]

CDwp=

0.135

3

√

CfSwwet

Swbase

Swbase

S(3)

where, for a smooth flat plate in a fully turbulent flow,

Cf = 0.455(log10(Re))−2.58.

Here, the area Swbaserepresents the area of a blunt trailing edge or the cross-sectional area of a

separated flow region. Since we are using a flat plate model for Cf , we may as well assume thatSwwet = 2S. Table 2 shows the parasite drag coefficient for a range of Reynolds numbers and baseareas.

Table 2: Wing parasite drag coefficient CDwpversus Reynolds number and base area.

Rew Swbase/S = 0.01 Swbase

/S = 0.05 Swbase/S = 0.10

105 0.0056 0.028 0.056106 0.0065 0.033 0.065107 0.0074 0.037 0.074

The term CDwiaccounts for drag due to the generation of lift. Small angle aerodynamic theory

suggests that, for a wing with an elliptical lift distribution,

CDwi=

C2Lw

πAR.

Using this expression, we obtain

CDw= CDwp

+C2Lw

πAR.

Clearly, this approximation is entirely unsuitable at large angles of attack. The fact that CLw= 0

when α = 90◦ suggests that the induced drag vanishes; however, one would expect a much larger

23

drag coefficient than CDw= CDwp

when α = 90◦. A more satisfying approximation for the dragcoefficient, which accurately captures the small angle relations, is

CDwi= CDwp

+C2Lw

2πAR(1 − cos 2α) .

Fuselage contributions. In preliminary sizing, we will ignore the contribution of the twin out-board hulls to vehicle lift and only consider their contribution to drag. With regard to pitchmoment, the largest effect of the hulls is captured by potential flow theory. This effect will beexpressed elsewhere in the dynamic model; we neglect the viscous component of pitch moment dueto the outboard hulls. A worst-case drag coefficient for one of the outboard hulls is that for ablunt-based body [65]:

CDhp=

0.029√

CfShwet

Shbase

Shbase

S.

The wetted area of a spheroidal hull is

Shwet= 2π

(

L

2

)2{

f2 +f

√

1 − f2arcsin

√

1 − f2

}

.

As a conservative assumption, suppose that a line of separation occurs at the maximum diameterof the outboard hull so that

Shbase= πf2

(

L

2

)2

.

The parasite drag for the combined hulls is then

2CDhp=

0.058π

4S(fL)2

[

2Cf

{

1 +1

f√

1 − f2arcsin

√

1 − f2

}]−12

(4)

0 10 20 30 40 50 60 70 80 900

0.5

1

1.5

2

2.5

3

α (deg)

CL

AR = 2AR = 4AR = 6AR = 8

Figure 18: Vehicle lift coefficient versus angle of attack.

24

Ignoring, for now, the effect of the empennage (the horizontal and vertical stabilizers), we maywrite for the whole vehicle:

CL = CLαα (5)

CD =(

CDwp+ 2CDhp

)

+C2

Lα

2πAR(1 − cos 2α) (6)

where CLα= CLαw

is given by (2), CDwpis given by (3), and 2CDhp

is given by (4).

0 20 40 60 800

0.5

1

1.5

2

α (deg)

CD

b = 1*(L/4)

0 20 40 60 800

0.5

1

1.5

2

α (deg)C

D

b = 2*(L/4)

0 20 40 60 800

0.5

1

1.5

2

α (deg)

CD

b = 3*(L/4)

0 20 40 60 800

0.5

1

1.5

2

α (deg)

CD

b = 4*(L/4)

AR = 2AR = 4AR = 6AR = 8

Figure 19: Vehicle drag coefficient versus angle of attack.

Figure 18 shows lift coefficient curves for various values of the wing aspect ratio AR. Figure 19shows corresponding drag coefficient curves, as well as the variation of drag coefficient with wingspan. For these computations, a speed of 0.5 meters per second was assumed; the correspondingwing Reynolds number varied with chord length, as determined by the aspect ratio and wing span.

The comparatively large drag coefficient for the smallest wing span is an apparent paradox; inreality, though, it is only a consequence of a smaller wing reference area. More informative thanthese plots is Figure 20, which shows the glide path angle for the various cases considered. Noticethat, as one would expect, the glide angle is shallower for larger aspect ratios and for larger wingspans. Moreover, the curves become more flat in the region of minimum glide angle for larger wingspans and aspect ratios. We note that glide path angle is solely a function of the lift-to-drag ratio.

Representative example. In order to obtain an estimate of the buoyancy lung volume required,we choose b = 3

4L and AR = 2 based, in part, on the preceding parametric study. At a speed

of U = 0.5 meters per second, the Reynolds number based on chord length is Re = 3.8 × 105.Assuming that Swbase

= 0.05S, we find

CL = 1.30 sin 2α and CD = 0.069 + 0.54 (1 − cos 2α) .

25

0 20 40 60 800

20

40

60

80

α (deg)

γ (d

eg)

b = 1*(L/4)

0 20 40 60 800

20

40

60

80

α (deg)

γ (d

eg)

b = 2*(L/4)

0 20 40 60 800

20

40

60

80

α (deg)

γ (d

eg)

b = 3*(L/4)

0 20 40 60 800

20

40

60

80

α (deg)

γ (d

eg)

b = 4*(L/4)

AR = 2AR = 4AR = 6AR = 8

Figure 20: Glide angle versus angle of attack.

Figure 21 shows the glide angle versus angle of attack for this choice of parameters. The minimumglide angle is γ = 12◦, occurring at an angle of attack of roughly α = 14◦.

Assuming that the SCALARS vehicle’s buoyancy is trimmed such that the vehicle is neutrallybuoyant when the buoyancy lung is half inflated, the buoyancy lung capacity is given by theexpression [24]

η =ρCD

V− sin γ

(

U

cos γ

)2

. (7)

If the given vehicle operates at its most efficient glide path angle, then we must have

η =(5.26 kg/m3)(0.13)

(1.9 m3)(0.21)

(

0.5 m/s

0.98

)2

= 0.45.

While a 45% volume change is ambitious, it is certainly physically achievable.

Having determined that the proposed vehicle configuration is feasible, several tasks remain:

• Develop a higher fidelity aerodynamic model.

• Derive the necessary wing geometry from performance requirements (as opposed to specifyingthe geometry a priori).

• Quantify the stability and control requirements and size the stabilizers and moving massactuators appropriately.

26

0 10 20 30 40 50 60 70 80 900

10

20

30

40

50

60

70

80

90

α (deg)

γ (d

eg)

Figure 21: Glide angle versus angle of attack: AR = 2 and b = 34L.

• Repeat the above analysis to determine the required buoyancy lung capacity with more pre-cision.

Some of these tasks are addressed in the discussion to follow while others are left for a future effort.

Vehicle dynamics. The dynamic model for a rigid SCALARS vehicle can be determined byextending the results of [71] to include two moving masses. The effect of wing twist can be incor-porated as a slight modification.

b1

I1

2

3

I

I

2

3

rr

p

b

b

rl

p

Figure 22: Illustration of reference frames and notation for SCALARS vehicle model.

Figure 22 is crude depiction of a SCALARS vehicle. A reference frame with orthonormal basis(I1, I2, I3) is fixed in inertial space.7 The other reference frame, with orthonormal basis (b1, b2, b3),is fixed at some reference point for the vehicle; b1 defines the longitudinal axis and b3 points outthe “belly” of the vehicle while b2 completes the orthonormal triad. A proper rotation matrix R

7At this point in the model development, we assume an inertially fixed reference frame rather than a frame fixedin the planet being explored.

27

transforms free vectors from the body reference frame to the inertial reference frame. The inertialvector X locates the origin of the body frame with respect to the inertial frame.

Let the body frame vectors ω and v represent the angular and linear velocity of the body withrespect to inertial space, respectively; both are expressed in the body frame. Next, define theoperator · by requiring that ab = a × b for vectors a, b ∈ �3. In matrix form,

a =

0 −a3 a2

a3 0 −a1

−a2 a1 0

.

The kinematic equations for the body are therefore:

R = Rω (8)

X = Rv (9)

Let the body vector vpldenote the velocity of the left-side moving mass particle with respect to

inertial space. The kinematic equation for the mass particle is

Xpl= Rvpl

. (10)

Now let the body vector rpldenote the position of the mass particle relative to the origin of the

body frame. An equivalent kinematic equation for the mass particle is

rpl= −ω × rpl

+ vpl− v. (11)

Rearranging (11) gives the familiar expression for the velocity vplof a point expressed in a rotating

reference frame. Because the point mass is constrained to move parallel to the vehicle’s longitudinalaxis, only the first component of the vector rpl

is free to vary under the influence of a control force.

We therefore define rpl= e1 · rpl

where e1 = [1, 0, 0]T . The kinematic model for the right-sidemoving point mass can be developed similarly (substituting “r” for “l” in the subscripts). In orderto determine the complete vehicle kinematic equations, define the generalized inertia matrix for thebody/particle system

�b/p =

Jb − mplrpl

rpl− mpr rpr rpr mbrcg + mpl

rpl+ mpr rpr mpl

rple1 mpr rpre1

−mbrcg − mplrpl

− mpr rpr m�

mple1 mpre1

−mpleT

1 rplmpl

eT1 mpl

0−mpre

T1 rpr mpre

T1 0 mpr

and the generalized added inertia matrix (due to motion through a fluid) as

�f =

J f Df 0 0

DTf M f 0 0

0T 0T 0 00T 0T 0 0

.

The total generalized inertia matrix is defined as the sum of the body/particle system and thegeneralized inertia due to the fluid,

�sys =

�b/p +

�f . The 8 × 8 matrix

�sys is positive definite and

depends on the vehicle geometry and mass distribution. The added inertia terms are significant forvehicles such as underwater vehicles and airships, where mass and displaced mass are of the sameorder.

28

The total body angular momentum hsys, the total body translational momentum psys, and themoving point mass momenta Ppl

and Ppr are defined as follows:

hsys

psys

ppl

ppr

=�sys

ω

v

rpl

rpr

. (12)

The complete equations of motion are

R = Rω (13)

X = Rv (14)

hsys = hsysω + psysv + m (15)

psys = psysω + f (16)

ppl= mprpl

+ e1 · (mprplω − v) (17)

ppr = mprpr + e1 · (mprprω − v) (18)

Steady Motions. In nominal operation, the SCALARS vehicle should perform steady, level glidingflight, either upward or downward. As for any flight vehicle, it is crucial to identify the steady flightconditions, as determined by various system parameters. The key parameters are position of themasses rpl

and rpr and the net weight (weight minus buoyant force). For symmetric (level) flight,the variables which these terms affect most are angle of attack (α), glide path angle (γ), and speed.The pitch angle θ is given by the expression θ = γ − α.

®°

zi

xi

xb

vb

zb

µ

Figure 23: Angles in the plane of symmetry.

In steady, level flight, ω = 0 and rpl= rpr = 0. Referring to equations (15) and (16), the steady,

level flight conditions give

0 = f (19)

0 = psysv + m (20)

Note that equation (20) simplifies to m = 0 when M = m�, that is, when added mass can be

neglected. This is the case, for example, for a conventional airplane flying in Earth’s atmosphere.

Numerical Simulation. With the equations of motion and equilibria determined, the system canbe numerically simulated to provide a preliminary assessment of stability and controllability. Thenonlinear equations of motion were implemented in Matlab, assuming a 10 kg vehicle operating inTitan’s atmosphere. Following are other relevant parameters:

29

Outboard Hull Inboard Wing

L = 1.9 m c =3

8L

D = 0.75 m b =3

4L

f = 4 AR = 2

Horizontal Stabilizer Vertical Stabilizer

cH = 0.3 m cV = 0.3 m

bH = 0.3 m bV = 0.3 m

ARH = 1 ARV = 1

−150

−100

−50

0

50

−40−20

0

0

20

40

60

80

100

120

y, m

t = 240

x, m

t = 180

t = 60

t = 120

z, m

Figure 24: Simulated flight path of a maneuvering SCALARS vehicle.

The equations of motion were integrated numerically, for two representative control sequences. Thefirst control sequence corresponded to a maneuver involving a downward glide and a banked turnto the left as shown in Figures 24. Initially, the vehicle is in level, equilibrium descending flight.The masses are then moved asymmetrically, with the left mass moving forward and the right massmoving back. This input induces twist in the wing which then rolls the vehicle to the left, eventually

30

resulting in a turn to the left. After completing the turn, the masses are returned to their original(symmetrical) position, and the vehicle returns to level, equilibrium gliding flight. A short whilelater, the masses are moved symmetrically forward and the vehicle pitches down and speeds up, asindicated by the slight steepening of the flight path after 180 seconds. Figure 25 shows a converseof this maneuver in which the vehicle begins in ascending gliding flight and performs a banked turnto the right. These simulations provide anecdotal evidence that the equilibrium motions of interestare stable and that the vehicle is controllable, in an informal sense. A thorough, formal analysis ofstability and controllability, such as the analysis presented in [19], would come in a follow-on effort.

−150

−100

−50

0

50

0

20

40

60

−80

−60

−40

−20

0

y, m

t = 240

x, m

t = 180

t = 120

t = 60

z, m

Figure 25: Simulated flight path of a maneuvering SCALARS vehicle.

Autonomous Control. The simulation results shown in Figure 24 represents open-loop control.The ambient fluid is assumed to be quiescent and there are no vehicle asymmetries or other dis-turbances. Naturally, this would not be the case on Titan. For an autonomous vehicle, feedback,or closed-loop, control is a basic requirement in order to reject disturbances and compensate formodel uncertainty. Feedback, in turn, requires sensors. For basic control, as described in [19] forconventional underwater gliders, the SCALARS vehicles would, at the very least, require altitudeand inertial navigation sensors for short-term navigation (“dead reckoning”). These sensors wouldlikely be small and inexpensive solid state, strap-down sensors. Obstacle avoidance sensors, possi-bly vision-based, would also be required as well as any special sensors necessary to enable dockingwith the base station. Depending on how the current technology matures, it may also be possibleto include local air-relative velocity sensors such as a Doppler velocity profiler.

For long distance navigation, dead reckoning is insufficient. Terrestrial autonomous vehicles can

31

use any one of several approaches for long distance navigation, including

• reference signals from the global position system (GPS) constellation, possibly integratedwith a geographic information system (GIS),

• reference signals from a surveyed long baseline (LBL) transponder network,

• vision-based mapping and localization schemes.

A more careful cost-versus-requirements analysis would be required to determine the most cost-effective solution.

Dynamic soaring. In general, the problem of dynamic soaring may be treated as a trajectoryoptimization problem with an aircraft performance model, that is, one in which the aircraft isassumed to be a point mass, subject to a three-dimensional control force. Working in such asetting, Boslough [6] presents an excellent overview of the basic requirements to accomplish dynamicsoaring. He reviews a number of fundamental papers on the topic, even reproducing and correctingsome of the prior work.

I1

X t( )

U X( )

X

U

V

I2

I3

Figure 26: Vehicle motion in a steady laminar shear layer.

One very useful observation, which Boslough attributes to Hendricks [22], is that the minimumwind gradient necessary to accomplish dynamic soaring is

dU

dh> 2CDV,

where U is the wind speed, as shown in Figure 26, h is altitude, V is the vehicle air-relative speed,and CD is the drag coefficient in a given flight condition. Thus, if wind speeds inferred fromthe recent Huygens descent probe data suggest that wind gradients are sufficiently large at lowaltitude, there is reason to suspect that a SCALARS vehicle could perform dynamic soaring forefficient locomotion.8 If not, the proposed system would still be quite effective, due to the efficientnature of buoyancy driven gliding, and drift induced by any prevailing winds could still be exploitedto improve locomotive efficiency.

Having developed a suitable vehicle design, one may proceed with numerical trajectory optimiza-tion techniques to determine the most efficient dynamic soaring patterns for the given vehicle.

8Wind-speed data had not been released at the time of this report.

32

Boslough’s work [6] reduces the rather complicated trajectory optimization problem to a parame-ter optimization problem by assuming a particular trajectory shape a priori. A more sophisticatedapproach, which might be pursued under a Phase II effort, would involve numerical trajectoryoptimization using a full vehicle dynamic model. Previous trajectory optimization results suggestthat, for the most efficient dynamic soaring trajectories, a bird or vehicle should travel more-or-less orthogonally to the direction of the wind [57]. If dynamic soaring is indeed possible using aSCALARS vehicle, the turning requirement determined from trajectory optimization would drivethe roll control requirements, particularly the torsional stiffness of the inboard wing.

4.3 Technology Hurdles

Several key technology areas have been identified for further investigation to determine if andwhere major technology advances will be necessary to enable SCALARS vehicle deployment andoperation. These areas include:

• Advanced materials for cryogenic environments.

• Aerodynamics of inboard wing aircraft [47, 64].

• Control for this new class of underactuated vehicle.

– Internal moving mass actuators [8, 60, 72].

– Using wing twist for lateral-directional control [14, 21, 48, 4, 58].

– Buoyancy driven gliding [13, 16, 20, 32, 63, 70].

– Vehicle autonomy

∗ real-time path planning and path following,

∗ advanced positioning and navigation (e.g., long baseline navigation, simultaneouslocalization and mapping algorithms, etc.),

∗ adaptation to parameter uncertainty and variations,

∗ autonomous dynamic soaring [55, 6, 62, 69], and

∗ health monitoring and robustness to component failures.

• System packaging and delivery (e.g., inflatable structures [9, 43]).

33

5 High Altitude Wind Energy Absorber

The high-altitude wind energy absorber concept originally was inspired by oscillating-wing energyconversion devices that have been researched and developed for tidal-stream energy conversionand micro-hydropower generation. These devices use actuators linked to a jointed-arm or pivoton a fixed frame of reference, and our proposal anticipated that such a configuration would notbe suitable for a high-altitude, tethered system. Instead, we proposed to use a similar twin-hull, inboard wing airframe as the SCALARS vehicle, with internal reciprocating linear inductiongenerators in each of the outboard hulls to convert vehicle pitching and plunging motions intoelectrical energy. To distinguish this vehicle from the SCALARS exploration gliders, we refer to itas the OAWEA, an acronym for Oscillating Airfoil Wind Energy Absorber.

Other elements of our Phase I high-altitude wind energy harvesting system include

1. linear induction generator modules mounted in the twin outboard hulls of the buoyantOAWEA,

2. a relatively short tether connecting the OAWEA to a mooring aerostat,

3. a much longer tether connecting the aerostat to the anchor base, and

4. a mooring aerostat, which provides the necessary buoyancy to support the mooring tethersand its own payload of sensors, communications and data processing equipment, and sufficientenergy storage to power these loads with variable wind energy input.

The state of technology for each of these elements is described in Section 5.1, which begins with anoverview of previous high-altitude wind energy generator concepts and the principles of oscillating-wing energy conversion in steady fluid flows. A preliminary analysis of the complete energy har-vesting system is given in Section 5.2, followed by identification of technology hurdles in Section 5.3.

Energy storage and electrical power conditioning on the docking base are covered in Section 6,which also describes the inductive charging of the SCALARS’ on-board power supplies. Theseaspects of energy management are covered in the next section of this report, while this sectionfocuses on energy absorption, conversion, and transmission to the docking station.


Previous High-Altitude Wind Energy Generators. The large power density of high-altitudewinds at the top of the planetary boundary layer has been described already in Section 2. Sev-eral concepts have been proposed to harness this resource for utility-scale power generation onEarth. These include conventional wind turbines mounted on tethered, buoyant aerostats [52],a series of kites mounted on a cable loop, 10 km long, driving a large ground-based genera-tor [45] (also see www.laddermill.com), and a counter-rotating autogiro system [54, 3] (also seewww.skywindpower.com).

Among these three basic concepts, the autogiro system developed by Sky WindPower is the mostadvanced in its engineering maturity, with a complete “wind-to-wire” demonstration of a 60 kWprototype with twin 4.6 m diameter rotors, flown at an elevation of up to 18 m in Australia; see

34

Figure 27: Deployment of Sky WindPower autogiro prototype in Australia.

Figure 27. The rotors of the Sky WindPower autogiro are inconsistent with our concept philos-ophy of having no moving parts exposed to potentially corrosive aerosols and particulates in theatmospheres of Titan and Venus. Moreover, methane icing of the autogiro rotor blades would be apotential concern on Titan.

Oscillating Wing Energy Conversion in Steady Fluid Flows. As an alternative to spinningrotor devices, we turned to oscillating-wing energy conversion devices, which have been modeledboth physically and numerically by Kevin Jones and his colleagues at the Naval PostgraduateSchool in Monterey, California [29, 27, 28, 33]. Completely apart from the work of Jones et al,an oscillating-wing tidal stream energy device known as the “Stingray” has been researched anddeveloped by The Engineering Business, Ltd, in the United Kingdom. Extensive numerical andphysical modeling was conducted [15], and a 150 kW prototype was tested in the Shetland Islands[67]; see Figure 28. Both of these sets of studies indicated that harvesting efficiencies of 15-25% areachievable (based on incident fluid energy flux on the rectangular area vertically swept by wing’sleading edge).

The problem with these technologies is that they have mechanical actuators linked to a fixedanchor base. Such a configuration would not be suitable for a high-altitude, tethered system, andit also would be inconsistent with our concept philosophy of having no moving parts exposed tothe atmosphere.

Based on anecdotal evidence of vertical oscillations reported in air-towed gliders and sailplanes, andin ship-towed underwater ocean instrumentation packages, it seemed likely that we could design ahigh-altitude, tethered buoyant wing that would undergo similar oscillations. These motions wouldbe converted to electricity via reciprocating, linear induction generators.

35

Figure 28: Stingray oscillating-wing tidal stream energy prototype in the United Kingdom.

Reciprocating, Linear Induction Generators. The key component that converts OAWEAmotions into electrical energy is a self-contained inductive power module configured as a canisterthat houses a permanent-magnet “proof mass” suspended by one or more springs that enable it tomove relative to a fixed stator coil. One of the challenges inherent in such an inductive generatoris the relatively long-period (and associated slow speed) of the exciting force (expected to be in thefrequency range of 0.05 to 0.5 Hz), relative to the high frequencies needed to generate significantvoltages (typically 2 to 20 Hz, depending on magnet and coil dimensions).

The principles of using linear induction generators to convert energy from such slow-period oscil-lations were first developed for ocean wave energy conversion in 1978 [46], and recent work on theArchimedes Wave Swing prototype now installed off the coast of Portugal has stimulated additionalrecent research on low-frequency linear induction conversion systems at Delft University of Tech-nology in the Netherlands [49], Uppsala University in Sweden [12], and the University of Durhamin the United Kingdom [41, 42].

Linear permanent magnet, inductive machines also have been proposed as regenerative force ac-tuators for the mitigation of earthquake disturbances in civil structures, which have similarly lowexcitation frequencies. Regenerative excitation makes it possible to isolate active-damping actua-tors from the external power grid, which is necessary during earthquakes, when the quality or eventhe mere availability of utility-grid power is questionable [61].

Electro-Mechanical Mooring Tethers. Catenary tether design tools have been developed forthe high-altitude Sky WindPower autogiro device [44], and these are equally applicable to ourproposed slack mooring of a high-altitude, buoyant oscillating wing. Furthermore, a variety oflightweight electro-mechanical tether cable configurations have been qualified for high-radiation

36

space environments [17, 11]. Fiber-optic tethers also have been developed [5], which would enabledata communications between the ground docking station and mooring aerostat, which could thenprovide long-range communications to and from Earth, either directly or via orbiting satellite.

Mooring Aerostat. Three alternative aerostat designs are available to support the OAWEAmooring tether and the long electro-mechanical tether from the aerostat to the docking station onthe ground: spherical, streamlined, and variable-lift. The spherical aerostat has no preference to aparticular orientation and does not produce any aerodynamic lift in a horizontal wind profile. Thestreamlined aerostat has an aerodynamic shape similar to a horizontal teardrop to reduce its winddrag, together with tail fins to maintain its orientation into the wind. A third type of aerostat istermed variable-lift, reflecting its ability to generate a significant amount of lift in a wind stream,through the addition of wings or kite-like lifting surfaces, or by using the hull to derive lift, or somecombination of these two.

Two different variable-lift aerostats, the SkyDoc9 and Helikite10 are illustrated in Figure 29. Theseaerostats have increased lift at higher wind speeds and are able to maintain a more vertical orien-tation, unlike a spherical or streamlined aerostat, in which the drag increases with wind speed, butnot lift, causing the aerostat to lose altitude.

Figure 29: Two variable-lift aerostats: Sky-Doc (left) and Helikite (right).

In support of the National Research Council of Canada’s Herzberg Institute proposal for a ’LargeAdaptive Reflector’ radio telescope, comprising a 200 m reflector and a receiver held aloft at analtitude of 500 m by a tethered aerostat, researchers at McGill University in Monreal evaluated theperformance of both the Sky-Doc and Helikite variable-lift aerostats. Field tests were performed bytowing the aerostats behind a small boat; see Figure 30. During the tests, both aerostats performedpoorly, becoming unstable at “high speeds” of 40 to 50 km/hr (11 to 13 m/s), sometimes divingviolently all the way to the water surface. Based on these observations, the McGill University teamexcluded variable-lift aerostats from further analysis and focused instead on more conventionalspherical and streamlined types [31].

According to [31], at a Reynolds number of 106, the drag coefficient for a spherical body is about0.15, while for a streamlined body of fineness ratio 2.4, it is approximately 0.058. When tail finsare added to the streamlined aerostat, its drag coefficient increases to 0.073, or about half thatof the spherical aerostat. The drag force on a conventional aerostat also is proportional to theprojected frontal area, and the streamlined aerostat has a frontal area 1.7 smaller than a sphere

9www.floatograph.com/oilspill/images.html10www.allsopp.co.uk/science/index.html

37

Figure 30: Field tests of Sky-Doc (left photo, upper aerostat) and Helikite (left photo, loweraerostat) in towing trials behind a small boat on the Columbia River (right photo).

having the same volume. The combined effect of lower drag coefficient and smaller frontal areayields a drag force on a streamlined aerostat that is about 3.5 times smaller than the drag force onan equal-volume spherical aerostat for a given wind speed.

Figure 31: TCOM, LP aerostat family (upper left photo), 15-m long aerial photo and advertisingplatform for sporting event (lower left photo), and 17-m tactical surveillance platform used bymilitary ground forces (right photo).

A leading-example of a streamlined aerostat designed to remain aloft in high winds has a bluntteardrop shape, with three large tail fins in an inverted-Y configuration [30]. These are commerciallymanufactured by TCOM Limited Partnership in Columbia, Maryland for a variety of commercial,military, and scientific applications; see Figure 31; also see www.tcomlp.com.

38

5.2 Concept Development and Preliminary Analysis

This section describes preliminary development of the OAWEA energy harvesting concept, so as toobtain order-of-magnitude estimates for platform and tether dimensions, as well as expected energyoutput as a function of wind speed. As with the SCALARS concept development in Section 4 of thisreport, the modeling methods used here are intentionally simplistic, and assumptions are sometimesmade on the basis of only a single literature reference or even arbitrarily, simply to have a startingpoint from which to begin the iterative design process.

Due to the revolutionary nature of both the SCALARS and OAWEA elements, and given theresources available in Phase I, we believe that an integrated approach that considers the impact ofone element’s performance on the performance of another element (e.g., OAWEA energy productionand SCALARS range) is more appropriate toward developing a novel architecture in its entirety,rather than expending our resources to develop any one element in great detail. Indeed, detaileddevelopment of a single element may prove to be wasted when eventual development of anotherelement suggests design or performance constraints on the original element that had been developedin detail.

We introduce this section by describing the theoretical principals of oscillating-wing energy conver-sion in steady fluid flows and their practical demonstration in the ”Stingray” tidal stream energydevice. We then develop an improved solution with a much lower drag penalty, based on har-nessing the long-period phugoid oscillations encountered in aircraft longitudinal stability problems.Next we estimate the along-chord and normal-to-chord accelerations of the OAWEA, assumingthat the restoring force of the catenary tether to the aerostat balances the aerodynamic drag onthe OAWEA, satisfying the necessary equilibrium between thrust and drag for a sustained phugoidoscillation in a turbulent wind regime. These acceleration profiles enable us to estimate the volt-age and current capability of a linear induction power module fixed at an appropriate orientationwithin the OAWEA hull. Finally we determine the number of modules to be connected in series inorder to obtain sufficient voltage for down-cable transmission to the docking station with minimalresistive losses, while ensuring that all components have sufficient buoyancy to remain aloft in thehighest expected wind speeds.

5.2.1 Physics of Oscillating-Wing Energy Absorption

As shown particularly in [27] and [28], key parameters that determine the energy absorption effi-ciency of an oscillating wing are the phasing between pitch and plunge and the amplitude of pitchactuation. Although these studies were of a device that is constrained in surge, it is useful tounderstand the basic physics of this simpler device before considering the OAWEA, which is onlyweakly restrained in surge and is expected to experience significant downwind-upwind oscillations.

This theoretical overview is derived from the mathematical description [28]. Consider the airfoilsection diagrammed in Figure 32, which is constrained horizontally, but which can oscillate in pitchand plunge, with an arbitrary phase angle between these two motions. Reference [28] demonstratesthat energy is absorbed from the flow when the pitch oscillation is approximately 90◦ out of phasewith the plunge oscillation. Under this condition, the extent to which the oscillating airfoil generatesthrust or absorbs power depends on the amplitude of its geometric pitch angle relative to horizontal.To illustrate the importance of this parameter, consider the three cases illustrated in Figure 32.

In Case A, there is no applied pitching moment, and the airfoil ”feathers” through the flow as an

39

Figure 32: Aerodynamics of an oscillating-airfoil, fixed in surge, but freely moving in pitch andplunge.

external force moves it up and down. Now consider the application of an external torque to changethe pitch of this oscillating airfoil, as shown in Cases B and C, such that the airfoil is horizontal atits extreme upper and lower positions in the oscillating plunge cycle, and attains its extreme pitchangle at the mid-points of the plunge cycle, creating a 90-degree phase difference between pitchand plunge.

In Case B, the applied pitching moment is applied in a down-nose direction as the airfoil rises oran up-nose direction as the airfoil falls. This generates a lift force counter to the plunge directionof the airfoil, such that an external vertical force must be applied to overcome it. References [29]and [28] demonstrate that Case B represents the required condition for generating thrust. Jonesand his colleagues have developed an evolving set of micro-air vehicle (MAV) designs based on thisprinciple, with a rear pair of such oscillating wings. Their latest design weighs 12 grams, has a25-cm wing span, exhibits stable flight at speeds of 2 to 5 m/s, and has a static thrust figure ofmerit 60% higher than propellers of similar scale and disk loading [26].

In Case C, the applied pitching moment is applied in an up-nose direction as the airfoil rises orin a down-nose direction as the airfoil falls. This generates lift in the same vertical direction thatthe airfoil is moving, and the only external force that must be applied is that needed to overcomefluid-induced pitching moment, which tends to feather the wing, as in Case A. References [29] and[28] demonstrate that Case B represents the required condition for absorbing energy, and Reference[28] describes a proof-of-concept experiment whereby net energy is absorbed from a steady flow ina laboratory flume and produces useful work.

40

An independent confirmation of this basic oscillating-wing theory is available from a series ofnumerical simulations, laboratory flume tests, and ocean trials of the “Stingray” tidal stream energyconverter [15, 67]. This device uses a small actuator to change wing pitch, which then induces large,reciprocating plunge motions of the wing. The wing strokes a double-acting hydraulic cylinder thatpumps fluid to a hydraulic motor/generator to produce electrical power. A 150-kW prototype,shown in Figure 28, has been tested in the Shetland Islands. It weighed 180 tonnes (including 80tonnes of deadweight ballast), had a rectangular seafloor “footprint” of approximately 400 m2, anda working height of 22 m above the bottom. The oscillating wing span was 15.5 m, with a chordlength of 3 m, and a swept area of approximately 180 m2. Sea trials of the prototype indicatedan energy absorption efficiency of 12% in a tidal stream depth-averaged velocity of 1.5 m/s [67].Numerical simulations of a similar device by Jones and his colleagues suggest that an efficiency of21% should be achievable with an optimized design [28].

A time-history simulation of the “Stingray” oscillating-wing through one plunge cycle is plotted inFigure 33. Note the similarities with Case C of Figure 32, namely that the wing pitches nose-upinto the flow as the wing rises and nose-down into the flow as the wing falls, and that the geometricpitch of the wing reverses at the top and bottom of the cycle. Also note that the pitch controlalgorithm for maximum energy absorption is to continually change the geometric pitch so as toachieve a relatively constant effective angle of attack of approximately 15◦, except for the brief timewhen the pitch is reversing. This involves steep geometric pitch angles (approximately 50◦) duringthe part of the cycle when the plunge velocity is greatest.

Figure 33: Time-domain simulation of “Stingray” oscillating-wing.

The OAWEA’s mooring arrangement does not rigidly constrain it in surge, so its dynamic behaviorwill differ from that shown in the above diagrams, although the same basic lift dynamics of apitching and plunging wing will still apply. The OAWEA is tethered to an upwind aerostat by aslack, catenary mooring cable, whose weight provides a restoring force proportional to downwinddisplacement. It nevertheless permits significant horizontal excursions, and its behavior is expectedto be similar to that of an underwater instrumentation package towed by a surface vessel, asdescribed in the next part of this section.

41

5.2.2 An “Inverted Two-Stage Tow”

Ocean scientists, hydrographers and geophysicists often must tow underwater instrumentation (suchas side-scan sonars, cameras, sound sources for bottom and sub-bottom acoustic reflection profiling,and current meters) at a relatively constant elevation above the seafloor, at depths of a few hundredmeters to several kilometers beneath the sea surface. An example of such a towing configurationis shown in Figure 5-8, which depicts the MARLIN towfish used by researchers at Oregon StateUniversity to investigate turbulence processes that lead to mixing the deep ocean near boundariesbetween water masses or along the margins of continental or island shelves or adjacent to submergedbanks and reefs. MARLIN is capable of making measurements of turbulence and water columnmicro-structure to depths of 3,400 m, with an electro-mechanical towing cable up to 9,000 m inlength [40].

In a collaborative effort sponsored by the National Science Foundation’s Ocean Technology Inter-disciplinary Coordination program, Virginia Tech has built a towfish to house a five-beam acousticDoppler current profiler (VADCP) for the purpose of measuring small-scale ocean turbulence [59].The towfish has a streamlined body, fixed fins and two independently actuated stern planes to pro-vide control and actively reject disturbances in pitch and roll. This towfish is expected to operatein coastal waters at depths up to 200 meters below the sea surface and must regulate its attitudewithin 1 degree of nominal at speeds ranging from 1 to 3 m/s, in order to permit uncorruptedmeasurements of ocean turbulence by the VADCP. The towfish is trimmed to have 5% positivebuoyancy so that the VADCP can be recovered should the towing tether fail.

Figure 34: A two-stage towing system for ocean science.

The towing assembly consists of an electro-mechanical tow cable attached to the VADCP towfish.A depressor weight is attached to the tow cable somewhere relatively close to the towfish. Theportion of tether between the towing vessel and the depressor is called the ”main catenary” whilethe portion between the depressor and the towfish is called the “pigtail” but it is important to notethat both segments are part of the same cable; see Figure 34. Typically, the length of the pigtailsegment can be adjusted to attenuate the effect of towing vessel motion on the towfish. While thistowing arrangement provides good passive disturbance rejection, it cannot compensate for pitchand roll biases due to towfish center of gravity (CG) offsets or asymmetric flow about the towfish.Therefore the towfish has servo-actuated stern planes to actively reject such biases and to enhancestability.

As with the SCALARS concept development, it is clear that the development of the OAWEAconcept can be well supported by design methodologies originating in the knowledge domain ofocean engineering, even though our intended application is atmospheric. The depressor weight

42

shown at the end of the primary tow cable in Figure 34 is provided to absorb the heaving andpitching motions of the surface tow vessel, such that they are not transmitted through the secondarycable, enabling the towfish to remain in a stable attitude, which is critical for accurate measurementsof turbulence and other water column micro-structures.

For our atmospheric application, the two-part tow system would be inverted: the depressor weightwould be replaced by a mooring aerostat, similar to those shown in Figure 30. Rather than tethera single OAWEA to the top of this aerostat (which would exactly mirror the towing arrangementshown in Figure 34), we envision one or more OAWEAs being tethered to the primary mooring cableat some distance below the mooring aerostat. A numerical simulation of this towing arrangementhas been developed at Virginia Tech [59] and this provides an excellent tool to determine what sortof changes might be made to accomplish the reverse goal, namely to destabilize the towfish andcreate a divergent, limit-state oscillation. A description of this simulation and some encouragingresults we obtained in Phase I are given below.

In a series of exploratory simulations of the above system, a variety of different fixed design features(longitudinal CG location, tail fin size, and pigtail length) were changed to see if the natural phugoidoscillation of the towfish could be enhanced to create a diverging oscillation that reached somesteady state limit. This proved to be possible, as shown in Figure 35.

Figure 35: A limit cycle oscillation in an unstable two-stage towing system.

The plot in Figure 35 shows the development of a self-sustaining phugoid-like oscillation having aperiod of about 25 seconds and amplitude of 4 meters for a steady vessel towing speed of 3 m/s.Note that this is a somewhat less energetic motion than the “Stingray” oscillating-wing exampleplotted in Figure 33, which has a period of about 16 seconds and amplitude of 6 meters for a tidalcurrent speed of 1.5 m/s. These results encourage further investigation into the use of an OAWEAfor harvesting wind energy on Titan.


As for the SCALARS vehicle concept, several key technology areas have been identified for furtherinvestigation with regard to the OAWEA concept; many of the areas overlap.

• Advanced materials (particularly elastic materials for very cold environments, for use in thetwisting inboard wing and in the variable buoyancy actuator) [9, 43]).

• Harvesting ambient fluid energy using inductive generators [11, 17, 25, 27, 41, 44, 15].

• System packaging and delivery (e.g., inflatable structures).

43

6 Docking Station

The SCALARS and OAWEA elements of our proposed architecture break considerable new groundand are highly evolutionary (if not revolutionary) themselves as stand-alone technologies. Sinceit is the docking station that brings these two elements together, we have adopted a lowest riskapproach toward developing this component. Although we had originally envisioned a drifting,airborne docking station, potentially building on the research of a previous NIAC project entitled“DARE,” this seemed far too risky, given the large uncertainty in the atmospheric dynamics onTitan and uncertainties in the development of the SCALARS and OAWEA elements.

Moreover the control challenges involved in docking a gliding vehicle are challenging enough whenthe docking station is fixed in space. Indeed, the problem of docking autonomous underwatervehicles (AUVs) is one of current interest and bears directly on the proposed architecture [56, 39, 51].

Figure 36: The Rosetta Lander.

This section begins with a technology review of the anchoring system of the Rosetta comet lander,shown in Figure 36, which was successfully launched by the European Space Agency in March2004. Unlike the other architecture elements, this component is quite well-developed (a NASATechnology Readiness Level (TRL) of at least 6). The unitized regenerative fuel cell (URFC) forstoring variable wind energy is at a somewhat lower TRL, but it has been proven in NASA’s high-altitude solar-powered airplanes in Earth’s atmosphere. Likewise, inductive couplers for rechargingthe SCALARS vehicles and enabling bi-directional transfer of data have been proven reliable ascommercial off-the-shelf (COTS) equipment for harsh industrial environments on Earth. Neitherthe URFC nor the COTS inductive couplers have been tested at the cryogenic temperatures to beencountered on Titan.

44


This section begins with a review of the anchoring system for the Rosetta comet lander, successfullylaunched by the European Space Agency in March 2004. Unlike our other architecture elements,which have not even progressed to laboratory-scale bench testing, this component is deemed to havea Technology Readiness Level (TRL) of at least 6. The unitized regenerative fuel cell (URFC) forstoring variable wind energy is at a somewhat lower TRL, but it has been proven in NASA’s high-altitude solar-powered airplanes in Earth’s atmosphere. Likewise, inductive couplers for rechargingthe SCALARS vehicles and enabling bi-directional transfer of data have been proven reliable ascommercial off-the-shelf-(COTS) equipment for harsh industrial environments on Earth. Neitherthe URFC nor the COTS inductive couplers have been tested at the cryogenic temperatures to beencountered on Titan.

Anchoring System. The 100 kg Rosetta Lander will be the first spacecraft ever to make a softlanding on the surface of a comet nucleus. The box-shaped Lander is carried in piggyback fashionon the side of the Orbiter until it arrives at Comet 67P/Churyumov-Gerasimenko. Once the Orbiteris aligned correctly, the ground station commands the Lander to self-eject from the main spacecraftand unfold its three legs, ready for a gentle touch down at the end of the ballistic descent. Onlanding, the legs have been designed to dampen out most of the kinetic energy to reduce the chanceof bouncing, and they can rotate, lift or tilt to return the Lander to an upright position. Given theuneven and uncertain topography at our proposed Titan deployment site, we propose to use thesame leg and anchor design.

Since the anchoring system was designed for the Rosetta mission, it is decidedly robust. Theharpoon design uses both a sharp point and shovel flukes in order to ensure acceptable anchoringin surfaces that have both high tensile strength and low density. The point can penetrate the icysurface of the comet, which is expected to have high tensile strength, and the shovel flukes willdevelop holding power in the low-density material beneath this crust. There is an accelerometeron each harpoon anchor for determining the success of the shot and estimating the required torquefor the screw auger anchors at the end of each leg.

We estimate this anchoring system to be at TRL of 6. The harpoon anchoring systems have beensuccessfully tested in a variety of relevant comet environments including inhomogeneous quantitiesof water, quartz sand, wood fiber, gravel, clay, and ice. However, because Rosetta has not landedon its target comet yet, the system still has some risk.

Dock Energy Storage. Unitized regenerative fuel cells (URFC) are proton-exchange membranestacks that can be operated reversibly as electrolyzers when connected to a power source, and as fuelcells when connected to an electrical load. These have been successfully prototyped aboard NASA’shigh-altitude solar-powered airplanes. Although the cryogenic environment of Titan is favorable forlow-mass storage containers for liquid hydrogen and liquid oxygen, liquid water storage will requirelocalized heating initially by radioisotope thermal generators and subsequently by thermo-electricheating powered by harvested wind energy. The water storage container must be thermally wellinsulated, and the operation of PEM stacks at such low temperatures remains to be researched inPhase II.

Ragone diagrams for URFCs, including storage components, have been developed for use in aeroshellpayload sizing. Similar diagrams have been developed for rechargeable lithium battery chemistriesto be installed on the SCALARS for powering their sensors, moving-mass actuators, and avionics.

45


Key technology areas requiring further investigation with regard to the docking station conceptinclude:

• Controlled docking with the base station [56, 39, 51].

– Sensing relative position with the necessary precision.

– Rejecting gust and steady current disturbances.

–

• Minimizing power and information losses through inductive couplings.

46

7 Extendability to Other Operating Environments

In addition to its potential use on Titan, the proposed architecture could also be used on Venus oreven in the atmosphere or hydrosphere of earth. Indeed, several of the core technologies employedby the proposed system are already being developed for use in Earth’s oceans. Table 3 shows someof the key properties of these potential operating environments.

Table 3: Planetary Atmospheric Properties

Property Earth [2] Venus [7] Titan [34]

Radius (km) 6378 6052 2575Mass (kg) 5.972e24 4.869e24 1.346e23Surface Gravity (m/s2) 9.81 8.87 1.35Surface Temperature (◦C) 14 457 -179Surface Density (kg/m3) 1.23 66.5 5.26Surface Pressure (bar) 1.01 94 1.44

Note that Titan’s atmosphere is roughly four times as dense as Earth’s at the surface and the grav-itational acceleration is almost an order of magnitude smaller. Moreover, the kinematic viscosityon Titan is also an order of magnitude smaller; see Section A. A consequence of these observationsis that a small, slow-flying, buoyancy driven glider could operate on Titan at quite low energeticcost. Moreover, if the glider is programmed with energy-saving behaviors, such as the ability to dy-namically soar in the PBL, the vehicle could cover a large area of terrain before having to rechargeits power supply.

7.1 Venus

In many ways, Venus is an even more appropriate environment than Titan for the proposed systemarchitecture. The density is an order of magnitude greater, reducing the total volume requiredfor neutrally buoyant flight as well as the change in volume required to accelerate upward ordownward. Gravitational acceleration near the surface is roughly the same as it is on Earth. Theambient pressure is 94 Earth atmospheres, roughly the same as the water pressure at a depth of940 meters in an Earth ocean. The fact that buoyancy driven gliders have already been operated atsuch depths suggests that technological obstacles are not insurmountable. The corrosive (sulfuricacid) atmosphere would have little effect on a vehicle which is internally actuated as there wouldbe no rotating shafts or other dynamic parts exposed to the ambient fluid. The greatest obstaclewould appear to be the extreme temperature. In fact, the surface temperature on Venus is higherthan that on the surface of Mercury, in large part because of the thick atmosphere [2].

Although an enormous amount of solar energy is reflected by Venus’ thick atmosphere, making itthe bright “morning star” that is plainly visible from Earth, Venus also absorbs a great deal ofsolar energy. External and internal energy combine to force some extremely dynamic atmosphericprocesses. Fluid kinetic energy is clearly available to be harvested on Venus, although solar energyis also readily available (despite the thick atmosphere). We point out that a solar energy harvestingsystem would need to be engineered so as to be impervious to corrosion, a feature which is inherent

47

Figure 37: Venus, as imaged by Galileo. (Image credit: NASA)

in the proposed system architecture. This may or may not be a significant obstacle for exploringVenus; engineering solar panels that can survive in Earth’s ocean environments has turned out tobe quite challenging. In any event, it appears that the proposed system architecture, or some setof component subsystems, could provide a means for exploring the terrain of planet Venus.

7.2 Earth Atmosphere

Assuming that the proposed architecture could be engineering for Titan, it is natural to conjecturethat it could also be deployed in Earth’s atmosphere. The idea has merit and deserves attention,but there are some key aspects of Earth’s atmosphere that might make such a system less feasiblehere. Earth’s atmosphere is even less dense than Titan’s meaning that a buoyancy driven glideror aerostat of given mass would be larger than its Titan counterpart. Moreover, Earth’s disparatefeatures of land and water and its proximity to the Sun lead to random atmospheric disturbances(gusts) which could make the design of such a system challenging. Having said this, lighter-than-air vehicles were successfully deployed on Earth prior to the advent of powered, fixed-wing flight.In fact, the idea of buoyancy driven gliding flight has already been proposed for use in air, forefficient transportation of people and cargo11, and a variety of techniques have been proposed andsuccessfully implemented for harvesting Earth’s wind energy, although not at the most energeticaltitudes.

7.3 Earth Hydrosphere

In fact, the problem of implementing a self-sustaining scientific monitoring system in Earth’s hy-drosphere motivated this entire project. Both the PI and the Co-PI have experience working withocean technology; buoyancy driven gliders and wave and tidal energy harvesting devices are focusareas for a great deal of current engineering research. Relative to Titan and Venus, the liquid waterof Earth’s hydrosphere provides an extremely dense medium in which buoyancy driven gliding isquite evidently feasible. (See Figure 14.)

11See http://www.fuellessflight.com/.

48

Figure 38: Earth, as imaged by Galileo. (Image credit: NASA)

Energy harvesting and storage is every bit as important and challenging in the ocean as it is onTitan or Venus. Solar arrays are subject to corrosion and biological fouling on short time scales,compared with the time scales of scientific interest. Other types of environmental energy, such aswaves or tides, exist only at the air-sea interface or in shallow regions. Thus, an energy harvestingdevice for exploring Earth’s oceans would have to operate at or near the surface or in a shallowregion.

7.4 Europa

The fact the components of the proposed architecture are either existing technologies or the focusof current applied research for ocean exploration raises another interesting possibility for planetaryexploration: Jupiter’s moon Europa. Quoting from the description at nineplanets.org [2]:

The images of Europa’s surface strongly resemble images of sea ice on Earth. Itis possible that beneath Europa’s surface ice there is a layer of liquid water, perhapsas much as 50 km deep, kept liquid by tidally generated heat. If so, it would be theonly place in the solar system besides Earth where liquid water exists in significantquantities.

To confirm the existence of liquid water elsewhere in the solar system would be a pivotal pointin the young history of space exploration. It raises the quite realistic possibility of ongoing pre-or protobiotic chemical activity. And, although the dogma of evolutionary biology held that liferequires not only water but sunlight, the relatively recent discovery of entire ecosystems whosefoundation is chemosynthesis has disproved that dogma. To many marine biologists, the onlyscientific event which could outshine the discovery of tubeworms surrounding deep-sea hydrothermalvents would be the discovery of living organisms in the depths of Europa.

To search for sites with interesting chemistry, or even living organisms, would require a system ofextremely efficient robotic explorers, such as the SCALARS vehicles proposed here. One valuablefeature of the proposed vehicles is that, because they are internally actuated, their operation would

49

Figure 39: Europa, as imaged by Galileo. (Image credit: NASA)

preserve the pristine environment; the only contaminants or impurities that would be introducedwould be carried on the vehicle’s external surface which would likely be sterilized in transit.

Figure 40: A deep-sea hydrothermal vent (left) and a colony of tubeworms (right).

Certainly a host of distinctly different engineering issues would have to be addressed to employSCALARS vehicles on Europa, most notably issues related to delivery and insertion, vehicle po-sitioning, and communication. With the exception of the first, these same issues arise in fieldingoceanic underwater vehicles so, again, there is reasonable hope that the necessary technologiescould be developed.

50

8 Conclusions and Recommendations

This report describes conceptual design of a revolutionary system for remote terrain exploration andenvironmental sampling on worlds with dense atmospheres. The proposed system is entirely self-sustaining, extracting energy from the planetary boundary layer for energy renewal, and potentiallyfor more efficient locomotion. The three major system components are: a fleet of rechargeable,internally actuated, buoyancy driven gliders; a tethered, oscillating wing whose motion is tunedto extract fluid kinetic energy from sustained high altitude winds; and an attached anchor andbase station to inductively recharge the gliders, upload science data, and download revised missioncommands. While the proposed system concept is novel, virtually all of the enabling technologiesexist currently or are in development.

Considering the proposed system architecture in the context of NASA’s shifting focus from aero-nautics to exploration raises two important observations. First, the proposed concept is quitewell-aligned with NASA’s stated program goals for planetary exploration. Second, the proposedconcept illustrates the need for NASA to preserve and cultivate its excellence in aeronautics. Afterall, the most intriguing environments for mankind to explore, and the most promising technologiesfor doing it demand fundamental expertise in aeronautical engineering.

The preliminary results described in this report suggest that the individual components of theproposed system architecture have great potential for use in planetary exploration. It is a steepclimb, however, from “a suggestion of great potential” to a thorough, rigorously developed roadmapof the enabling technologies. Section 8.1 describes the accomplishments of this Phase I effort whichwould directly support a follow-on effort to develop such a technology roadmap. Section 8.2 providesrecommendations to guide the follow-on effort.

8.1 Phase I Accomplishments

Following is a list of the more significant accomplishments of this Phase I effort.

1. Developed a mathematical model for the SCALARS terrain exploration vehicle.

• Developed a basic aerodynamic model for a twin-hulled, inboard wing, buoyancy drivenglider for preliminary sizing calculations.

• Developed a nine degree of freedom nonlinear dynamic model, incorporating the aero-dynamic model mentioned above.

– Implemented the model in Matlab to illustrate the nature of vehicle locomotion,demonstrating basic stability and controllability.

– The mathematical model can be used, along with existing numerical trajectoryoptimization software, to determine the most efficient gliding and soaring behaviors.

– The nonlinear model can be linearized to obtain modal information, particularlywith respect to the phugoid mode which is expected to play an important role inenergy-harvesting by the OAWEA vehicle.

2. Identified a necessary condition for dynamic soaring by SCALARS vehicles. (See [6].) Thiscondition can be easily tested for the given vehicle concept once wind data from the Huygensdescent are made publicly available.

51

3. Determined how to couple the long period (phugoid-like) mode of the OAWEA vehicle to anumerical model of a PM linear induction generator to estimate high-altitude wind energyconversion efficiency. (Tether dynamics were ignored in this Phase I effort.)

4. Established engineering feasibility for anchoring the OAWEA system and providing a dockingand charging station for the SCALARS vehicles. Such a station could incorporate commercial,off-the-shelf components for inductive power transfer and bi-directional data communication.

8.2 Recommendations for Future Concept Development

8.2.1 SCALARS Vehicle Concept Development

Further development of the SCALARS vehicle concept should focus on the following topics.

• Identify specific scientific mission and operational hardware needs for budgeting vehicle cost,weight, and volume.

• Reiterate SCALARS vehicle design procedure, using higher fidelity modeling assumptions andtechniques, to place final bounds on design parameters and to increase confidence in vehicleperformance and the results of parametric studies.

– Improve aerodynamic model for twin-hull, inboard wing vehicle (by modifying an existingvortex lattice code, if necessary).

– Refine/formalize stability and controllability analysis.

• Compare SCALARS concept with possible competing concepts in terms of energy efficiencyand performance.

• Enumerate and investigate potential failure modes of the SCALARS vehicle.

• Review and project the state of the art in vehicle autonomy, focusing on

– advanced positioning and navigation without global references,

– real-time path planning/optimization and path following,

– coordination of multiple vehicles for mapping and sampling,

– vision-based vehicle control,

– adaptation to model uncertainty and parameter variations, and

– health monitoring and fault tolerance.

8.2.2 Wind Energy Conversion Concept Development

Development of the OAWEA wind energy harvesting concept should focus on the following topics.

• As for the SCALARS vehicle, identify specific operational hardware needs for budgetingvehicle cost, weight, and volume.

52

• As for the SCALARS vehicle, reiterate OAWEA vehicle design procedure, using higher fidelitymodeling assumptions and techniques.

– Model the dynamic coupling between the OAWEA vehicle and an internal array of linearinduction generators.

– Determine bounds on design parameters (empennage size and shape, “pigtail tether”length, etc.) to ensure optimal energy transfer in the expected range of wind speeds.

• Study the various energy transfer pathways, including the two principal approaches to ex-tracting fluid energy:

– ambient excitation of an unstable mode into a limit cycle oscillation, or

– low control-energy excitation of a dynamically stable mode which is forced by the ambientflow.

• Investigate fluid-structure interactions to design against structural failures.

• Review and project the state of the art in energy generation and storage

8.2.3 Docking Station Concept Development

Development of the anchor and docking station concept will continue along the following avenues.

• Perform a careful preliminary design, respecting deployment limitations and anticipating andincorporating hardware requirements that include:

– energy storage components,

– hardware for two-way Titan-to-orbiter and/or Titan-to-Earth communications,

– hardware for inductively charging and communication with SCALARS vehicles,

– soil science equipment as dictated by science mission requirements.

• Review and project the state of the art in autonomous vehicle docking in dynamic environ-ments (such as shallow-water underwater vehicle docking) and related enabling technologies(such as active vision-based control).

• Identify and compare alternative concepts.

53

A Operating Environment on Titan

Many of the following properties were taken from Appendix A of [35].

Temperature. In the lowest few kilometers of the atmosphere (< 100 km), the temperature varieslinearly with altitude from a surface temperature around T0 = 94 K. Modulo small variations, thetemperature is

T = T0 − 1.15h K,

where h is altitude in kilometers. Above 100 km, the temperature is around 170 K.

Density. For low altitudes (between zero and forty kilometers), Lorenz suggests using

ρ = 10[A+Bh+Ch2+Dh3+Eh4] kg/m3,

where the coefficients are

A = 0.72065 B = −1.28873 × 10−2

C = −3.254 × 10−4 D = 2.50104 × 10−6

E = 6.43518 × 10−5

Pressure. The pressure at a given altitude can be determined from the ideal gas law

P = ρRT

where, approximating the atmosphere as pure nitrogen, we have

R =R

mN2

=8.3144 J/mol K

28.01 g/mol

1000 g

1 kg= 296.84

N m

kg K.

Viscosity. Again, approximating the atmosphere as pure nitrogen, the dynamic viscosity is givenby the equation

µ = 1.718 × 10−5 + 5.1 × 10−8(T − 273) Pa s,

where the temperature T is given in Kelvins. Thus, at Titan’s surface, the viscosity is about8 × 10−6 Pascal-seconds. Thus, the viscosity is roughly half that of Earth’s atmosphere. (Note:µ = νρ where ν is the kinematic viscosity.)

Figure 41 shows the behavior these four principal properties in the lowest 10 kilometers of Titan’satmosphere. It is evident from the plot that, for engineering purposes, it may be sufficient toconsider the lower atmosphere as having constant density and constant viscosity. In some cases,constant pressure may also be an appropriate assumption.

Speed of Sound. The specific heat of nitrogen at constant pressure is

cp = 1.04J

g K.

Given R above, we compute

cv = cp − R = 0.74316J

g K.

54

80 90 1000

1

2

3

4

5

6

7

8

9

10

T (K)

Alti

tude

(km

)

0 10 200

1

2

3

4

5

6

7

8

9

10

ρ (kg/m3)

0 2 4

x 105

0

1

2

3

4

5

6

7

8

9

10

P (Pa)6 8 10

x 10−6

0

1

2

3

4

5

6

7

8

9

10

µ (Pa s)

Figure 41: Atmospheric properties of Titan below 10 kilometers.

Thus, the ratio of specific heats is

γ =cp

cv= 1.40

(which is identical to the ratio of specific heats for air). The speed of sound at the surface of Titanis therefore

a =√

γRT

=

√

(1.40)

(

296.84N m

kg K

)

(94 K)

= 197.6m

s.

55

References

[1] Visionary challenges of NASA strategic enterprises for the NASA Institute for Advanced Con-cepts. Technical report, NASA, January 2004.

[2] B. Amett. The nine planets: A multimedia tour of the solar system.URL: www.nineplanets.org.

[3] Anonymous. Professor plans flying power station. BBC News Online, Science and Technology,March 2001. URL: http://news.bbc.co.uk/1/hi/sci/tech/1248068.stm.

[4] D. Cowan B. Sanders and L. Scherer. Aerodynamic performance of the smart wing controleffectors. Journal of Intelligent Material Systems and Structures, 15:293–303, 2004.

[5] R. C. Bendlin. Development of a fiber optic aerostat tether. In 11th AIAA Lighter-Than-AirSystems Technology Conference, Clearwater Beach, FL, May 1995. AIAA Paper No. 95-1618.

[6] M. B. E. Boslough. Autonomous dynamic soaring platform for distributed mobile sensorarrays. Technical Report SAND2002-1896, Sandia National Laboratory, Albequerque, NM,June 2002. Unlimited release.

[7] S. W. Bougher, D. M. Hunten, and R. J. Phillips, editors. Venus II. University of ArizonaPress, Tucson, AZ, 1997.

[8] F. Bullo, N. E. Leonard, and A. D. Lewis. Controllability and motion algorithms for underac-tuated Lagrangian systems on Lie groups. 45(8):1437–1454, 2000.

[9] D. Cadogan, T. Smith, F. Uhelsky, and M. MacKusick. Morphing inflatable wing developmentfor compact package unmanned aerial vehicles. In AIAA Conference, number 1807, 2004.

[10] C. F. Chyba1, W. B. McKinnon, A. Coustenis, R. E. Johnson, R. L. Kovach, K. Khurana,R. Lorenz, T. B. McCord, G. D. McDonald, R. T. Pappalardo, M. Race, and R. Thomson.Europa and Titan: Preliminary recommendations of the Campaign Science Working Group onPrebiotic Chemistry in the Outer Solar System. In Lunar and Planetary Science Conference,March 1999.

[11] M. L. Cosmo and E. C. Lorenzini. Tethers In Space Handbook. NASA Marshall Space FlightCenter, third edition, 1997.

[12] O. Danielsson, E. Sjostedt, K. Thorburn, and M. Leijon. Simulated response of a linear gener-ator wave energy converter. In 14th International Offshore and Polar Engineering Conference& Exhibition, 2004.

[13] R. E. Davis, C. C. Eriksen, and C. P. Jones. Autonomous buoyancy-driven underwater glid-ers. In G. Griffiths, editor, Technology and Applications of Autonomous Underwater Vehicles,volume 2, chapter 3. Taylor and Francis, 2002.

[14] S. M. Ehlers and T. A. Weisshaar. Static aeroelastic control of an adaptive lifting surface.Journal of Aircraft, 30(4):534–540, 1993.

[15] Ltd. Engineering Business. Research and development of a 150 kW tidal stream generator.Technical Report ETSU T/06/00211/00/REP, UK Dept. of Trade and Industry, SustainableEnergy Programmes, 2002. Publications URN No. 02/1400.

56

[16] C. C. Eriksen, T. J. Osse, R. D. Light, T. Wen, T. W. Lehman, P. L. Sabin, J. W. Ballard,and A. M. Chiodi. Seaglider: A long-range autonomous underwater vehicle for oceanographicresearch. Journal of Oceanic Engineering, 26(4):424–436, 2001. Special Issue on AutonomousOcean-Sampling Networks.

[17] M. J. Fitzpatrick. APTUS: Application of tether united satellites. In AIAA/USU Conferenceon Small Satellites, 2001.

[18] David E. Glass. Development of a flexible seal for a 60 psi cryogenic pressure box. TechnicalReport NASA/CR-1998-207669, Analytical Services & Materials, Inc., Hampton, VA, May1998.

[19] J. G. Graver. Underwater Gliders: Dynamics, Control, and Design. PhD thesis, PrincetonUniversity, 2005.

[20] J. G. Graver, J. Liu, C. A. Woolsey, and N. E. Leonard. Design and analysis of an underwaterglider for controlled gliding. In Conference on Information Sciences and Systems, pages 801–806, 1998.

[21] M. Abdulrahim H. M. Garcia and R. Lind. Roll control for a micro air vehicle using activewing morphing. In AIAA 2003-5347, 2003.

[22] F. Hendricks. Dynamic Soaring. PhD thesis, University of California, Los Angeles, 1972.

[23] J.I. Lunine J. S. Kargel, S.K. Croft and J.S. Lewis. Rheological properties of ammonia-waterliquids and crystal liquid slurries: Planetological applications. Icarus, 89:93–112, 1991.

[24] S. A. Jenkins, D. E. Humphreys, J. Sherman, J. Osse, C. Jones, N. Leonard, J. Graver,R. Bachmayer, T. Clem, P. Carroll, P. Davis, J. Berry, P. Worley, and J. Wasyl. Underwaterglider system study. Technical report, Office of Naval Research, 2003.

[25] Z. Jiang and R. A. Dougal. Strategy for active power sharing in a fuel-cell-powered chargingstation for advanced technology batteries. In IEEE Power Electronics Specialists Conference,Acapulco, Mexico, 2003.

[26] K. D. Jones, C. J. Bradshaw, J. Papadopoulos, and M. F. Platzer. Improved performance andcontrol of flapping-wing propelled micro air vehicles. In 42nd Aerospace Sciences Meeting &Exhibit, Reno, NV, 2004. (AIAA) Paper No. 2004-0399.

[27] K. D. Jones, S. Davids, and M. F. Platzer. Oscillating-wing power generation. In ASME/JSMEJoint Fluids Engineering Conference, San Francisco, CA, 1999.

[28] K. D. Jones, K. Lindsey, and M. F. Platzer. Fluid Structure Interaction II, chapter An inves-tigation of the fluid-structure interaction in an oscillating-wing micro-hydropower generator,pages 73–82. WIT Press, Southampton, UK, 2003.

[29] K. D. Jones and M. F. Platzer. Numerical computation of flapping-wing propulsion and powerextraction. In 35th AIAA Aerospace Sciences Meeting, Reno, NV, 1997. (AIAA) Paper No.97-0826.

[30] S. P. Jones. Aerodynamic characteristics of a new aerostat with inverted-y fins. In 6th AIAALighter-Than-Air Systems Technology Conference, Norfolk, VA, June 1985. AIAA Paper No.85-0867.

57

[31] C. Lambert, A. Saunders, C. Crawford, and M. Nahon. Design of a one-third scale multi-tethered aerostat system for precise positioning of a radio telescope receiver. In CASI FlightMechanics and Operations Symposium, Montreal, Canada, April 2003.

[32] N. E. Leonard and J. G. Graver. Model-based feedback control of autonomous underwatergliders. Journal of Oceanic Engineering, 26(4):633–645, 2001. Special Issue on AutonomousOcean-Sampling Networks.

[33] K. Lindsey. A feasibility study of oscillating-wing power generators. Master’s thesis, NavalPostgraduate School, 2002.

[34] R. Lorenz and J. Mitton. Lifting Titan’s Veil. Cambridge University Press, New York, NY,2002.

[35] R. D. Lorenz. Post-Cassini exploration of Titan: Science rationale and mission concepts.Journal of the British Interplanetary Society, 53:218–234, 2000.

[36] R. D. Lorenz. Flight power scaling of airplanes, airships, and helicopters: Application toplanetary exploration. Journal of Aircraft, 38(2):208–214, 2001.

[37] R. D. Lorenz. The seas of Titan. Eos, 84(14):125, 131–132, 2003.

[38] J. C. Luby. Personal communication. March 2005.

[39] P. R. Blankinship M. D. Feezor, F. Y. Sorrell and J. G. Bellingham. Autonomous underwa-ter vehicle homing/docking via electromagnetic guidance. Journal of Oceanic Engineering,26(4):515–521, 2001.

[40] J. Moum, D. Caldwell, J. Nash, and G. Gunderson. Observations of boundary mixing over thecontinental slope. Journal of Physical Oceanography, 32:2113–2130, 2002.

[41] M. A. Mueller and N. J. Baker. A low speed reciprocating permanent magnet generatorfor direct drive wave energy converters. In International Conference On Power Electronics,Machines And Drives, Bath, UK, 2002.

[42] M. A. Mueller, N. J. Baker, and P. R. M. Brooking. Electrical power conversion in direct drivewave energy converters. In 5th European Wave Energy Conference, 2003.

[43] J. E. Murray, J. W. Pahle, S. V. Thornton, S. Vogus, T. Frackowiak, J. Mello, and B. Norton.Ground and flight evaulation of a small-scale inflatable-winged aircraft. Technical ReportTM-2002-210721, NASA, 2002.

[44] R. A. Murthy. Dynamics of tethering cables for a flying electric generator. Master’s thesis,University of Western Sydney, 2000.

[45] W. J. Ockels. Laddermill: A novel concept to exploit the energy in the airspace. AircraftDesign, 4:81–97, 2001.

[46] T. Omholt. A wave-activated electric generator. In OCEANS ’78, pages 585–589, Washington,DC, December 1978.

[47] Matthew W. Orr. Aerodynamic properties of the inboard wing concept. Master’s thesis,Virginia Polytechnic Institute and State University, 2000.

58

[48] E. Pendleton, P. Bessette, P. Field, G. Miller, and K. Griffin. Active aeroelastic wing flightresearch program: Technical program and model analytical development. Journal of Aircraft,37(4):554–561, 2000.

[49] H. Polinder, M. Damen, and F. Gardner. Linear PM generator system for wave energy con-version in the AWS. IEEE Transactions on Energy Conversion, 2004.

[50] C. C. Porco and et. al. Imaging Titan from the Cassini spacecraft. Nature, 434:159–168, 2005.

[51] N. Forrester T. Austin R. Goldsborough B. Allen R. Stokey, M. Purcell and C. von Alt. Adocking system for REMUS, an underwater vehicle. In MTS/IEEE OCEANS ’97, October1997.

[52] G. Riegler, W. Riedler, and E. Horvath. Transformation of wind energy by a high-altitudepower plant. Journal of Aircraft, 7:92–94, 1983.

[53] M. Robertson-Tessi and R. Lorenz. Exploring Titan.URL: www.lpl.arizona.edu/ rlorenz/titanart/titan1.html.

[54] D. C. Rye, J. Blackler, and B. W. Roberts. The stability of a tethered gyromill. In 2ndTerrestrial Energy Systems Conference, Colorado Springs, CO, December 1981. AIAA PaperNo. 81-2569.

[55] D. D. Moerder S. C. Beeler and D. E. Cox. A flight dynamics model for a small glider inambient winds. Technical Report NASA/TM-2003-212665, NASA Langley Research Center,Hampton, VA, November 2003.

[56] S. Briest S. Cowen and J. Dombrowski. Underwater docking of autonomous undersea vehicleusing optical terminal guidance. In MTS/IEEE OCEANS ’97, October 1997.

[57] G. Sachs. Minimum shear wind strength required for dynamic soaring of albatrosses. Ibis,147:1–10, 2005.

[58] B. Sanders, F. E. Eastep, and E. Forster. Aerodynamic and aeroelastic characteristics of wingswith conformal control surfaces for morphing aircraft. Journal of Aircraft, 40(1):94–99, 2003.

[59] E. M. Schuch. Towfish design, simulation and control. Master’s thesis, Virginia PolytechnicInstitute & State University, 2004.

[60] C. Schultz and C. A. Woolsey. An experimental platform for validating internal actuatorcontrol strategies. pages 209–214, 2003.

[61] J. Scruggs and D. Lindner. Smart Structures and Materials 1999: Smart Systems for Bridges,Structures, and Highways, chapter Active energy control in civil structures, pages 194–205.SPIE.

[62] S. A. Shaffer, D. P. Costa, and H. Weimerskirch. Behavioural factors affecting foraging effortof breeding wandering albatrosses. Journal of Animal Ecology, 70:864–874, 2001.

[63] J. Sherman, R. E. Davis, W. B. Owens, and J. Valdes. The autonomous underwater glider“Spray”. Journal of Oceanic Engineering, 26(4):437–446, 2001. Special Issue on AutonomousOcean-Sampling Networks.

59

[64] M. L. Spearman and K. M. Feigh. A hybrid airship concept having twin-hulls and an inboard-wing. In AIAA 99-3914, 1999.

[65] R. F. Stengel. Flight Dynamics. Princeton University Press, Princeton, NJ, 2004.

[66] W. R. Thompson and C. Sagan. Organic chemistry on Titan: Surface interactions. In Sym-posium on Titan, September 1992.

[67] T. Trapp. Stingray - tidal power generator. In Wave and Tidal Technology Symposium, 2004.

[68] Unknown. Robot crosses the Gulf Stream underwater. Robotics Daily, December 2004.URL: www.roboticsdaily.com.

[69] S. Vogel. Life in Moving Fluids: The Physical Biology of Flow. Princeton University Press,second edition, 1994.

[70] D. C. Webb, P. J. Simonetti, and C. P. Jones. SLOCUM: An underwater glider propelled byenvironmental energy. Journal of Oceanic Engineering, 26(4):447–452, 2001. Special Issue onAutonomous Ocean-Sampling Networks.

[71] C. A. Woolsey. Reduced Hamiltonian dynamics for a rigid body/mass particle system. Journalof Guidance, Control, and Dynamics, 28(1):131–138, 2005.

[72] C. A. Woolsey and N. E. Leonard. Moving mass control for underwater vehicles. pages 2824–2829, 2002.

[73] L. A. Young, E. W. Aiken, V. Gulick, R. Manicinelli, and G.A. Briggs. Rotorcraft as Marsscouts. In 2002 IEEE Aerospace Conference, Big Sky, MT, 2002.

60

Documents

1010 woolsey[1]