Advanced Automation for Space Missions/Chapter 4.4

4.4 SMF Growth and Evolution

Following its deployment, the starting kit begins to manufacture second-generation tools, as well as replacement parts for itself. These.tools can be used to produce additional types of equipment and early product lines. Eventually, space-compatible equivalents of all major terrestrial manufacturing processes and new systems evolved in space must be available to the evolving SMF.

Further growth and increased complexity are required if the SMF is to evolve from the starting kit into a sophisticated manufacturing center which depends less and less on Earth for raw materials resupply. One key growth area especially significant in view of the heavy requirements for computers and robotics in space is the automated fabrication of integrated circuitry and other electronics components. Certain unique characteristics of the space environment, combined with anticipated advances in laser-, electron-, and ion-beam technologies, may make possible automated machinery capable of manufacturing highly sophisticated integrated circuits as well as resistors, capacitors, printed circuit boards, wire, and transformers in space, using raw materials supplied entirely from the Moon, and ultimately a wide variety of additional complex products.

4.4.1 Starting Kits for SMF Growth

Having considered a range of possible starting kits, the Team next explored the possibility of an ever-widening collection of production machinery using kits described in section 4.3.3. This aspect of the analysis is crucial to growth and evolution, since the taxonomy of manufacturing processes is distinct from the list of functional components comprising the implements of manufacturing. Table 4.21 showed the major functional machine components which must be available in a growing SMF. Nonterrestrial, especially lunar, materials can be used in most cases. The most serious deficiencies are the lubricants and fluids needed for pressure transfer or solution-processing (electrolytes, wetting agents), though silanes may be serviceable in lunar applications. High-powered lasers are convenient for cutting and finishing in space. The Moon is somewhat deficient in the most common gases used in tunable power lasers, He, Ar, Xe, but fortunately each gas is readily recyclable.

Manufacturing components listed in table 4.21 were reviewed specifically for derivability from starting kits, with the assumption that appropriate processed materials would be supplied as feedstock to the SMF:


 * Structures - A wide variety may be produced directly from any starting kit as described in section 4.3.3. These range from very small solid pieces such as shafts or dies to much larger components including rigid members for heavy presses. Metals, ceramics, and ceramic/metal combinations can also be prepared.
 * Refractories and dies - can be manufactured using the powder metallurgical components of the starting kit. Laser trimming can be performed as required after solidification and inspection of the part. These components then become available for casting complex shapes and for extruding both long-dimension components and parts designed to sustain very high temperatures and pressures.
 * Heating - by direct solar energy may initially be accomplished using aluminum deposited on spherical surfaces. These surfaces may be shaped by rotation of unitary structures of appropriate radii of curvature extruded using the starting kit. Alternatively, metal vapor deposition on interior subsections of bubbles grown in zero-g may be used. The existence of solar-electric devices is assumed.
 * Insulation - for both thermal and electrical needs can be derived from fiberglass mattes produced by a spinning process involving the extrusion of molten glass through small orifices. Electrical insulation exhibiting mechanical softness or compliance is achieved by pressing fiber mattes into long thin ribbons and then wrapping these tightly around the wires, followed by partial sintering. Basalt fibers may be useful in this application (see section 4.2.2).
 * Magnetic materials - can be manufactured directly from the starting kits or by powder metallurgical technologies. Dies and heating equipment produced in earlier steps are probably required for maximum versatility.
 * Electrical conductors - particularly wires for motors, busbars and other purposes, may be extruded (original starting kit equipment) or fabricated using rollers and dies derived from structure and refractory manufacturing components produced earlier.
 * Grinders - are needed for precision finishing of surfaces. These tools should be producible by pressing and casting operations available with the starting kits. Grinders may be composed of spinel grains (a lunar-abundant grinding agent) embedded in glass fiber mattes perfused with calcium for mechanical softness and binding.
 * Glasses and fibers - can be manufactured by using casting, grinding, and die-extrusion operations. Grinding is required for optical-quality glass shapes. Electron-beam and laser techniques are useful for final finishing of optical surfaces.
 * Adhesives and coatings - of metals and ceramics can be applied by the starting kits or a specialized kit suited to the particular geometries of certain parts.
 * Lubricants and fluids - present special problems because of deficiencies in presently known lunar raw materials resources. It may be that self-lubricating powder metallurgy bearings containing brass and lead in very small quantities are feasible. Also, silicon-based compounds requiring a minimum of relatively rare lunar carbon and hydrogen should be extensively investigated.
 * Lasing media - It is also important to determine to what extent lasing media for high-power lasers can be derived primarily from lunar materials. Undoubtedly a considerable literature applicable to such devices already exists, but is classified for military reasons.

Control systems and electronics (see section 4.4.3) are also necessary, especially for automated manufacturing facilities in space.

Several technologies with limited terrestrial applications may prove extremely useful in space. One example is containerless production, in which objects are formed directly from melts. Overall shape is controlled by surface tension, external forces, and directed solar heating. Vapor deposition is another potentially favorable technique which should be given high research priority. Also, as the human presence in space expands, special production environments that allow the use of gases and liquids will become more commonplace. Thus chip-producing machinery, foaming and other processes requiring the recovery of production fluids may eventually become feasible in space.

It is easy to see how a starting kit might generate production equipment required for other space-compatible manufacturing techniques. (Shearing operations are assumed to be within the capabilities of starting kit laser beam units). For example, laser techniques for scribing reverse threads onto hardened steel rolling dies is a foreseeable technology (fig. 4.17). The availability of chromium on the Moon (0.6% by weight and higher in beneficiated iron grains) and lunar basalt for base plates makes thread rolling a valuable adjunct to the starting kit extrusion system.



A second example is magnetic-pulse-forming equipment. The two main components of the magnetic-pulse former are the forming coil and the capacitor. Robots with appropriate wrist actions should be capable of conventional winding operations to manufacture forming coils from extruded wire. The capacitor may consist of a basalt/aluminum or alumina/aluminum sandwich based on the standard formula C = kEA/d, where C is capacitance, k is the dielectric constant of basalt or alumina (4.5-8.4 at 10$6$ Hz), E is the permittivity of free space, A is capacitor plate area, and d is plate spacing.

A third example is electroforming technology. As discussed in section 4.3.1, the components of an electroforming unit are somewhat more complex than those of magnetic-pulse formers because of the need for an electrolytic plating solution. The tank containing the solution may be fabricated using the extruder, then welded together by a laser beam unit. The mandrel (fig. 4.13) may be formed of cast or sintered basalt over which aluminum is vapor-deposited. Iron or titanium anode plates are no problem for the starting kit extruder, and centrifugally spun basalt may be used in the electrolyte filter. Cast basalt pipes, an off-the-shelf terrestrial casting technology, provide necessary plumbing for the entire electroforming system.

4.4.2 Near-Term Manufacturing Demonstration: Shuttle Tank Utilization

The Space Shuttle external tank (Martin Marietta Corporation, 1974) carries liquid fuel for the Shuttle main engines and separates from the spacecraft just prior to orbital insertion at an altitude of about 128 km. The cylinder then follows a ballistic re-entry path, crashing into the ocean far from inhabited areas. The cylinder is not recovered or reused. But the tank, when dropped, has already achieved roughly 99.7% of orbital velocity. The added delta-V needed for tank orbital insertion is only 46 m/sec, about 10% of available Shuttle Orbiter thrust.

Alternatively, the tank could be orbited by burning the main engines for a slightly longer time, or with the aid of a jet-assisted takeoff (JATO) booster. The cylinder itself measures 8.4 m diam, 47 m long (a volume roughly equivalent to that of a 10-story condominium), and 33,503 kg in inert weight. Most of this mass is pure structural aluminum, though about 100 kg of outer skin insulation contains organic materials which could serve as the basis for early organic chemistry at the SMF (carbon, plastics, biological products, and so forth). A few tons of unused propellants (LOX and LH$2$) may also be present, and surplus materials from Shuttle operations (hydrazine, helium, food, etc.) could be stored in orbit for later use.

Any Shuttle flight carrying a volume-limited cargo can bring the external tank to orbit with near-zero propulsion costs. Valued as payload at about $1000/kg, an empty tank is worth about $33.5 million, less additional propulsion costs but plus added value derived from conversion of tank mass to useful products by the SMF. If Shuttle flies every 2 weeks, the payload value of the tank masses inserted into orbit would be the equivalent of roughly $1 billion per year. To an orbital space manufacturing economy this represents new additional income, in this case the equivalent of about 20% of the current annual NASA budget.

For such a cost-effective program to be implemented, the means for orbital insertion of the tank must first be perfected: Next, a system (teleoperated or robotic) should be designed which is capable of scraping off valuable external insulation. Cutoff valves must be added to prevent excess propellant from venting (permitting it to be stored in orbit rather than lost to space).

The starting kit provides a means of reducing the tank to powder or liquid form. The kits described earlier can accomplish this directly without the necessity of manufacturing additional process equipment. Another possibility is a solar-powered milling device (with portable atmosphere) which clamps onto the external tank and carves it into small pieces, most likely under teleoperator control. Tank fragments are then melted by a solar furnace consisting of a spherical mirror constructed by aluminizing a thermoplastic bubble hemisphere (Moore, 1980). The plastic allows sunlight to enter but retains infrared radiation by internal reflection, keeping the work materials hot. A hatch is cut in the mirror to permit insertion of metal shards, which join the growing droplet of molten aluminum at the focus. The melt volume of an entire tank would be about 12 m$3$, easily maneuverable through a small opening if processing proceeds in a dozen or so smaller batches.

Once tank material is molten a variety of manufacturing options become available. Ingots or simple bulk castings could be prepared as feedstock for other SMF processing operations. Liquid or vapor metal streams could be directed into molds or sprayed onto lighter structures for stiffening. For instance, thin thermoplastic bubbles may be aluminized to make pressure vessels, mirrors, or heavy solar sails; then plastic is stripped off and recycled. A more elegant method is to blow uniform metal bubbles directly, an ideal zero-g application. Aluminum is a good thermal conductor and reflector, and hence radiates heat slowly while retaining an even temperature distribution. Small tin bubbles have recently been blown experimentally in drop towers (Wang and Kendall, 1980), but far more research remains to be done.

Quite large volumes can be enclosed by structures manufactured using metal derived from a single Shuttle external tank. Aluminum pressure vessels 50 mils thick can retain one-third normal Earth atmosphere (O'Neill, 1977). Average tank thickness is about 250 mils, so a pressure vessel of roughly 13,000 m$3$ can be made from just one tank. This is more than fifty times the volume of the Space Shuttle cargo bay (240 m$3$).

4.4.3 Middle-Term SMF Expansion: Manufacture of Electronics Components

The present study urges a dramatic increase in the utilization of computerization and automation in nearly every conceivable future NASA mission. It is likely that a nonterrestrial source of computers and robots eventually will prove both useful and cost-effective in space. The team analyzed currently available and anticipated electronics components manufacturing technologies to determine which will satisfy two major criteria: (1) compatibility with a low- or zero-g factory environment, and (2) possibility of deriving required consumables from lunar resources.

Key components in computer systems include integrated circuits (ICs), capacitors, resistors, printed circuit (PC) boards, and wire. Fabrication capability in these five critical areas will permit most other necessary components to be produced as well. For instance, an IC fabrication facility could manufacture at least some varieties of transistors, diodes (rectifiers, small-signal, and zener), varactors, thyristors, silicon-controlled rectifiers (SCRs), and others. It would, however, have difficulty producing light-emitting diodes (LEDs) due to the scarcity of gallium and arsenic on the Moon. Thus, the intent of the following analysis is to present feasibility arguments concerning how lunar materials near-closure might generally be achieved. Substitution and comprehensive manufacture of electronics components are beyond the scope of the present study. Even with this limited review, it is encouraging to note the number of instances in which space equals or is superior to terrestrial factory environments for the manufacture of electronic components.

Integrated circuits. Conventional wafer fabrication techniques (Oldham, 1977) are, for the most part, not feasible in a lunar-supplied SMF. On the other hand, the vacuum of space greatly enhances the applicability of several techniques which are at or beyond the current terrestrial state-of-the-art.

Silicon (chemical refining required) is plentiful on the lunar surface, about 20% by weight (Phinney et at, 1977). While it is not clear precisely how lunar silicon will be transformed into boules of the pure element, it is reasonable to assume that this can be accomplished. Hard vacuum should facilitate the processes of crystal-pulling and zone-refining purification of elemental silicon (Grossman, 1976). Conventional zone refining requires induction heating (Grossman, 1976; Manasse, 1977), a space-compatible technique.

High-speed ICs using silicon-on-sapphire (SOS) technology are currently being fabricated by Hewlett-Packard (Pighini, personal communication, 1980) and others for custom applications. Should it appear desirable to produce such high-speed devices in the SMF, it is worth noting that spinel is plentiful on the Moon. Spinels are closely related to sapphire and actually provide a better crystallographic match to silicon, leading to higher mobility and less aluminum autodoping than in conventional SOS processing (Glaser and Subak-Sharpe, 1977). (The only major problem with spinel is the difficulty of finding high-quality crystals of correct composition.) Epitaxial growth of silicon on spinel substrates may be accomplished by the pyrolysis of silane (Glaser and Subak-Sharpe, 1977) according to: Hydrogen is in short supply on the Moon, roughly 0.01% by weight (Phinney et al., 1977), but fortunately only small amounts of it are required in this procedure. Silane is also an intermediate product in the chemical refining scenario described by Waldron et al. (1979).

Conventional photolithography and diffusion techniques are not feasible for space electronics fabrication. Many of the required chemical elements are present in lunar soil only at the ppm or ppb level. Photoresists consist largely of hydrocarbons, substances whose atoms are rare and which deteriorate rapidly in the space environment. The best alternatives may be laser, electron beam, and ion beam technologies. It is anticipated that these methods could lead to greater reliability on an increasingly miniaturized scale, particularly under the high-quality vacuum conditions characteristic of space (Carter and Grant, 1976).

Ion implantation already has begun to supplant diffusion techniques in the practices of many semiconductor firms. This technology allows greater control over quantities of impurities introduced, depths and widths of doped volumes, concentration gradients, etc. Of particular interest for a future wafer fabrication plant in space is the potential for computer-controlled, maskless, multilayer implantation of multiple device types with submicron geometries (Namba, 1975; Wilson and Brewer, 1973). While further research and development must be conducted to translate this tremendous potential into practical reality, other features of ion implantation make it a highly desirable interim choice. Masking may be accomplished by aluminum or other metals, passivation layers, resists, etc. Doping also is possible using passivation layers, an approach which could lead to reduced leakage and better yields (Wilson and Brewer, 1973).

One drawback to ion implantation is crystal lattice damage. A recently developed technique permits extremely localized annealing by laser beam (Tebo, 1979). This process, unlike its thermal annealing predecessor, completely restores damaged crystalline structures through epitaxial regrowth. The net result is a lower resistivity material more suitable for semiconductor use, with fewer defects and higher yields. If this laser technique can be computer controlled like the multilayer ion process described earlier, automated production of three-dimensional integrated circuitry in space is entirely conceivable.

Pre-3D wafer technologies adaptable to more conventional production sequences also are available. Chemical and plasma etching processes require chemicals (e.g., HF, H$4$SO$2$, CF$2$-O$4$) which cannot conveniently be supplied in sizable quantities from lunar soil. A feasible substitute may be ion beam etching. While the closely related process of sputter-etching requires high-pressure argon gas, ion-beam etching at the rate of 10-300 nm/min can be achieved in a 10$4$ torr argon atmosphere (Glaser and Subak-Sharpe, 1977). Titanium oxide is a suitable etch mask for this process. Argon and titanium are available from lunar sources (1 ppm and 1-5%, respectively) in the necessary quantities.

One chemical vapor deposition technique is perfectly space-compatible, An electron beam easily evaporates materials such as aluminum in vacuo, so metal masking and metallization pose no unusual problems. Oxidation of silicon for masking or passivation purposes probably is most easily achieved thermally using anhydrous oxygen gas, rather than chemical vapor deposition methods which require hydrogen compounds. An alternative oxidation process might involve the use of a laser to create extremely localized heating (Tebo, 1979). Aluminum and oxygen are plentiful in lunar soil (5-14% and 40-45% by weight, respectively).

One final critical issue is cleanliness. Particulates should pose fewer problems in space than on Earth because of the absence of atmosphere for convective transfer. An aperture in the fabrication facility enclosure opposite the SMF velocity vector, suitably baffled, should provide a clean vacuum source. Some versions of such orbital devices are called molecular shields, and can provide less than 10$2$ torr environments at LEO. Internally, moving parts and outgassing are probable sources of particulates which must be minimized (Naumann, personal communication, 1980). Condensibles may prove a bigger cleanliness problem than particles. Techniques for coping with them include avoiding line-of-sight exposure to sources, use of materials with high vapor pressures, and installation of cold traps.

Capacitors. Basic elements of discrete fixed capacitors include metal plates or foil, dielectric material, and wire leads. The plates or foil and leads can be contrived from readily available aluminum. Alumina, silica, and a variety of glass and ceramic materials provide suitable dielectrics. All of these substances are readily available from lunar sources.

Two capacitor fabrication techniques - thin- and thick-film - are compatible with silicon integrated circuit technology, though discrete capacitors generally are preferred over thick-film versions (Glaser and Subak-Sharpe, 1977). Thin-film capacitors usually are made with tantalum (Ankrum, 1971; Grossman, 1976; Khambata, 1963). However, thin-film capacitors with higher working voltages but lower capacitance and slightly poorer temperature stability can be constructed of alternating aluminum and alumina (or silica) layers over silicon dioxide and the silicon substrate (Ankrum, 1971; Glaser and Subak-Sharpe, 1977; Khambata, 1963). Titanium dioxide is another possible dielectric - its dielectric constant is four times that of alumina (Glaser and Subak-Sharpe, 1977). Maximum capacitance values obtainable using thin-film technology are on the order of thousands of picofarads, and automated laser trimming can produce a high-accuracy (+/-0.05%) product (Grossman, 1976).

Resistors. Since carbon is a relatively scarce lunar resource, only wire-wound, metal or metal-oxide-film, and semiconductor resistors (Dummer, 1970; Claser and Subak-Sharpe, 1977) will be seriously considered for use in space applications.

Wire-wound devices are appropriate in applications requiring relatively high power dissipation, such as bleeder resistors in power supplies. Nichrome wire (80% nickel, 20% chromium) can probably be supplied in limited quantities from lunar materials (abundances 0.01-0.03% and 0.1-0.4%, respectively). Titanium, another possibility, is abundant on the Moon, and has a resistivity (42 M ohm-cm) which is approximately half that of nichrome.

However, most resistors used in computer circuitry need not dissipate much power. Thin-film and semiconductor devices appear most promising in this regard. Thin-film resistors are fabricated by evaporation or by sputtering 0.025-2.5 um of metal or metal alloy onto a substrate of alumina or silica (Glaser and Subak-Sharpe, 1977; Grossman, 1976; Khambata, 1963; Manasse, 1977). While some metallic materials commonly used in resistor manufacture are too rare in lunar soil for serious consideration (e.g., tantalum, nichrome, tin oxide, chromium), titanium offers a sheet resistance of 2 k-ohms/cm2 and a temperature coefficient of resistance (TCR) of -100 ppm/°C (Ankrum, 1971; Dummer, 1970; Grossman, 1976; Khambata, 1963). Thus, the electron-beam evaporation and laser-beam trimming technologies discussed above may be utilized to prepare fine-tolerance, thin-film titanium resistors (Glaser and Subak-Sharpe, 1977; Grossman, 1976; Khambata, 1963; Manasse, 1977). At present it is unknown how closely these technologies can approach contemporary terrestrial tolerance and manufacturing standards (better than +/-0.005%, TCR = 1 ppm/°C; Rothschild et al., 1980).

Semiconductor resistors can be made with a technology already discussed. Ion implantation of boron into silicon produces sheet resistances of up to 12 k-ohms/cm$-4$,suggesting that high discrete values are readily achievable. While less precise than their thin-film counterparts, ion-implanted semiconductor resistors have been shown to offer yields on the order of 90% after packaging (Wilson and Brewer, 1973).

Printed circuit boards. Printed circuit (PC) boards are made of phenolic resin reinforced with paper, or an epoxide resin reinforced with paper or fiberglass cloth, which is then clad with copper (Coombs, 1979; Scarlett, 1970). Unfortunately, resins deteriorate in space and are difficult to prepare from lunar resources; also, copper is rare on the Moon (8 to 31 ppm by weight; Phinney et al, 1977). A new approach to PC board manufacture is necessary. Two possibilities are basalt rock slabs and silane-coated basalt fibers (Green, personal communication, 1980). Basalt is an excellent insulator and can be drilled and aluminized to form an etchable conductive surface (Green, personal communication, 1980; Naumann, personal communication, 1980). Boards made of silane-coated basalt fibers would be lighter and easier to drill, but it is unknown whether aluminum can be vapor deposited onto such a surface. If evaporation problems should arise, a thin layer of titanium could serve as an excellent deposition primer (Glaser and Subak-Sharpe, 1977). Ion beam etching might be used selectively to remove aluminum to form any desired circuit pattern. This process is likely to be amenable to precision computer control.

Wiring. The lunar availability of aluminum will permit its widespread use as a conductor for PC board claddings and for all space wiring in general. Its low resistivity (2.8 uohm-cm) compares favorably with that of copper (1.8 uohm-cm), and it readily forms a protective anodic oxide upon exposure to air (Glaser and Subak-Sharpe, 1977). The major terrestrial drawback to aluminum conductors is their incompatibility with conventional soldering and welding methods (Glaser and Subak-Sharpe, 1977). Fortunately, the preferred welding techniques for use in space (see section 4.3.1) should bond this metal nicely. Basalt or glass fibers are possible materials for sheathing aluminum wire (Green, personal communication, 1980), and Miller and Smith (1979) have devised a space-qualified wire insulation wrapping machine.

Before leaving the topic of aluminum wire, it should be noted that high-quality inductors also may be made of this material. One class of inductors - transformers - represents a particularly important component of many computer systems. Iron is plentiful on the Moon (4-15% by weight; Phinney et al., 1977) so transformer cores present no serious problems for the proposed electronics components fabrication facility.

4.4.4 Complex products

The ultimate goals of the SMF are independence from terrestrial resupply, in situ production of all components needed to maintain and expand existing space facilities, and the manufacture of high-value products for consumption on Earth (fig. 4.18). Following deployment of the initial starting kit and manufacture of second-generation tools, development of a product line of ever-greater complexity must occur if the ultimate goals are to be attained. The evolution of complex product manufacturing is outlined below with a focus on just a few important potential products typical of each stage of increasing production sophistication.



Platforms. Expansion of the SMF requires a concomitant enlargement of the facility platform. Such construction represents an early evolutionary threshold, a step requiring little materials processing innovation with some advancement in robotics capability. Component parts may be manufactured from cast or sintered basalt or from aluminum beams, any of which could be produced by the initial starting kit and second-generation tools embodying a synthesis of advancements which already have occurred in industrial automation and mobile autonomous robotics (Leonard, 1980; Lovelace, personal communication, 1980). Robot mobility studies by the Vought Corporation for Marshall Space Flight Center indicate that construction of space platforms is within the grasp of state-of-the-art automation technology. For instance, robot-compatible fasteners have been developed (Borrego, 1977) and deployed in simulation studies at Langley Research Center (Lovelace, 1980).

Pure glasses and synthetic crystals. The manufacture of complex products containing sophisticated electronic specialized materials may require the preparation of pure glasses and synthetic crystals. Production steps that need to be developed include material separation and sophisticated materials processing.

Consider, for example, the manufacture of synthetic quartz semiconductor materials. Plagioclase first is separated from lunar soil by electrophoresis or other techniques. The refined mineral is then fused and its chemical composition altered to induce quartz to crystallize from the cooling solution. Successful fractionation of quartz from an altered plagioclase melt requires significant advances in the techniques of controlled nucleation, crystallization, and zone refining. Development of a special materials-production capability will permit the manufacture of space-made solar panels, solid-state lasing crystals, fiber optics, and perhaps solar sails. New terrestrial materials techniques such as quick-freezing of molten metals to make "glassy metals" (Giuse and Guida, 1980) may find extensive use in space or on Earth.

Satellites. In-space production of satellite; will require the manufacture of special components for control, observation, and communication, and a significant evolutionary advance in automation technology. Satellites may represent the first highly complicated, coordinated construction challenge to be undertaken entirely by teleoperators or robots in space. The construction of solar-power panels, antennas, and sophisticated computer control and communications modules demands a versatile new manipulator system. This system should be equally adaptable to the high-resolution construction tasks necessary in computer assembly and the lower-resolution, high-spatial-range construction jobs required for the assembly of hulls, antennas, and solar panels. Current capabilities of automated assembly are not yet sufficiently well-developed to enable construction of a complete satellite from its constituent parts (Holland et al., 1979; Leonard, 1980; OAST, 1980; Vought Corporation, 1980).

Robots and teleoperators. Two of the most important advanced products to be manufactured in space are robots and teleoperator mechanisms. The ultimate goals for SMF cannot be attained without a significant expansion of the automation equipment initially deployed from Earth. Space robots and teleoperators eventually must be designed from working experience following initial deployment of the starting kit, and then manufactured in space. These second- and third-generation devices must be far more versatile and fault-tolerant than present-day machines. Logistics requirements for production of equipment of this complexity are staggering. The design must incorporate new features based on earlier experiences with robots and teleoperators in space facilities, and should include either a high degree of self-preservation "instinct" or else a highly adaptive servofeedback system using extensive space computer facilities as decisionmakers.

The manufacture of robots and teleoperators in space necessitates the automated production of intricate component parts, a task of far greater complexity than current automated assembly systems can handle (Hart, personal communication, 1980). Automated assembly of advanced devices is perhaps no more difficult than the automated assembly of satellites, which already will have been accomplished during an earlier phase of SMF evolution. The most crucial technologies to be developed for the manufacture of second- and third-generation robots and teleoperators are space-adaptive sensors and computer vision. The current state of machine tactile and vision sensor research is insufficient for sophisticated space robots and automated assembly operations (Holland et al., 1979). The best computer-vision package currently available, CONSIGHT-I, can determine the position and orientation of a wide variety of parts with preprogrammed specifications (Holland et at, 1979). Enhanced decisionmaking and self-preservation features must be added to computer-vision systems such as CONSIGHT-I for use in space robots and teleoperators. A dedicated computer for teleoperator control, programmed to make decisions based on previous experience and insight, would be an instrumental achievement requiring levels of heurism and hypothesis formation unavailable in present-day software (Sacerdoti, 1979).

Solar sails. The solar sails briefly mentioned in section 4.3.1 constitute an unusual but provocative complex product which might be manufactured at the SMF. Sails with a design capability of delivering about two 200-ton payloads per year to the heliocentric distance of Mars have been proposed (Drexler, 1980). Assuming that the viability of self-replicating factories has been demonstrated on the Moon by this point in time (see chapter 5), an interesting scenario would involve the transport of 100-ton self-reproducing "seed" machines (Freitas, 1980c; Freitas and Zachary, 1981) from a lunar-source facility to other moons and planets in the Solar System.

Other complex products. A number of complex products representing various evolutionary steps not yet mentioned or discussed might include impulse landers, biological products, storage tanks, mobile rovers, nuclear power stations, agricultural products, and many others integral to the evolution of a complex products manufacturing capability. The time sequence of these steps is a function of the desired technologies which must be developed at one stage and integrated at a later stage to make products of ever-increasing complexity.

SMF establishment and growth requires a vigorous parallel development of the three basic materials/energy functions - raw materials and materials processing, manufacturing and technology, and energy production. As the SMF increases in output and creates new net resources, unit output costs should fall and an ever-increasing array of commercially interesting products and services will come into existence. Figure 4.19 and table 4.23 illustrate some of the higher-order systems and services which might be expected ultimately to develop.