Kinematic Self-Replicating Machines

© 2004 Robert A. Freitas Jr. and Ralph C. Merkle. All Rights Reserved.

Robert A. Freitas Jr., Ralph C. Merkle, Kinematic Self-Replicating Machines, Landes Bioscience, Georgetown, TX, 2004.


3.9 Space Manufacturing Systems with Bootstrapping (1977-present)

One of the earliest advocates of self-replicating manufacturing systems in space was Freeman Dyson, who, in lectures in 1970 (Section 3.6) and 1972 [2934], observed that space might be a more appropriate venue for replicators than the verdant hills of Earth: “The true realm of self-reproducing machinery will be in those regions of the solar system that are inhospitable to man. Machines built of iron, aluminum, and silicon have no need of water. They can flourish and proliferate on the moon or on Mars or among the asteroids, carrying out gigantic industrial projects at no risk to the earth’s ecology. They will feed upon sunlight and rock, needing no other raw material for their construction. They will build in space the freely floating cities that Bernal imagined for human habitation. They will bring oceans of water from the satellites of the outer planets, where it is to be had in abundance, to the inner parts of the solar system where it is needed. Ultimately this water will make even the deserts of Mars bloom, and men will walk there under the open sky breathing air like the air of Earth.”

Following the publication of papers and books [942-945] on the colonization of space by physicist Gerard K. O’Neill (1927-1992) [946]) and others [947-951] in the mid-1970s, the potential utility of manufacturing in space using nonterrestrial materials (materials found in space) became clearer. Since the cost per kilogram to reach low Earth orbit even today remains relatively high (~$10,000/kg), it makes sense to search for ways to ship the smallest possible package of manufacturing equipment into space, and then to use that minimal package to gradually expand its own capabilities over the years, a process called bootstrapping that would not require the construction of large monolithic space colonies [952].

Early proposals advanced by O’Neill in 1977 assumed manned “construction shacks” [948] and made little mention of automation [947]. For example, one proposal [953] envisioned an initial lunar supply base with a mass of “less than 1000 tons, deployed by 25 persons in 4 months, maintained by 10 persons.” The buildup rate for industry in space was explored with the assumption that about 3000 tons of equipment would need to be brought up from Earth. These studies [949-951] had included, but only in a limited way, the concept of “bootstrapping”: the use of industry in space to yield, as finished products, some components identical to those of the industry itself. Partial self-replication was a key to cost-cutting and to rapid exponential growth of industry in space. Earlier studies [2688, 2689] from which the concept of bootstrapping had been excluded had already concluded that 90-96% of the required factory mass could be derived from lunar sources.

The Space Studies Institute (SSI), founded in 1978 by O’Neill to further his research, then turned to the question of reducing the cost of space bootstrapping by resorting to some degree of automation. During 1978-1980, SSI organized a series of workshops [954, 2690] aimed at defining more precisely the minimum size of an industrial facility capable of processing lunar materials into pure elements to serve as feedstock for industry in space. The workshops considered scenarios involving manned and unmanned facilities with preliminary estimates of 15-107 tons for partially self-replicating lunar factories of several different types that could grow with a doubling time of 90 days (0.25 year). Instead of the initial cost estimates in the hundreds of billions of dollars, O’Neill found [954, 2690] that self-replicating systems “appear capable of achieving high levels of productivity for investments considerably less than $10 billion. They would be in the range of the Alaska pipeline ($7 billion), and much lower than the Churchill Falls, Quebec, electric power system, both of which were private ventures.”

Two years after the well-known NASA study on self-replicating lunar factories (Section 3.13) was published, in 1984 O’Neill stated in a book [955] that within about four years SSI would release plans for a space program including self-replicating space manufacturing facilities. Hewitt claims [956] these plans never materialized, but that he published [957] “a logistic analysis of such programs, similar to what SSI had intended.” However, SSI notes [958] that in 1988 “the Institute conducted its third major systems study which, in particular, looked at opportunities to bootstrap space industry from low-Earth orbit to the lunar surface as a prelude to large-scale space industrialization.” The results of this study were published by the Lunar and Planetary Institute in 1988 [959].*

* O’Neill was still interested in replicating systems as late as 1990, when he wrote [960], under the subtitle “Manufacturing Economically Productive Structures in Space: Self-Replication of General Purpose Production Machines,” the following recommendation for the decade ahead: “In its early application in space, self-replication is likely to be employed for three general purpose facilities: mass-drivers, processing plants to generate industrial feed-stock from lunar materials, and general purpose, teleoperated ‘job shops’ capable of building more mass-drivers, processing plants and job shops. As noted earlier, it is not cost effective to carry self-replication to the 100% level. Many of the components of all three types of facilities are complex but light in weight and therefore inexpensive to lift from the Earth. Self-replication should be confined to the heavy, repetitive components of the production facilities. The logic of self-replication for production facilities in space is that a ‘seed’ facility consisting of a mass-driver, processing plant and job shop on the Moon and a processing plant and job shop in space could replicate itself in the sequence 1, 2, 4, 8, 16... A series of seven doublings, starting with an initial set of facilities built on the Earth and weighing less than fifty tons, would lead to the capability of processing about 100,000 tons of lunar material per year into completed structures in space. Because of its urgency in terms of potential payback and its derivation from already existing scientific work and profitable industrial practice, a realistic time scale for the earliest pilot plants which are partially self-replicating industrial systems on the Moon and in space is ten years – the year 2000. As doubling times would be as short as two months, the years 2002-2005 could yield a full-scale system processing 100,000 tons per year or more.”

As for autonomous self-replication, the SSI workshop [2690] judged that “it would be uneconomical and unnecessary to push artificial intelligence to the limit of total machine autonomy, because…a great many jobs can be controlled by humans on Earth, through radio/video links…The indispensable person in space will be the repairman, and it is going beyond present technology to think of replacing him – in every eventuality – by another machine.” An extensive multi-year study published by Criswell [961] in 1980 showed that it was extremely likely that self-contained automated processing plants could produce 99% pure elements and large quantities of oxygen from lunar soil, and the 1980 NASA Summer Study (Section 3.13) provided an opportunity to begin defining the precise nature of a “starting kit” [962] for an orbital space manufacturing facility for the processing of nonterrestrial materials (Figure 3.31). The ARAMIS study [963, 964] during 1981-1983 investigated additional specific methods for automating space manufacturing tasks, including especially the use of teleoperation or telepresence. Other studies of various concepts for automating assembly operations in space [965-967], autonomous systems theory [968], planting robotic “seeds in space” [969], teleoperated microrobots [970], and bootstrapping space industry [957, 1063] or solar power satellite production [1007, 1008] with varying degrees of self-replication have continued sporadically throughout the 1980s and 1990s to the present day. In 1984, roboticist Hans Moravec [971] wrote enthusiastically: “I visualize immensely lucrative self-reproducing robot factories in the asteroids. Solar powered machines would prospect and deliver raw materials to huge, unenclosed, automatic processing plants. Metals, semiconductors and plastics produced there would be converted by robots into components which would be assembled into other robots and structural parts for more plants. Machines would be recycled as they broke. If the reproduction rate is higher than the wear out rate, the system will grow exponentially. A small fraction of the output of materials, components, and whole robots could make someone very, very rich.”

Most recent studies have included minimizing the size of bootstrapping starting kits using freeform fabrication [997], molecular nanotechnology [972-976], and architectures for nanotechnology-based space manufacturing systems such as McKendree’s “logical core architecture” [975, 976]; replicators and nanorobotics in space utilization [977], orbital tower construction [978, 979], lunar development [980, 981] and manned mission support [982]; self-replicating human-inhabited O’Neill space colonies [983, 984]; and other speculations [985, 986] including interstellar exploration [987] and communication (Section 3.11). Friedman [988] reported in 1996 that the board of directors of SSI had approved “Quest for Self-Replicating Systems” as a worthwhile project, but through 2003 [573, 989] still had not yet found any specific research proposals that met the criteria for financial support.

In 1996, Raj Reddy [990] at Carnegie Mellon University said of the field of self-replicating systems in space: “There have been several theoretical studies in this area since the 1950’s. The problem is of some practical interest in areas such as space manufacturing. Rather than uplifting a whole factory, is it possible to have a small set of machine tools that can produce, say, 80% of the parts needed for the factory, using locally available raw materials and assemble it in situ? The solution to this problem of manufacturing on Mars involves many different disciplines, including materials and energy technologies. Research problems in AI include knowledge capture for replication, design for manufacturability, and design of systems with self-monitoring, self diagnosis and self-repair capabilities.”


Last updated on 1 August 2005