JAMES A. CUTTS, KERRY T. NOCK, JACK A. JONES, GUILLERMO RODRIGUEZ AND J. BALARAM Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, 91109 U.S.A.
Abstract. Planetary aerobots are a new type of telerobotic science platform that can fly and navigate in a dynamic 3-dimensional atmospheric environment, thus enabling the global in situ exploration of planetary atmospheres and surfaces. Aerobots are enabled by a new concept in planetary balloon altitude control, developed at JPL, which em- ploys reversible-fluid changes to permit repeated excursions in altitude. The essential physics and thermodynamics of reversible-fluid altitude control have been demonstrated in a series of altitude-control experiments conducted in the Earth's atmosphere, which are described. Aerobot altitude-control technology will be important in the exploration of seven planets and satellites in our solar system. Three of these objects Venus, Mars, and the Saturnian satellite Titan--have accessible solid surfaces and atmospheres dominated by the dense gases nitrogen or carbon dioxide. They will be explored with aerobots using helium or hydrogen as their primary means of buoyancy. The other four planets--Jupiter, Saturn, Uranus, and Neptune--have deep atmospheres that are predominantly hydrogen. It may be possible to explore these atmospheres with aerobots inflated with atmospheric gas that is then radiatively heated from the hotter gaseous depths below. To fulfill their potential, aerobots to explore the planets will need autonomous state estimators to guide their observations and provide information to the altitude-control systems. The techniques of acquiring these data remotely are outlined. Aerobots will also use on board altitude control and navigation systems to execute complex flight paths including descent to the surface and exploiting differential wind velocities to access different latitude belts. Approaches to control of these systems are examined. The application of aerobots to Venus exploration is explored in some detail: The most ambitious mission described, the Venus Flyer Robot (VFR), would have the capability to make repeated short excursions to the high-temperature surface environment of Venus to acquire data and then return to the Earth-like upper atmosphere to communicate and recool its electronic systems. Finally a Planetary Aerobot Testbed is discussed which will conduct Earth atmospheric flights to validate autonomous-state-estimator techniques and flight-path-control techniques needed for future planetary missions.
Keywords: balloons, mobile robots, planetary exploration, navigation, aerobot
1 Introduction
The exploration of the solar system has developed in several phases beginning with planetary flybys, or- biters, atmospheric probes, landers, and mobile vehi- cles that operate on the surfaces and in atmospheres. For the most accessible planetary bodies--the Moon, Venus, and Mars--we have begun the phase of mobile exploration of the surface and atmosphere (Robertson, 1994). The use of telerobotic and autonomous aerove- hicles or aerobots and their use in planetary exploration is described along with the plans to demonstrate the needed technologies.
The earliest solar system exploration missions were flyby missions that made no attempt to orbit or land on the targeted object. Typically, flyby missions conduct observations for a few days around closest approach.
Later, using more sophisticated technology, orbiter missions were developed that observed the planet for months or years from close range, acquiring detailed maps of the surface and characterizing diurnal and sea- sonal variations in any atmosphere. To date, orbital missions have been carried out only for the Moon, Mars, and Venus. An orbital "tour" of the Jupiter satellite system is planned on the Galileo mission be- ginning in December 1995 and of the Saturn system in 2004. Surface landers and atmospheric probes have conducted much closer range observations. Landers have now been placed on the surfaces of the Moon, Mars, and Venus, obtaining images and geochemi- cal and atmospheric data about the immediate regions around the landing sites. Atmospheric entry probes are planned for Jupiter on the Galileo mission in 1995 and for Saturn's large moon, Titan, in 2004 on the
262 Cutts et al.
ESA Huygens probe to be deployed from the Cassini orbiter.
In the early 1970s, during the Apollo program, mo- bile surface exploration began when the U.S. deployed rovers to the lunar surface that were used by the as- tronauts. In the same time period, the Soviet Union carried out unmanned rover missions to the lunar sur- face. To date, no rover mission has been successfully carried out to any body other than the Moon, although the U.S. currently has firm plans to launch a Microrover Flight Experiment (MFEX) to Mars as part of the Mars Pathfinder mission in 1996. In addition, Russia is de- veloping its own Mars rover, Marsokhod, which will fly to Mars at a later date.
In some respects, mobile atmospheric exploration of the planets is ahead of mobile surface exploration. In 1985, the Soviet Vega mission successfully deployed two balloons into the upper atmosphere of Venus. A Soviet-French-American experiment tracked these bal- loons for two days on the side of the planet visible from Earth. They floated at an altitude of 54 km, de- termining wind velocities, characterizing atmospheric turbulence, and measuring solar insolation. A French- Russian team has been developing the Mars Aero- stat mission. This balloon system will be equipped with imaging cameras and meteorological and geochemical sensors.
2 Background
At the Jet Propulsion Laboratory (JPL), we are now involved in planning and developing technology for the next phase of planetary exploration using buoy- ant vehicles. This phase will draw on the technical experience of earlier missions but will employ tele- robotic and autonomy technologies to control motion in all three dimensions. There are significant parallels in these systems to the capabilities needed for mobile surface vehicles. However, there are also significant new challenges in atmospheric exploration that demand distinctly different approaches.
In this paper, we introduce the concept of robotic aerovehicle or aerobot as a powerful new approach to in situ planetary exploration. We distinguish an aerobot from a conventional balloon when it has one or more of the following four characteristics:
1. The ability to autonomously determine its position, altitude, and velocity without intervention from the ground or by a support spacecraft.
2. The means of executing cyclical altitude variations about a mean altitude in the atmosphere.
3. The capability of controlling altitude and executing a designated flight path within the atmosphere.
4. The capability of landing at a designated surface lo- cation.
The original motivation for developing this new class of buoyant vehicle was to advance the exploration of Venus. Following the exploration of the surface of Venus by short-lived Soviet landers and the Vega bal- loons, JPL carried out the Magellan mission, which mapped the surface of Venus using radar sensors. The radar revealed a surface with a great variety of struc- tural and volcanic features. There has been no clear pathway, however, to follow up the Magellan mission with a long-lived in situ mission.
Venus, Earth's estranged sister planet, has a dense atmosphere exceeding 92 bars in pressure and sur- face temperatures in excess of 460°C (733 K). Its sur- face is obscured from view at visible wavelengths by high-altitude haze and clouds as well as the molecu- lar scattering of the clear atmosphere beneath. The Soviet Venera landers were able to function for less than two hours exposed to the high-temperature envi- ronment on the Venus surface. With advanced ther- mal techniques and the use of vacuum insulation, it may be possible to extend surface lifetime to a few days. Much longer lived systems, however, will require radioisotope power and temperature control systems which will be costly and present Earth environmental concerns. For the same reasons, rovers appear even less practical at this time and have not been the subject of serious study.
An aerobot on the other hand can turn the environ- mental challenges of Venus to advantage. The Venus Flyer Robot (VFR) concept, conceived at JPL in 1993, could make brief excursions to the hot surface envi- ronment of Venus to acquire data and return to higher altitudes to cool down and telemeter those data to an orbiting relay station or directly to Earth. This con- cept takes advantage of new technologies in lightweight and low-power electronics and instruments that were used in the microrover systems also developed at JPL and in the MFEX/Pathfinder mission. However, in the case of the Venus aerobot, the attributes of low power and mass are even more critical.
The need for a lightweight payload and control sys- tem on an aerobot is obvious. The need for low power is also apparent; but for Venus exploration, it has a signif- icant new dimension. The lifetime of a thermally insu- lated gondola in the Venus lower atmosphere is limited
Planetary Exploration by Robotic Aerovehicles 263
not only by heat leaks from the high-temperature envi- ronment, but also by power dissipated by the electron- ics. The power required for information acquisition systems can be reduced substantially. However, the power for communications systems is beginning to approach theoretical limits and must be much larger. Hence a strategy of acquiring data near the surface and telemetering it from a higher altitude, where it is cooler, makes practical sense.
Although the original motivation for the aerobot was for Venus exploration, these vehicles are becom- ing recognized as powerful tools for exploration of all planets with substantial atmospheres. While the short- lived entry probes to be deployed on the Cassini and Galileo missions sample the planetary atmosphere at only one place and one time, long-lived aerobots can circumnavigate the planet many times, change altitude and access different latitude zones. Planetary aerobots represent the same kind of advance in exploration po- tential over single-shot entry probes that planetary or- biters bear to planetary flyby spacecraft. They are also engaging mission concepts that will capture the imag- ination of the public by their ability to explore hitherto inaccessible regions of our solar system.
In this article, we will discuss the opportunities for using telerobotic and autonomously controlled aero- bots that are capable of vertical mobility in atmospheres using altitude-control systems. We will also examine the use of prevailing wind patterns to enable global ex- ploration. We will then consider the challenges of con- trolling the flight path for precision landing of aerobots. Emphasis is placed on approaches to autonomous navi- gation for achieving desired latitudes and longitudes in planetary atmospheres using on-board flight dynamic models, real-time sensory perception (surface topog- raphy, balloon state, and atmospheric conditions), and occasional independent position updates.
A planetary aerobot testbed vehicle is also described that will conduct a series of terrestrial technology demonstrations that; (1) move gradually from remote control of the robotic vehicle to fully autonomous alti- tude change and landings; and (2) achieve increasingly long-range mobility from widely separated launch and landing sites (first predicted sites followed by design- ated sites).
3 Aerobot Mobility
Eight solar system bodies have sufficient atmosphere for exploration with buoyant vehicles. They include
the four major planets--Jupiter, Saturn, Uranus, and Neptune; the three terrestrial planets--Earth, Venus, and Mars; and Titan, the satellite of Saturn. Aerobots are lighter-than-air vehicles that include a primary buoyancy system for supporting the mass of the sci- entific payload, communications, and a closed-system reversible-fluid buoyancy-control system. In this sec- tion, we describe the primary buoyancy and buoyancy- control approaches applicable to exploration of the eight solar system targets. A series of Earth demon- strations of reversible-fluid buoyancy control that were conducted during the last year are also discussed. In addition, the method by which buoyancy controls alti- tude and enables horizontal mobility is described.
3.1 Primary Buoyancy' Systems
Although heavier-than-air vehicles have been consid- ered for planetary exploration, lighter-than-air vehi- cles have clear advantages. First, they are much more suitable for long-duration flight because they can re- main aloft without consuming energy. Their lift derives from the displacement of the atmosphere (Table 1) by a lighter gas in a balloon envelope. Heavier-than-air vehicles, in contrast, must generate lift by consum- ing significant amounts of energy. For a long-duration aerovehicle, a renewable source of energy such as so- lar power is needed. Solar-powered aerovehicles have been examined for operation at Mars. Solar-powered aircraft might also operate at Venus above the cloud layers but would be unable to penetrate the deep atmo- sphere, which is the region of primary interest. For the outer planets and Titan, the low solar intensities render solar-powered aircraft entirely impractical.
There are two general approaches to a primary buoy- ancy system for lighter-than-air systems: inflation with a gas that is inherently less dense than the surrounding atmosphere or inflation with gas from the surrounding atmosphere whose temperature is raised to lower its density. Mars, Venus, and Titan all have solid surfaces with atmospheric pressures ranging from less than 1% of that of the Earth to 100 times higher (at Venus). Like Earth, these atmospheres are comprised primarily of carbon dioxide or nitrogen, which have comparatively large molecular weights that determine their inherent density. Inflating the primary balloon with a very low- density gas such as helium or hydrogen is practical. Ammonia or water are also satisfactory in the dense, high-temperature Venus environment.
For the outer planets, whose compositions are domi- nated by hydrogen, the light gas approach involves
264
Table 1.
Cutts et al.
Means of primary buoyancy.
Surface pressure Surface temperature Primary Molecular Means of primary Planet (Bars) (K) composition weight buoyancy
Mars <.01 200-250 CO2 44 H2, He Earth 1 ~300 N2 28 H2 or He Venus 100 750 CO2 44 H2, He, H20 or NH 3 Titan 1.5 90 N2 29 H2 or He
Jupiter 10' 300 t H2 2.2 Pure H2 or Saturn 10" 300 t H2 2.0 "Hot air balloon" or Uranus 10" 300 t H2 2.3 "Infrared balloons" Neptune 10" 150 t H2 2.3
*These planets have no surface; temperatures are shown for the 10 bar-level. tEstimates based on model extrapolations. Jupiter will be measured in 1995, Saturn in 2008.
delivering very large amounts of pure hydrogen to fill the balloon. The alternative approach of inflating the balloon with gas from the surrounding atmosphere and raising the gas temperature to lower its density appears feasible and may be more efficient. One approach to ac- complishing this is to use infrared radiation upwelling from the hotter regions of the atmosphere to heat the balloon. The outer part of the upper hemisphere of the balloon is made highly reflective to infrared radiation; the inner surface highly absorptive. As a result, in- flared radiation from below is absorbed by the balloon and raises the temperature of the enclosed gas. This concept, which originated in France as the infrared Montgolfiere balloon, has been demonstrated on the Earth in a series of flights (Malaterre, 1993). Explo- ration of the outer planets to pressures greater than 10 bars, where temperatures are in the vicinity of 300 K appear feasible; measurements made by the Galileo entry probe when it enters Jupiter later this year may confirm its feasibility for that planet.
3.2 Buoyancy Control Using Reversible Fluids
Aerobots designed for long-term operation in plane- tary atmospheres require methods of altitude control that are both energy efficient and involve minimal ex- penditure of consumables. A closed-system reversible- fluid balloon uses the naturally occurring atmospheric temperature variation with altitude to drive a heat en- gine, providing the mechanical energy needed for alti- tude change.
A reversible fluid is either a gas or a liquid, depend- ing on pressure and temperature. It is this phase change which can be used to control the buoyancy of a bal- loon system. When the reversible fluid is in the gas phase, the balloon has a lower average density than the
surrounding atmosphere thus providing a net increase in lift. Conversely, when the fluid is in the liquid phase, the balloon has a higher average density than the sur- rounding atmosphere thus providing a negative lift.
Reversible fluid balloons were first proposed by the French (Rougeron, 1969) in order to stabilize a bal- loon at high altitude without the use of superpressure balloon envelopes, which can be heavy and difficult to build. A two-balloon system was suggested in which a main balloon, not quite buoyant enough to ensure equi- librium, is attached to a stabilizing balloon containing a reversible fluid. At low altitudes, where tempera- tures and pressures are higher, the reversible fluid turns to vapor. This phase change adds to the lift, causing the balloon system to ascend. At high altitudes, the reversible fluid condenses, thus taking away lift and causing the balloon to descend. The French (Romero, 1980, 1981) developed balloon performance models and studied the operation of such balloons at Venus us- ing a number of different main balloon buoyancy gases and reversible or stabilization fluids. Those studied in- clude water with octane, ammonia with cyclohexane, and helium with toluene. Note that with the use of a primary balloon containing a light gas, the stabilization fluid in its gaseous state can be denser than the Venus atmosphere.
In 1981, the Russians (Moskalenko, 1981) devel- oped an idea for a single reversible-fluid balloon which could descend to the Venusian surface. This concept employed a single balloon fitted either with water alone or water with various other fluids for buoyancy. These fluids (all lighter than CO2 when vapor) were origi- nally proposed in order to eliminate the high penalty of carrying helium to Venus in heavy pressure tanks. In addition, it appeared to be possible to change flight alti- tude utilizing the difference in temperatures at different
Planetary Exploration by Robotic Aerovehicles 265
Fig. 1.
eo / _ _ l _ .
-
''-"°" \
I REVERSIBLE FLUID "<~'/ / / )k | | oI'AI RAT'°"cuRvE [ i _
0 ° 500 ° TEMPERATURE, °C TIME
Venus balloon equilibrium altitude and cyclic motion.
altitudes. Moskalenko recognized that such balloons would oscillate about equilibrium altitudes, the exact altitude depending upon fluid mixtures. By trapping the condensed fluid in a pressure vessel before it passed below its equilibrium altitude, a balloon could be made to descend all the way to the surface. Once on the sur- face, a "throttle" valve would release this gas into the balloon so the system could reascend. In this fashion, such a balloon would utilize the "generous" energy source available in the atmosphere itself to rise in al- titude, relying on gravity to return the vehicle to the surface. In fact, Moskalenko even suggested extract- ing some of this energy by means of wind turbines to generate electrical energy as the system bobs up and down. Because these were single balloons rely- ing primarily on water, the equilibrium altitudes were about 40 km, still in a relatively high-temperature en- vironment. There appears to have been no follow up to Moskalenko's work in the early 1980s in Russia. However, Venus balloon studies and experiments on the phase transition of liquids to vapor have been de- scribed by the Japanese (Nishimura, 1990), followed by an attempt to fly this type of stabilization balloon at Earth (Akiba et al., 1992), with mixed results.
In 1993, JPL began exploring concepts for achiev- ing both the high, cool altitudes for balloon oscillation, and the ability to trap reversible fluids for balloon de- scent to the surface where scientific observations can be made (Jones, 1995). The first concept considered by JPL used a two-balloon system similar to the French concept, with the main balloon filled with helium and the secondary balloon filled with a reversible fluid like methylene chloride (heavier than CO2 as a vapor). The system would be designed to be neutrally buoyant when about half the reversible fluid is condensed. Such a balloon would exhibit forced oscillations about an
equilibrium altitude of about 56 km at Venus. Descent would be initiated by trapping the methylene chloride in a pressure vessel before the balloon descends below 56 kin. The amplitude of oscillation is expected to be a few kilometers above and below the equilibrium al- titude. Vaporization is enhanced below 56 km by use of a heat exchanger design, which prevents the system from dipping too low every cycle. The high-altitude oscillation phase can be used to generate electrical en- ergy from solar cells, transmit data to Earth, and re- cool both electronic packages and phase-change-based heat sink materials (e.g. wax) for later descents to the surface. Opening a valve releases the gas to refill the secondary balloon causing the system to return to alti- tude. Figure 1 illustrates the concept of dual-balloon, reversible-fluid altitude control.
Studies at JPL for Venus aerobots are now con- sidering water/ammonia mixtures in dual, and-single balloon configurations. A single water/ammonia bal- loon has significant advantages in terms of total mass delivered to Venus due to the absence of helium pres- sure tanks as suggested by the early Russian work. It has a further advantage. Taking into account the par- tial pressures of each buoyant gas component, a wa- ter and ammonia system could actually have a much higher altitude of oscillation than one would expect when considering their fluid properties independently of each other. In other words, when the water vapor is mixed with ammonia (always a vapor at Venus), the balloon must rise well above 42 km in order to con- dense all the water vapor to liquid, perhaps to 60 km. However, once the water is liquid, and not influenced by the ammonia vapor, it will not vaporize until the system reaches 42 kin. A heat exchanger is employed to enhance the vaporization below 42 km so the vehi- cle does not descend on a regular basis to depths where
Fig. 2. Comparison of venus and earth atmospheres note, temper- ature units are different than those in Fig. 1.
it is hot and the pressures are high. A passive free- flight Venus balloon technology experiment called B al- loon Experiment on Venus (BEV), without the robotic controls, has been proposed by JPL (DiCicco, 1995), which could demonstrate high-altitude cycling of this water/ammonia system at high altitude.
3.3 Flight Tests of Dual-Balloon, Reversible-Fluid Systems
Because the Venus high-altitude atmosphere is simi- lar to the Earth's in temperature and pressure, we can demonstrate reversible-fluid altitude control tech- nology in our own atmosphere. Figure 2 is a Van't Hoff plot showing the Venus high-altitude atmosphere model along with a typical profile for Earth's lower at- mosphere. During 1993 and 1994, JPL carried out a series of four flight demonstrations of reversible fluid control systems. These Altitude Control Experiment (ALICE) tests were performed with purely passive dual-balloon systems using helium and Freon Rl14 (Nock, 1995).
In the ALICE project, a very small (total system mass <3 kg) two-balloon system is being tested. The primary balloon is filled with helium and the buoyancy- control balloon is filled with a commercial refriger- ant called R114, which is about 7 times heavier than air. At Earth atmospheric conditions, R114 becomes a liquid above 4000 to 7000 meters depending on weather conditions. A typical ALICE balloon sys- tem includes a helium balloon, radiosonde, and a R114 buoyancy-control balloon. Both rubber latex and clear
polyethylene helium balloons have been flown. The ra- diosonde is a slightly modified commercial unit which provides an 8-channel capability for balloon telemetry in addition to measuring normal pressure, ambient tem- perature, and humidity. The R 114 balloon or bag (since it hangs from the system) is constructed from clear, 2- mil-thick, seamless, 3-feet-wide lay-flat polyethylene film, which is heat sealed to achieve the proper bag configuration. In the most recent flights, the balloon system is fully instrumented to continuously monitor the temperatures of the helium gas and the R114 as it changes from a gas to a liquid. These temperatures are measured by very small (14 x 20 mil) thermistors, some of which are in protective gold-plated cages to reduce the effect of solar radiation on the temperature measurements.
Extensive balloon performance modeling has been carried out in the ALICE Project in order to charac- terize the thermodynamics and aerodynamics of the dual-balloon system in a given environment (Wu & Jones 1995). This modeling is based upon extensive experience gained in the NASA Scientific Ballooning Program (Needleman et al., 1993; Carlson et al., 1983).
The first two flights were launched during the day and employed standard 200 to 300-g rubber latex helium balloons. In both flights, the balloon ascent rates were seen to slow at a higher-than-expected con- densation altitude. However, the balloons did not exhibit oscillatory behavior. Extensive balloon ther- modynamic and aerodynamic modeling and balloon envelope thermodynamic parameter testing suggested several problems including higher-than-expected solar heating of the helium balloon (causing greater lift).
The third flight began after sunset in order to better decouple model parameters relating to effects of forced convection and drag. This third flight was identical to the first two flights except that the balloon system was fully instrumented to continuously monitor helium temperature and R114 temperature as it changed phase from gas to liquid. After reaching about 6500 m al- titude, the balloon descended as predicted by perfor- mance model estimates. Telemetry was lost at about 2600 m as the balloon system descended below a moun- tain range. Reascent before impact was predicted to be unlikely for this flight because the equilibrium altitude during the winter was only about 4000 m and the R114 bag did not incorporate a heat exchanger to facilitate liquid boiling•
The configuration for the fourth flight, ALICE 0/D is illustrated in Fig. 3. It had two new features, a 0.8- mil clear polyethylene helium balloon and an integrated
Planetary Exploration by Robotic Aerovehicles 267
m
5
~ H e l i u m
- 5 m V Bau°°n
m
ll
R a d i o s o n d e
Revers ible F lu id B a g ' ~
m m
m Strobe
1
Connector Fill Tube _ - -"~'-.,~ 3b~Gromet ( ~ b e ~ " - ~ ~
off and seal .....~3],~ a ~ - ,, f'~-- b , f o , , I ~.-_ ............................. I _ . . . .~ : ' . . 1
Feedthro~a (PE loop heat sealed at first 2
seals)
Fig. 3. ALICE 0/D flight configuration RTG denotes R114 temperature, gas RTL denotes R114 temperature, liquid PE indicates polyethylene.
heat exchanger built into the R114 bag to facilitate fluid boiling at low altitudes. The total balloon system mass was about 3 kg (see Table 2 for a detailed mass break- down of this system). At 9 p.m. the night of July 24, 1994, the balloon system was filled with buoyant gases, assembled to its telemetry system, and released into the clear night sky from the foothills above JPL.
The ALICE 0/D balloon system was tracked throughout the night as it flew over the San Gabriel Mountains and Mojave Desert. It was sighted at day- break the next day at Daylight Pass on the northeast rim of Death Valley. Ground vehicle chase terminated at about 8:00 a.m. on July 25, 1994 as the balloon flew over Nellis Air Force Range and Nevada Nuclear Test Site. Data acquisition continued until loss of signal occurred at 11:20 a.m. July 25 as the balloon passed over the horizon. The 14 hours of tracking data contain valuable information on balloon dynamic and thermo- dynamic conditions. Four complete oscillations be- tween 5 and 9 km in altitude were recorded.
Figure 4 shows the mission profile from this ALICE 0/D flight conducted in the California and Nevada
Table 2. ALICE 0/D mass breakdown.
Equipment Mass, g
Helium balloon system 919 Helium balloon 815 Helium balloon fill plug 75 Helium gas temp. probe 29
R114 Bag assembly 206 Polyethylene bag 157 Temperature probes 36 Miscellaneous 13
Buoyancy gases 1409 Helium 409 R114 Freon 1000
Total balloon system mass 3006
desert; the thin line is actual data with the post-flight model fit shown as a bold line. The bottom altitude profile is the topography under the ground track. The
middle line shows estimated updrafts and down drafts during the flight used in the model fit. One complete oscillation occurred after sunrise, which should pro- vide important insights into the performance of this new design during the day.
Further flights of this purely passive system are be- ing planned that will incorporate an ultra-low-power imaging sensor to demonstrate navigation techniques applicable to other planets for which there are no ade- quate radio-positioning systems.
3.4 Using Altitude Control to Achieve Lateral Mobility
By controlling vertical mobility, aerobots can select altitudes where wind speeds and directions provide a wide range of horizontal mobility. This is espe- cially true at Mars and Venus where the occurrence of altitude-variable wind gradients enables near-global planetary access.
The Vega balloon experiments explored the Venus middle cloud layer at 50-55 km altitude. Vertical winds were found to be large (3 m/s) and variable, with tur- bulent episodes lasting about an hour. East-to-west average zonal winds of about 69.4 m/s for Vega 1 and 66.0 m/s for Vega 2 were detected. Both Vega bal- loons drifted about 11,000 km from the local midnight meridian into the late morning sky, carried by strong, predominantly zonal east-west winds.
Vega 1 initially encountered weak southward winds, which changed to northward winds later in the mission. These meridional (north-south) winds produced north- south displacements in the Vega 1 trajectory that never exceeded 50 km. The Vega 2 meridional winds were consistently northward with a mean velocity near 2.5 m/s (Crisp et al. 1990), which produced more than400 km of displacement toward the north pole.
The zonal (east-west) wind profiles as a function of altitude shown in Fig. 5 were obtained from the several Venus probes and landers (Moroz, 1994). These pro- files indicated that above 10 krn altitude there is a mono- tonic increase in zonal wind velocity with increasing altitude up to near the top of the clouds (~70 km). Observed variations among the different data sets are possibly due to variations in probe entry position, ac- cording to the local time of day on Venus, and the phase of a wave motion that circles the planet with a 4-day period. These variations illustrate the need for better information on the global atmospheric circulation and is the objective of a proposed Venus Multiprobe Mis- sion (VMPM) recently selected by NASA for Phase A studies.
One postulated model for the circulation of the at- mosphere of Venus (Schubert 1983) appears in Fig. 6. On Venus, the atmospheric and surface temperatures at the poles are very little different from those at the equator. Since Venus is a slowly rotating planet, the transport of heat from equator to pole is believed to
Planetary Exploration by Robotic Aerovehicles 269
Iv
G
E~try Pro6e R e s ~ z l t s : Z o n a l C o ~ n p o ~ z ~ t On/s) 1o ~ ~ 40 se 60 7e 90 90 ioo 11o ~2o
involve a series of Hadley cells. Hadley cells are circu- lating regions extending from equator to poles in which winds flow towards the pole at one altitude and away at another altitude (Houghton, 1977). Balloon flight ex- periments in the Venus winds will be needed to charac- terize the parameters of the circulation and will provide the knowledge necessary to use these winds for global mobility in more sophisticated surface reconnaissance missions in the deep atmosphere.
3.5 Vehicle Mobility-Augmentation Mechanisms
Although reversible-fluid altitude controls is an extraordinarily powerful technique, it has some limitations. One is its inability to modulate the rate of descent to a planetary surface except by evaporat- ing fluid. The use of variable geometry surfaces might
augment this capability by allowing an increase in nom- inal descent rate. A second limitation is that reversible fluid altitude control works only in planetary tropo- spheres where temperatures drop with increasing alti- tude and where reversible fluids can be used that change phase within the altitude range that is to be explored. This is the case for the tropospheres of Venus, Earth, Titan, and the outer planets. However, reversible-fluid altitude-control techniques cannot be used in planetary stratospheres where atmospheric pressures are low and the temperature changes very little with increasing alti- tude and can even increase. For example, the thin atmo- sphere of Mars has no troposphere, in effect prohibiting the use of reversible fluids for altitude control. Accord- ingly, other approaches to reversible altitude control are needed, such as reversible chemical reactions with the atmosphere.
270 Cutts et al.
CLOU HADL
Wl SUP[
Fig. 6. Postulated Venus global circulation model
Altitude control indirectly allows the control of horizontal longitudinal motion along the wind-driven path. Horizontal control tangential to the motion along the wind-driven path is more challenging to achieve but is not as critical as longitudinal control enabled by vertical mobility. Futuristic options (Vorachek, 1970) for lateral maneuverability include deployable surfaces like tethered kites and parafoils. Another pos- sible lateral control mechanism uses the Magnus ef- fect (Koshlyakov, 1984), in which a rotating body in a wind field provides motion perpendicular to the wind. The main drawback is the need for external moving parts.
3.6 Landing
Soft landing of science payloads on planet surfaces is achieved by means of a flexible and possibly robotic landing "snake" tethered below the vehicle. As the snake contacts the surface, it relieves the vehicle of some of its gravity load. This enables the gondola to "hover" at a fixed distance away from the surface with- out impacting it. Robotic snake concepts have been
conceived for Venus whiCh use ambient atmosphere in a pneumatic control system to achieve desired articula- tion (Hadaegh & Chirivella, 1994). Passive snake con- cepts applicable to Mars have been successfully tested in a series of balloon and ground tests (Anderson, 1991).
4 Mission Opportunities
There are opportunities for using aerobot technology at Venus, Mars, Titan, and the outer planets as well as terrestrial applications. In this section, we describe several aerobot mission concepts that take advantage of reversible-fluid balloon systems to provide cyclic altitude variation or active control of altitude. Alti- tude control provides a capability for brief sorties to high-temperature or radio-opaque atmospheric regions (important at Venus and outer planets), extension of latitude coverage by exploiting variations in the merid- ional component of wind with height, control of flight path for optimizing imaging of surface targets, and controlled descent to designated landing sites. The characteristics of prior, planned, and future missions
Planetary Exploration by Robotic Aerovehicles 271
involving balloons and aerobots are summarized in more detail below.
4.1 Venus
Venus is the only planet which has been explored with aerovehicles. It is an attractive target for future aer- obot missions because of its harsh, high-pressure, high- temperature environment. An aerobot can enable many kinds of observations which may be impractical to ob- tain by any other means.
4.1.1 Vega. The Vega-1 and Vega-2 balloons de- ployed in 1985 to obtain information about Venus atmospheric circulation (Science, 1986; Crisp et al., 1990) were entirely passive vehicles. The Vega bal- loons operated at a constant altitude of 54 km for about two days before radio contact was lost when the bat- teries failed. During this period, each balloon travelled about 11,000 km in an east-west direction relative to the Venus surface. They were tracked by radio but had no autonomous tracking capability or ability to respond to commands.
4.1.2 Balloon Experiment at Venus (BEV). An in- expensive flight demonstration of reversible-fluid al- titude control, called Balloon Experiment at Venus (BEV) has been developed as a possible piggyback to a NASA Discovery Program mission (DiCicco et al., 1995). Discovery is a new NASA program of moderate-cost solar system exploration missions. The BEV flight demonstration would refine on-board navi- gation designs by characterizing the robotic vehicle and on-board sensor operation. It would relay data acquired with a mapping sensor directly to Earth.
4.1.3 Venus Flyer Robot. This aerobot would incor- porate all four of the aerobot attributes discussed in Section 3. It would use autonomous navigation and control to enable repeated short observations of the Venus surface over long duration. The Venus Flyer Robot (VFR) would conduct remote-sensing visual and infrared-imaging observations from the middle atmo- sphere and make brief excursions to the surface to sam- ple the surface and near-surface atmosphere using a balloon envelope capable of operating at those tem- peratures (Yavrouian et al., 1995). To achieve global maneuverability, the VFR would exploit its altitude- control capabilities to access regions of the atmosphere with favorable north-south winds (Cutts et al., 1995). Using autonomous navigation capabilities and these
winds, VFR could move to particular sites of interest to make remote observations and to land.
A Venus Aerobot has been identified by a NASA science group as a high- priority candidate for the next mission to explore the Venusian surface. Maat Mons, a volcano 8 km above the mean surface level that is about 60 K cooler than the mean surface temperature, has been proposed as a site of scientific significance and accessibility.
4.2 Mars
Mars is likely to be the next planet after Venus to be explored with balloons. Simple, passive Mars bal- loons have been proposed by a joint Russian/French team as noted above and by others (Blamont, 1993; Zubrin, 1993). Aerobot vehicles using sophisticated telerobotic navigation approaches, have great potential at Mars, although the reversible-fluid approach to alti- tude control is impractical. Concepts for long-duration flight at Mars have already been developed at JPL and alternate long-lived altitude control systems using re- newable energy resources may be developed in future. Long-duration Mars aerobot missions would be able to survey much of the Martian surface and map available subsurface water resources if they exist. As with all in-situ missions, they would require the use of orbital data relay to achieve efficient data return and could use data links provided for the Mars Surveyor program, an ongoing program of Mars missions planned by NASA and JPL for the late 1990s and beyond that includes orbital and landed payloads.
4.2.1 Mars 1998 Aerostat. The Russian-French Martian Aerostat mission, as originally conceived for a 1994 launch, used a zero-pressure balloon and made diurnal visits to the surface as a result of the dramatic diurnal variations in the temperature of the atmosphere. There was no ability 1;o select where those landings oc- curred. When the mission was delayed to 1998, it was modified to avoid hazardous diurnal visits to the sur- face given the higher surface windspeeds expected for the 1998 opportunity. Because this modification in- volved venting helium and dropping ballast each day, the duration of the mission, is limited to a few days. At the end of the mission, the balloon is to initiate a high- risk descent to the surface for an important microwave sounding experiment in search of subsurface water. It will relay data to one of the Mars Surveyor orbiters, which are now being planned by the U.S.A. (Sirmian et al., 1995).
272 Cutts et aL
4.2.2 Mars Atmospheric Platform (MAP). Thepro- posed MAP mission (Zubrin, 1993) would use a super- pressure balloon to execute a constant-pressure altitude reconnaissance of Mars in order to acquire remote- sensing data. The superpressure balloon would retain lift at night and, as a result, the vehicle would not reach the surface but would complete many circumnaviga- tions of the planet in a period of a few months. MAP would have no means to control altitude and no au- tonomous navigation capability.
4.2.3 Future Mars Aerobots. These mission con- cepts, which could fly as early as 2001, would in- corporate autonomous navigation and conventional altitude-control subsystems to allow semi-controlled flight within the atmosphere for remote sensing, in situ sensing, and deployment of surface packages. The Mars Aerobot Experiment (MAX), would have the ability to autonomously measure position as it circum- navigated Mars and thereby image targets of opportu- nity. It would also have an altitude-control capability (ballast/or buoyant gas release) to allow modification of the ground track to improve the ability to acquire these targets. The Mars Flyer Robot (MFR), like the Venus Flyer Robot, would have the capability to exe- cute landings at designated target locations.
4.3 Titan
able to make extended stays at the surface provided the wind conditions were favorable.
At the distance of Titan, a communications relay is critical for a viable mission. The 1997 Cassini mis- sion offers an opportunity for a low-cost technology demonstration using the available Titan Probe commu- nications link. Cassini will be in Saturn orbit from 2004 to at least 2008. Frequent Titan flybys are planned, which enable a possible relay from a Titan aerobot to the Earth through the Cassini spacecraft. A small Titan aerobot could be launched in 2000 on a short trajectory timed to arrive at Titan shortly after Cassini begins its orbital operations around Saturn.
4.4 Outer Planets
Similar aerobot technologies may apply to the outer planets of Jupiter, Saturn, Neptune, and Uranus. These balloons could to use the infrared Montgolfier approach to achieving primary buoyancy using the internal radi- ated heat of the planet (Malaterre, 1993) as discussed in Section 3.1. A variety of options exist for altitude con- trol fluids. The primary motivation for an outer planet balloon is exploration of the deep atmosphere. The Jupiter Deep Atmosphere Balloon (JDAB) would pen- etrate to depths inaccessible to radio communications. Detailed feasibility studies of these types of balloon systems are yet to be performed.
Titan is the large moon of Saturn and the only satellite in the solar system with a substantial atmosphere that is primarily nitrogen and a surface which is believed to consist of ice continents and methane oceans below the perpetual haze layers. A Titan aerobot equipped with a reversible-fluid buoyancy-control system using argon (or a mixture of argon and nitrogen) as a reversible fluid would be able to repeatedly explore the surface.
Two mission concepts have been examined. The Balloon Experiment in the Titan Atmosphere (BETA), uses a passive reversible-fluid system like that used by the BEV to execute cyclic forced oscillations in al- titude. In the case of Titan, the equilibrium altitude can be made low enough for the aerobot to make un- controlled descents to the surface. The large excur- sions in temperature experience on Venus would not be encountered by a Titan aerobot. The Titan Flyer Robot (TFR), with capabilities similar to its Mars and Venus analogs, would have the capability for control- ling altitude on command or autonomously, modifying its flight path, and descending to designated surface features. Unlike the Venus Flyer Robot, it would be
4.5 Summary
A comparison of the objectives and capabilities of the aerobot missions to earlier balloon missions appears in Table 3. All the true aerobots possess a capability for autonomous state estimation during flight. They range in capability from those with passive (cyclic) altitude variations (BEV and BETA) to those able to actively control altitude and flight path (VFR, MAX, MFA, TFR, and JDAB). Of these vehicles, three-- VFR, MFR, and TFR have the further ability to per- form landings at designated locations for prescribed periods.
5 Navigation, Control, and Mission Planning
In the last section, different types of robotic capa- bility were described: the autonomous state estima- tion, cyclic altitude variation, altitude control, and controlled flight and landing. The term "aerobot" has been adopted to refer to a lighter-than-air vehicle with
Planetary Exploration by Robotic Aerovehicles 273
Table 3. Aerobot capabilities for planetary exploration.
Aerobot capabilities
Flight Landing
Target Autonomous Altitude Flight path Location Duration Body Mission Objective state estimator variation con t ro l control control
Venus V E G A Atmospheric circulation 1 No No No N/A N/A BEV Navigation experiment Yes Passive No N/A N/A
(cyclic) VFR Remote & in situ sensing Yes Active Yes Yes Yes
Mars CMA Remote surface sensing 2 No Passive No No No MAP Remote surface sensing 1 No No No N/A N/A MAX Remote surtace sensing, Yes Active Yes N/A N/A
small station deployment MFR Remote surface sensing,
in situ sampling Yes Active Yes Yes Yes Titan BETA Remote & in situ, Yes Passive No No No
surface sensing (cyclic) TFR Remote & in situ Yes Active Yes Yes Yes
surface sensing Outer JDAB Deep atmospheric Yes Active Yes N/A N/A
planets exploration
IThe original design of the CNES Martian Aerostat mission involved descents to the surface of MARS each night. Thus, it experienced cyclic (diurnal) altitude variations. For the FY'98 opportunity, it will land at the end of the Mission (see text). 2Both the VEGA and MAP use a superpressure balloon for altitude stabilization.
one or more of these four capabilities. In this section, we describe how these capabilities can be implemented, focusing primarily on aerobots that use the reversible- fluid approach to cyclic altitude variation and altitude control. A comparison of mobility, control and naviga- tion for planetary aerobots and planetary, rovers appears in Appendix A.
5.1 Autonomous State Estimator
An on-board Autonomous State Estimator (ASE) is needed for all of the future aerobot missions described in the last section. The state variables that would be estimated include position, altitude, velocity, and angu- lar velocity. Not all of these will be needed in a given aerobot mission and the accuracy required may vary from mission to mission. For those mission concepts that do not incorporate active altitude control (BEV and BETA), the ASE enables remote-sensing targets of op- portunity to be identified from an on-board prioritized target list. For those missions that have altitude control and the potential for modifying the flight path (VFR, MAX, MFR, TFR, and JDAB), it provides the essential data for navigating a desired flight path. For those mis- sions involving landings, additional capabilities will be needed for the terminal descent phase.
An ASE block diagram is shown in Fig. 7. Ra- dio metric position and velocity measurements can be made by observing the aerobot radio signal from the Earth or from a communication relay orbiter. These measurements will not usually be available continu- ously and the results must be communicated to the aerobot before they are useful. In addition, for radio metric measurements made on Earth, there will be a significant delay in measurement associated with the round-trip light time. Accordingly, an ASE relying on aerobot sensors, is necessary for timely and complete information for targeting and/or flight path control.
On-board sensors could include solar or star trackers, surface imagers, surface radar range/Doppler sensors, magnetic field sensors, and inertial sensors. Tracking an orbiter radio beacon from the aerobot (similar in concept to Earth-based GPS systems) to determine the aerobot's state variables, is another option. The sensors and associated signal-processing electronics must be compact and require very low power in order to be accommodated on the aerobot.
Output of the ASE is provided to the controller of the remote-sensing payload which enables the acquisition of targets of opportunity from a prioritized target set stored in the aerobot memory. If the aerobot has an altitude control capability (VFR, MAX, MFR, TFR,
274 Cutts et al.
Fig. 7.
E A R T H - B A S E D R A D I O M E T R I C ,
P O S I T I O N & V E L O C I T Y ,
FROM ON-BOARD SENSORS ' • Sun/Stm" Tracker • Surf , tee Images • Surface Range/Doppler Radar • Gyro/Aceelerometer • Inc l inometer • Magnet ic ~ e l d Direction • Radio Range/Doppler
A u t o n o m o u s state de te rmina t ion .
. . . . ! f [ I I I I I |
I I ! I I I
I I + l l A U T O N O M O U S
S T A T E E S T I M A T O R
S U R F A C E F E A T U R E D A T A B A S E
O R B I T E R - B A S E D ' R A D I O M E T R I C
P O S I T I O N & V E L O C I T Y
_~TO R E M O T E S E N S I N G I N S T R U M E N T S SEQUENCING • S e ~ o r P la t fo rm Al t l tude /AWtude • Balloon Ground Track Podt /on/Vdoci ty
T O G L O B A L T R A J E C T O R Y G E N E R A T O R • In i t ia l Locat ion and V d o d t y • Actual FllRht pa th
JDAB), the state variables will also be passed to the Global Trajectory Generator (GTG) discussed later.
Selection of on-board sensors for position and ve- locity measurement will depend on the frequency of independent measurements from the ground or from the communications orbiter. Current trends in deep- space missions are to rely increasingly on autonomous tracking in order to reduce operations costs. Accord- ingly, a system that can minimize or eliminate the need for Earth updates is highly desirable. The sensor selec- tion likely to depend on the altitudes to be flown and the characteristics of the surface and atmosphere to be observed.
For Venus or Titan missions, a cloud-shrouded at- mosphere makes celestial references impractical, but surface referencing is attractive. For Venus, imaging in the infrared region of the spectrum yields maps of sur- face temperature that correlate strongly with altitude. It has been demonstrated that it is possible to match these data with a low- resolution map of Venus topogra- phy obtained by the Magellan mission (Gaskill, 1994). Monitoring the radio metric parameters of a known radio beacon of an orbiter provides further informa- tion that can be used in the velocity estimator process. Velocity measurements can be acquired from frame- to-frame correlation of repeated images as the aerobot drifts over the surface. For both measurements, gyro and accelerometer measurements of the rotational state of the aerobot can be used to establish the orientation and pointing of the sensor.
For a Mars mission (MAX, MFR), the absense of clouds in the atmosphere makes celestial references (Sun, stars) practical although the occurrence of dust storms may degrade stellar visibility during both day and night. The use of a simple Sun sensor such as the "computer vision sextant" proposed by Cozman & Krotkov (1995) for rover missions has merit. This sen- sor idea is attractive because the location of the aerobot can be constrained by measuring the elevation of the Sun at a known time. Sun sensors weighing as little as 120 grams have been designed by industry as part of a JPL program on microrovers. A definitive posi- tional measurement is feasible if the solar azimuth can be measured with a stable azimuthal reference. Simul- taneous measurements with an imaging sensor during the day could be used to determine velocities with a similar approach to that discussed for Venus.
For Titan, the Cassini mission is expected to provide radar maps similar to those already obtained for Venus by the Magellan mission. However, thermal imaging of altitude at Titan by the aerobot, is impractical due to the extremely cold conditions. Radar altimetry or visible imaging may permit positional referencing to the radar map, but they have not been studied. Velocity measurements could be acquired with Doppler radar or with repeated correlated visual imaging.
For outer planet missions, with no surface references and no access to celestial references, other sensing ap- proaches are needed. Fortunately, all these planets have strong magnetic fields offset from their rotation axes.
FROM ON-BOARD SENSORS • Aerobot System Temperatures/Pressures • ~nblent Atmosphere Temperatures/Pressures
Va/~ Ac~t/on
- - P r e d ~ d Fi~M park Updates
[ Val~ Aetaatiom
L~t $or Actual FllgM P~A Updates
I FLIGHT PATH
-[ RECOVERY SYSTEM
ALTITUDE ] CONTROLLER
Altitude control and navigation.
This should enable measurements of both latitude and longitude to be made with useful precision as well as providing the azimuthal orientation of the aerobot sen- sor platform.
The preceding discussion of sensors was primarily in the context of global positioning and navigation. When landing aerobots at designated landing sites, reference to detailed surface topography or imaging of the target area and its surrounding is needed. Accordingly, a ref- erences database would be loaded in the aerobot prior to the terminal descent phase.
5.2 Altitude Control and Navigation
The block diagram in Fig. 8 illustrates the major fea- tures of an on-board Altitude Control and Navigation (ACN) system that guides the aerobot to a desired loca- tion above or at the planetary surface. This destination is referred to in the diagram as the target. The system receives the location of the target expressed in latitude, longitude, and altitude coordinates from the ground station on Earth.
This target-site-location information is processed by a GTG, which predicts a flight path that the aerobot must achieve in order to get to the desired terminal de- scent entry corridor and generates a command list for the altitude controller that is designed to realize this flight path. The GTG uses a simplified on-board atmo- spheric model based on the best available information (global circulation model) about prevailing wind condi- tions at various altitudes and a model of the reversible- fluid altitude-control system. Other inputs to the GTG
are thermodynamic parameters of the ambient atmo- sphere and the altitude control system from on-board sensors and the initial location and velocity of the aer- obot from the ASE.
Inevitable uncertainties in the wind model mean that the predicted flight path will have significant uncertain- ties. The Flight Path Recovery System compares the predicted flight path with the actual flight path from the Autonomous Position and Velocity Estimator and, when deviations exceed a prescribed threshold, issues commands to the flight controller to update the flight path so that the aerobot achieves its targeted destina- tion. Changes to the profile will not be made continu- ally but only at a small number of discrete points in the trajectory.
The Altitude Controller is at the core of the over- all altitude control and navigation systems. Control in altitude is essential to achieve controlled trajecto- ries over the planetary surface. Semicontrolled move- ment lateral to the uncontrolled flight path is achieved by combining altitude control with knowledge of the wind direction as a function of altitude. Results dis- cussed earlier suggest that on Venus, for example, the circulation is inherently more predictable than that of the Earth. However, the wind models will inevitably have significant errors, and the accuracy of the result- ing aerobot trajectory will degrade as it is projected further into the future.
Aerobot navigation is a very different class of guidance-and-control problem than is faced by the navigator of a spacecraft who deals with highly deter- ministic gravitational effects and well defined control
276 Cutts et al.
impulses. Fortunately, unlike the spacecraft naviga- tor, the aerobot will have more than one opportunity to perform the maneuver needed to view targets or to reach a terminal descent region. Several circum- navigation trajectories may be necessary to gradually reduce the trajectory errors and reach the designated destination.
While the goal of controlling the trajectory and land- ing location of a lighter-than-air vehicle is challenging, a promising precedent exists in terrestrial teleoperated balloon experiments which have used helium venting and ballast dropping for altitude control. These experi- ments were motivated by atmospheric scientists inter- ested in examining more than one vertical slice through the atmosphere and by engineers interested in using balloons as missile targets. Statistics from missile tar- get flights indicate that, with a launch site 80 km away, 60% of the flights came within 10 km of the desired tar- get center and 70% came within 15 rain. of the desired overflight time (Gildenberg, 1970).
5.3 Aerobot Operations
In this section, aerobot capabilities are examined that can enhance the exploration potential and the scientific return from a mission.
5.3.1 Observing Targets of Opportunity. A conven- tional balloon mission, such as the Martian Aerostat and the proposed Mars Airborne Platform, makes ob- servations that are essentially random along a poorly known flight path. An aerobot, equipped with a basic capability for autonomous position determination, can provide a major gain in balloon capability by being able to observe targets of opportunity.
As previously discussed, ASE output can be made available to the controller of the remote-sensing pay- load, which also has access to the locations of a priori- tized set of targets of opportunity stored in the aerobot main memory (Fig. 7). In this way, scarce data-return resources (storage and telecommunications) are only allocated to the highest priority targets traversed by the aerobot.
Even without an ability to change flight path, this method represents a substantial increase in capability over a conventional balloon. In the case of BEV, which executes cyclic altitude variation, an additional condi- tion for data collection would be the altitude at which data are acquired. Images from the lowest points in the aerobot trajectory would be expected to be of much higher quality than data from higher altitude.
This level of vehicle control can also be used to deploy small instrument packages to the surface of planets provided the change in buoyancy can be accom- modated or even used in lieu of a reversible altitude- control system.
5.3.2 Control of Flight Path. The goal offlightpath control is to maneuver the aerobot from its initial 3-D atmospheric location to an approach position where low-altitude observations of a designated target can be conducted or landing operations can begin. Periodic command and navigation updates will be required. For planetary aerobots, communications are possible only when the aerobot is on the nearside of the planet unless an orbital relay satellite is used. Even then, communi- cations may be limited. For an aerobot in the upper at- mosphere winds of Venus 50-60 km altitude, the Earth is out of view for 3 to 4 days during every 6 to 8-days "orbit" around the planet.
Flight-path control of an aerobot is analogous to a microrover navigating inaccurately modeled static ter- rain, but it is even more challenging, because the wind environment is dynamic and 3-dimensional. The tradi- tional robotics problem of going from Point A to Point B is very different in atmospheric robot navigation. Moving from one location to another involves changing altitude to exploit one or more wind-vector directions.
The aerobot uses a simple on-board wind pattern model for trajectory generation. This wind model is analogous to the world model embedded in a conven- tional robot to map obstacles. An aerobot's wind model captures planetary wind behavior in terms of prevailing wind directions, wind velocities vector directions as a function of altitude, vertical down/up drafts caused by winds passing over the surface caused by topography, cloud-cover, and day/night insolation models. These comprise relatively simple computer models suitable for on-board use and are not complex meteorological models.
Data from prior planetary missions, e.g., Venus (Crisp et al., 1990), provide a start for such models. Challenging maneuvers include long longitudinal tra- verse, equator crossing, and latitude change. Initially, there will be large uncertainties in global navigation. Early efforts will therefore focus on reaching regions of fairly large size, or of reaching targets of opportu- nity, instead of trying to achieve pin-point landing at specific sites.
5.3.3 Terminal Descent Strategy. For a Venus aer- obot, terminal descent starts at a corridor upwind of the
Planetary Exploration by Robotic Aerovehicles 277
LANDING PREVAILING . , SiTE WIND DIRECTIOn/ AT LANDING SI ~ m
/ / I t l ,~ DESCENT / * / TRAJECTORY / /e STAR'I IN TERMINAL . ~, CORRIDOR DESCENT PHASE
TRAJECTORY IN GLOBAL NAVIGATION PHASE
A. TRAJECTORY PHASES
WIND VECTOR FIELD "TACKING" MOTION AT ALTITUDE 1 OF ROBOT BALLOON
/
WIND t~ VECTOR FIELD AT ALTITUDE 2
B. s'n~ERING CONCEPT
Fig. 9. Venus robot trajectory and steering concepts.
desired landing site, as shown in Fig. 9. This phase, in which the robot uses sensor data from the environ- ment to guide itself to the target, is analogous to a conventional robot performing a task in contact with its environment. However, the descent task is more difficult because of the inherent uncertainty of the ac- tual wind profiles. For Venus, previous Venera and Pioneer Venus probes have determined a vertical pro- file in wind speed ranging from about 100 rrds at about 65 km to near zero at the surface (Crisp et al., 1990). The current uncertainty in wind velocity profiles is about 20%.
Uncontrolled descent only lands the vehicle some- where along its wind-driven path. Controlling the de- scent rate allows the target to be achieved, provided wind and thermodynamic profiles are nominal and the descent start corridor is chosen correctly. To land at the required target, the nominal descent rate is matched with the expected horizontal wind velocity. A more advanced approach uses on-board sensors to iteratively estimate the likeliest landing point, and continually ad- just the descent rate to guide the vehicle to the pre- scribed site. This increases the reliability of achieving the target along the wind-driven path, but also increases system complexity. Controlled descent is particularly challenging with reversible-fluid altitude control since the mobility mechanism available to the robot as it goes down is one-sided. The descent rate can only be made slower, not faster. This is a classic unidirec- tional control problem where one targets to undershoot and adjusts to hit the target. An analogy is the Viking lander descent profile. During descent the Viking lan- der could not fall any faster than free-fall at Mars. The propulsion system thrusters were used to slow the ve- hicle down for landing. Augmentation of reversible fluid altitude control with variable geometry surfaces to increase descent rate has been considered and was discussed earlier.
6 Planetary Aerobot Testbed Vehicle
The Planetary Aerobot Testbed (PAT) was conceived to carry out proof-of-concept tests for the ASE and the ACN subsystems described in the last section. It will build on the accomplishments a series of Altitude Control Experiments (ALICE) conducted last year that proved the concept of using reversible fluids to induce cyclical altitude variations about a stabilizing altitude. The testbed includes two principal subsystems: the aerobot vehicle itself and a workstation used for control and display.
PAT will primarily address the challenges of aerobot missions to planets and satellites with solid surfaces-- Venus, Mars, Titan. In demonstrating the ASE func- tions, it will fly some sensors and emulate others. It will use the same reversible fluids proven in the ALICE pro- gram, whose thermophysical properties are very simi- lar to fluids planned for a Venus aerobot.
6.1 Planetary Aerobot Testbed Vehicle
The PAT vehicle (Fig. 10) uses two attached balloons: helium in one provides most of the buoyancy, while a second, smaller balloon, provides altitude control by using a reversible fluid selected as described earlier for ALICE. Several reversible fluids are possible for use on Earth, each with a different condensation equilib- rium altitude. Freon R114, with a condensation alti- tude of about 7 kin, depending on season, is a suitable fluid. With an appropriate amount of reversible fluid, the vehicle oscillates about the equilibrium altitude of the fluid. These oscillations are "forced" by the evap- oration of the fluid at low altitude, which increases buoyancy and causes the balloon system to rise and by the condensation of the fluid at high altitude, which reduces buoyancy, allowing gravity to pull the system down. For the helium/R114 system, the amplitude of this oscillation is expected to be about 4 or 5 km.
To descend, the liquid condensing in the cold upper half of the altitude cycle is trapped i~'side a small pres- sure vessel thus creating negative lift. The system then descends into warmer lower altitudes, and eventually settles on the surface. A landing "snake" keeps the gon- dola hovering offthe surface. At any point in a descent, valves can be opened to allow the now super-heated liq- uid to boil and reinflate the small, buoyancy-controlled balloon. With a net positive lift, the system goes up to the cooler upper altitudes and resumes oscillation about the equilibrium altitude. It is this vertical con- trol, combined with a varied and rich wind structure,
that can enable long term, cyclic operations to and from landing sites on Earth.
The entire PAT vehicle mass is expected to be 15 to 25 kg depending upon the degree of autonomy incor- porated and the efficiency of the heat exchanger sys- tem. The gondola system includes a remotely operated or autonomous controller, flight telemetry subsystem, Global Positioning System (GPS) receiver, geosyn- chronous satellite-communications subsystem, Federal Aviation Administration (FAA) transponder, redun- dant balloon cut-down subsystem controller, struc- ture/insulation and batteries. The reversible-fluid heat
HI °~ t WIND DIRECTION ~ / T ..ys..,a.-i,.,.,-,m~ .+ • ~ ' J ~ L ~ m ~ . . , ~ I C #1
f ,.,,~/¢" LANDING ~ PREDICTS
i ' ~ , GPS DATA i
LONGITUDE i tm
f + T t f t T _ _ - . ~ - - . ~ - - . - ~ - -
NEUTRAL liUOYANCY. LINE ~ I
~t / +,~ it.)+ "%...++<.t c,~.~>j "<>
TIME
Predictive display simulation.
exchanger system includes valves, actuators, heat ex- changer fins, a reservoir, and plumbing.
The PAT vehicle also has a sensor complement and an on-board controller/computer. It will operate in two modes: telerobotic control in which commands from the workstation are used to operate the altitude control system and autonomous control in which these com- mands are issued by the vehicle computer based upon real-time sensory perception of the balloon thermo- dynamic state, ambient atmospheric conditions, and the desired actions, i.e., landing at a desired site.
Initially, many of the ASE and ACN functions will actually be implemented in the PAT workstation. These functions will be migrated to the PAT vehicle as its computational capabilities are upgraded.
6.2 PAT Workstation
The PAT workstation will be able to display com- mands and vehicle status and will support the ASE and ACN functions at high rates. Complementary
Planetary Exploration by Robotic Aerovehicles 279
functions performed on the aerobot vehicle will incor- porate faster-than-real-time flight dynamics models of aerobot flight paths. The workstation will be used in the initial teleoperation phase to display the actual bal- loon trajectory and predicted trajectories based upon various control scenarios, Vertical profiles and views of planned balloon ground tracks will be generated. Figure 11 illustrates various outputs of this display system. Figure 1 la shows the actual balloon vertical profile and tile predicted profile assuming no control ac- tion is initiated ("no action predict"). Alternate vertical profiles can be shown which describe balloon behavior if various control strategies are employed, e.g., clos- ing the reversible-fluid container valve before or after reaching peak altitude. Figure 1 lb illustrates this same actual and predicted behavior in plain view. This view shows the effect of variable wind speed and wind direc- tion with altitude. Two different simulated predicted landing sites are shown depending on control strategy.
7 Conclusions
The development of small planetary aerobots is opening up a new phase in planetary exploration-- the long-duration, detailed study of planetary envi- ronments using 3-dimensional mobility. Compact, yet powerful exploration systems will be able to carry out comprehensive imaging exploration from low altitudes of Mars, Venus, and Titan and make excursions to the surface to sample surface materials. The deep atmo- spheres of the four outer planets are also accessible to exploration with this approach. The development of new approaches to altitude control in planetary at- mospheres creates substantial challenges in navigation and autonomous control. These problems are being ad- dressed at JPL but can benefit from research throughout the telerobotics community.
Appendix A--Comparison of Planetary Aerobots and Rovers
One of our goals in the aerobot program is to draw upon the systems and technology heritage obtained in almost two decades of rover development at NASA cul- minating in the Microrover Flight Experiment (MFEX) on Mars Pathfinder in 1996. Thus, it is important to understand the relationship between the mobility, con- trol, and navigation functions that are employed in the aerobot with those for planetary rover.
Mobility. Fundamental differences in the mobility of planetary rovers and aerobots determines give rise to differences in their control, navigation, and path- planning needs. Planetary rovers are restricted to motion on a 2 dimensional surface. Except on steep downward slopes, power is needed to move over that surface to overcome gravitational forces and rolling resistance. When power is removed, the vehicles typi- cally come to an abrupt halt. Because of rough terrain and limited power, vehicle speeds are very low. Early planetary rover missions will be limited to distances of a few tens of meters; traveling tens or hundreds of kilometers will require a mission length measured in months or years.
Planetary aerobots, on the other hand, can move in all three dimensions above the surface. With a reversible- fluid altitude-control system, power is needed only to switch the buoyancy state. As a result, the aerobot can move horizontally relative to the surface with negli- gible energy when carried by horizontal wind forces. However, there is no way to stop the vehicle except by descending to the surface. The aerobots envisaged here will move horizontally with the atmospheric flow, but will have small up or downward velocities relative to the atmosphere, depending on the buoyancy state. Although the use of power or airfoils for horizontal motion relative to the atmosphere can be envisaged, these will require substantial energy input and are not required for any of the aerobot mission concepts dis- cussed here. Indeed, the atmospheric motions alone allow global coverage--in two days, the 1985 Vega balloons traversed 11,000 km relative to the Venusian surface.
Control: Control of planetary rovers is achieved by motion and turning of the wheels aided primarily by visible and other information on local terrain charac- teristics. Various sophisticated mobility systems have been devised to surmount obstacles, deal with soft ter- rain, and accommodate steep slopes.
For control of the aerobots envisaged here, the prin- cipal mover is buoyancy change and knowledge of the atmospheric environment which allows rates of as- cent or descent to be predicted. Coupling this with information on the windfields in the regions traveled leads to a 3 dimensional path through the atmosphere.
Navigation: The differences in mobility and control between planetary rovers and planetary aerobots dictate major differences in navigation approach, although some of the sensors used in rover missions may be
280 Cutts et al.
appropriate to aerobots. For planetary rover missions,
the essential reference frame is provided by images ac-
quired by the lander from the rover. Depending on the
range of the mission, it may not be necessary or possi- ble to reference the vehicle to a regional or global map.
Steroimaging using laser-ranging sensors is important
for navigating through a field of obstacles. For planetary aerobot mission, a global perspective
is required. Radiogacking from either an orbital space-
craft or from a ground-based VLBI array, as was used
on the Vega balloon mission, is a proven approach that
provides accurate positional and velocity data. How-
ever, these data will only be available when the aerobot
is in view; sometimes the aerobot may be inaccessible
because of its depth within the atmosphere. More-
over, the information must be communicated from the
ground station or the orbiter to the robot and may not be available in a timely fashion. Accordingly, other ap-
proaches that involve only sensing and computation on
the aerobot itself are needed. As discussed, these may include the use of passive visible and infrared imag-
ing, radar ranging, and Doppler, solar, and stellar map- pers and trackers, and inertial sensors. As we shall
see later, some recent work done on sensors for long-
range rovers (Cozman & Krotkov, 1995) is applicable
to aerobot navigation needs.
Acknowledgments
Many people have contributed to the ideas developed
cribed in this paper was carried out by the Jet Propul-
sion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space
Administration.
References
Akiba, R., Hinada, M., and Nakajima, T. 1992. Simulation study of venus balloon system. Paper IAF-92-0559 presented at the 43rd Congress of the International Astronautical Federation.
Anderson, C.M. 1991. Wind, Sand and Mars: The 1990 Tests of the Mars balloon and snake. The Planetary Report, 11(1):12-15.
Biregovoy, S.N. 1993. The hopping motion balloon station in the lower layers of the venusian atmosphere. Informal Technical Brief, Babakin Center, Moscow.
Blamont, J. et al. 1993. Balloons on planet Venus: Final results. Proceedings on Scientific Ballooning. Adv. Space Res., Pergamon Press, 13(2):145-152.
Blamont, J. 1981. Balloons on other planets. Adv. Space Res., 1:63-69.
Blamont, J. 1993. Development of Mars network by balloons. Intl. Workshop for Coordination of Mars Exploration, Germany, CNES/DG/CN No. 1631, Paris.
Blamont, J. 1993. Balloons for the exploration of Mars. Adv. Space Res., 13(2):137-144.
Blamont, J., Heinsheimer, T., and Pommerean, J.P. 1974. Nouvelle methode d'etude de la dynamique de la stratosphere; principe et premiers resultats. Academie des Sciences (Paris), Comptes Rendus, Seris B Sciences Physiques, 278(7):249-254.
Blamont, J.T. 1985. The exploration of the atmosphere of venus by balloons. Space Research. 5(9):99-106.
Carlson, L.A. and Horn. W.J. 1983. New thermal and trajec- tory model for high-altitude balloons. AIAA Journal of Aircraft, 20:500-507.
Crisp, D., Cozman, and Krotkov, E. 1990. Vega balloon meteorolog- ical measurements. Adv. Space Res., 10(5):109-124.
CuRs, J.A., Nock, K.T., Jones, J.A., Rodriguez, G., Balaram, J., Powell, G.E., and Synnott, S.P. 1995. New means for mobility. 1EEE Conference on Robotics & Automation, Nagoya, Japan.
Desjean, M.C. 1981. A balloon and its basket in the Venus atmo- sphere. In Proceedings of the 2nd Int'l Conf. Numerical Methods in Thermal Problems, 2:1302-1313~Italy.
DiCicco, A., Nock, K.T., and Powell, G.E. 1995. Balloon experiment at Venus (BEV). In Proceedings 11th Lighter-Than-Air Technol- ogy Conference, AIAA 95-1623, Clearwater, FL.
Gildenberg, B.D. 1970. General philosophy and techniques of bal- loon control. 6th AFCRL Scientific Balloon Symposium.
Houghton, J.T. 1977. The Physics of Atmospheres, Cambridge Uni- versity Press, p. 203.
Hunter, D.M., Colin, L. Donahue, T., and Moroz, V. 1983. Venus, The University of Arizona Press.
Jones, Jack A. 1995. Reversible fluid balloon altitude control con- cepts. In Proceedings of the 11th Lighter-Than-Air Systems Tech- nology Conference, AIAA-95-1621, clearwater, FL.
Koshlyakov, V.N. 1984. The Magnus effect. DOP-UKR-A, 4:39-43. Krotkov, E., Hebert, M., and Simmons, R. 1996. Stereo percep-
tion and dead reckoning for a prototype lunar rover. Journal of Autonomous Robots, Special Issue on Autonomous Vehicles for Planetary Exploration.
Malaterre, P. 1993. Long duration balloon flights in the middle strato- sphere. Adv. Space Res., 13(2):107-114.
Matijevic, J. and Shirley, D. 1996. Mars pathfinder rover. Journal of Autonomous Robots, Special Issue on Autonomous Vehicles for Planetary Exploration.
Matthies, L., Gat, E., Harrison, R., Wilcox, B., Volpe, R., and Litwin, T. 1996. Mars microrover navigation: Performance evaluation and enhancement. Autonomous Robots, Special Issue on Autonomous Vehicles for Planetary Exploration.
Moroz, V.I., 1994. VA-94 Venus atmosphere model for discovery venera project. Space Research Institute of Russian Academy of Science (IKI), Profsojuznaja, 84/32, Moscow.
Moskalenko, G.M. and Troshin, V.S. 1980. Izbytochnoye devleniye v aerostaticheskoy obolochke i vozmozhnost' ego regulirovaniya pri plavanii v atmosfere Venery. Excess Pressure in an Aerostatic
P l a n e t a r y E x p l o r a t i o n b y R o b o t i c A e r o v e h i c l e s 281
Shell and the Possibility of its Regulation During Floating in the Atmosphere of Venus, Kosmicheskiye Issledovaniya, 18:474-476.
Moskalenko, G.M. 1981. Dirizhabl' dlya Venery (Dirigibles for Venus), Nauka i Zhizn, 9:85-87.
Needleman, H.C., Nock, R.S., and Bawcom, D.W. 1993. Status of the NASA balloon program. Adv. Space Res., 13(2):69-76.
Nishimura, J., Yajima, N., Fujii, M., and Yokota, R. 1990. A model experiment of the Venus balloon. In Proceedings of the Interna- tional Symposium on Space Technology and Science, Tokyo,pp. 1589-1592.
Robertson, D.E 1994. To boldly go: A new generation of rovers are poised to explore surfaces on the Moon, Venus, Mars and beyond. Astronomy, 22(12):.
Romero, M. 1980. Balloon sur Venus--Gonflage sous Parachute et Oscillations de Balloon Ludion (Balloons over Venus-- Inflation under Parachute and Ludion Balloons Oscillations). Centre d'Etude et due Recherches de Toulouse, ONERA, CR/BAL/THER/49.
Romero, M. 1981. Balloon sur Venus--Simulations de Divers Modes de Stabilization (Balloons over Venus)--Simulations of Various Stabilization Methods). Convention 81/CNES/0728, Etude Cert 412, CR/BAL/THER 151.
Rougeron, M. 1969. CNES Article 329/DET/TO/BA/EE. Schubert, G., 1983. General circulation and the dynamical state of
the Venus atmosphere. Venus, The University of Arizona Press, Tucson, pp. 681-765
Science, 1986. Special issue on venus ballooning. Sirmain, C., Evard, J., and Vega, J. 1995. Martian aerostat deploy-
ment: Analysis and test paper. Presented at AIAA 1 lth Lighter- Than-Air Conference, Clearwater, FL.
Vorachek, J.J. 1970. A comparison of several very high altitude sta- tion keeping balloon concepts. 6th AFCRL Balloon Symposium, pp. 355-381.
Wu, J.J. and Jones, J.A. 1995. Performance models for reversible fluid balloons. In Proceedings 11th Lighter-Than-Air Technology Conference, AIAA- 95-1623, Clearwater, FL.
Yavrouian, A., Plett, G., and Yen, S.S. 1995. High temperature bal- loon materials for Venus balloon envelopes. In Proceedings of the 11th Lighter-Than- Air Systems Technology Conference, AIAA- 95-1617, Clearwater, FL.
Zubrin, R., Price, S. Clark, B. Cantrell, J., and Bourke, R. 1993. A new Mars aerial platform. Aerospace America, pp. 20-24.
scientific investigator on the Mariner 6, 7 and 9 missions to Mars as well as on the Viking Mars program and lead the planetary sci- ence department at Science Applications International Corporation during much of this period. Since returning to JPL in 1982, he has served as the Chief Technologist of the Observational Systems di- vision, Deputy Director of the Center for Space Microelectronics Technology and the Manager of the Technology for Advanced Opti- cal Systems program. Prior to his current position he was Program Manager for Advanced Science Instruments and Space Technology. He holds an MA degree in Natural Sciences (Physics) at St. Johns College, Cambridge University, England, and both as MS degree in Geophysics and a Ph.D. degree in Planetary Science from the California Institute of Technology. He also received the Executive Management Program Certificate from the University of California at Los Angles.
Kerry T. Nock graduated with a MS in Space Science and Engineer- ing from UCLA. He has worked at JPL since 1969 on the Mission Design of various flight projects including the Mariner 9 mission to Mars, Mariner 10 mission to Venus and Mercury, and the Voyager and Galileo missions. Mr. Nock has managed several advanced planetary mission studies at JPL including early studies which lead to the Magellan Venus mapping mission and the Joint NASA/ESA Cassini mission to Saturn and Titan. He has also studied many other advanced missions to the Moon, Venus, comets, asteroids and the outer planets. In addition, Mr. Nock has supervised JPL's Advanced Projects Group, supported the Paine Commission on Space and led a number of advanced space technology studies involving solar and nuclear electric propulsion and small spacecraft technologies. He is now the manager of the JPL Planetary Aerobots Program.
James A. Cutts has been Manager of the Advanced Concepts Pro- gram at the Jet Propulsion Laboratory since 1994. He served as a
Jack A. Jones MSME Rice University, NSF Fellow, 1973: BSME, Rutgers University, James Slade Scholar, 1970. Member of Technical Staff at JPL since 1979 performing thermal analysis, R&D, and developing engineering solutions for spacecraft cryogenic refrigeration systems, balloon systems, and ground-based refrigera- tion systems. Senior Mechanical Thermodynamic Engineer at Gar- rett AiResearch from 1975-79 performing thermal analyses and beat
2 8 2 Cutts et al.
transfer designs for ground-based and spacecraft systems. Over 70 professional publications, including over 30 NASA Tech Briefs, with 17 patents and 3 major NASA Invention Awards for fluid refrigera- tion systems. Numerous professional honors and organizations.
Guillermo Rodriguez has been at the Jet Propulsion Laboratory since graduating from UCLA in 1974 with a Ph.D. degree in control engineering. He has participated in the development of autonomous control systems for several planetary spacecraft. He also has been involved in research programs in space robotics and control. His main research activities are in autonomous control architectures, es- timation theory, and robot dynamics.
J. Balaram received the B. Tech degree in Mechanical Engineering from the Indian Institute of Technology in 1980, and the MS and Ph.D. degrees in Computer and Systems Engineering from Rensse- laer Polytechnic Institute in 1982, and 1985, respectively. He has been at JPL since 1985 and has worked in the area of task planning and machine vision for several JPL projects in telerobotics. He is cur- rently involved in the design and implementation of the perception and navigations systems for future Mars rovers. His interests in- clude machine vision, real-time architectures, and task-level robotic control.
