© 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.

3.19 Self-Replicating Robots for Space Solar Power (2000)

A 3-day NSF/NASA workshop [18] was held in Arlington, Virginia in April 2000 to identify fundamental and applied research issues associated with manufacturing and construction of very large systems in space, especially systems designed to convert and deliver massive amounts of solar energy to the Earth’s power grids from either Earth orbit or the Moon. The manufacture and assembly of such systems, as well as the production of photovoltaic cells, major structures, transmission subsystems, and robotic facilities from lunar materials will probably require the use of large numbers of semi-autonomous yet cooperating robots. Reminiscent of the 1979 Telefactors Working Group of the NASA New Directions Workshop (Section 3.13), this new NSF/NASA Workshop generated no new replicator designs but acknowledged that if “means were found to make the manufacturing robots partially or mostly self-replicating, the amount of material brought from Earth for such manufacturing could be greatly reduced.” The Workshop produced a detailed R&D funding recommendation* for research on self-replication totaling $8.2 million [19]. Workshop participants were told that a lunar power system [1007, 1008] might require only about 5,000 robots – not so many that self-replication is absolutely necessary, but certainly a large enough number to require new research in “teleautonomy.”

* Research recommendations on self-replicating systems for lunar manufacturing with funding estimates by the NSF/NASA workshop [19] included the following:

1. Define system-level measures of effectiveness and models which couple the effect of alternative technologies and designs to investment costs and total system effectiveness; ~$0.2M

2. Develop a simulation capability with sufficient detail to determine manufacturing flow networks which are both transitive and intransitive and permit the determination of such parameters as closure, investment costs, production amplification and the degree of autonomy from human involvement; ~$0.5M.

3. Employing the simulation capability in item 2, examine a robotic hierarchical society which involves the natural extrapolations of self and mutual diagnostics, self and mutual repair, parts replacement from spares provisioning, parts replacement from scavenged parts of failed robots, and the construction of entirely new robots – including the reconfiguration of healthy robots – to serve emerging needs. The sea of parts available in the construction vicinity – although a long “bill of materials” – should enable a far longer list of possible robot species. Since this concept is not autotrophic it will not be completely self-replicating, but it should substantially increase the total system reliability and flexibility; ~$1M.

4. Employing the simulation capability in item 2, define the entire process from the mining of available raw material to the finished useful product, including the robotic society. Assuming that the “genome” of the robots will be under full human control, examine alternative approaches to accomplish replication closure: how are the robotic subsystems manufactured and how are their parent machines manufactured, etc. Determine producer/product cost and mass ratios and alternative investment costs for each level of closure. Estimate the optimum level of self-replication investment to maximize mission levels of effectiveness; ~$2M

5. Repeat item 4, except that varying degrees of autonomy – or internalization – of the robotic genomes shall be considered; ~$3M.

6. Migrate the understanding of self-replication attained by the past few decades of research on cellular automata to the understanding of the kinematic model; ~$0.2M

7. Continue to mine von Neumann’s intellectual heritage through scholarly reviews of his work on the general theory of automata, complexity, reliability of large systems with unreliable components, and evolution; ~$0.1M

8. Examine additional biological and “super-biological” analogies which may benefit the mission, including epigenesis (growth and development), immune systems, learning, Lamarkian evolution (passing on acquired characteristics to progeny), language acquisition and reconfigurability; ~$0.2M

9. Encourage the molecular nanotechnology community to accelerate their research into universal constructors useful to the mission; ~$1M

The Workshop [18] identified three principal enabling robotics technology areas common to all space solar power (SSP) generation scenarios, the first two being:

In-situ production of robot parts. “This area involves the automated manufacture, assembly and repair of robot parts and subassemblies, a first step towards making a self-replicating robot…At some point it will become highly desirable to have the machines be able to reproduce themselves. Self-replicating robots will greatly lower the required launch mass to produce an SSP generator. The same technology that is needed to manufacture in-situ parts for the generator should eventually lead to the in-situ production of robot parts and robots.”

Large numbers of cooperating robots. “This area involves the capability of having multiple robots cooperate in completing a task that cannot readily be done by a single robot. Research is needed in both groups of heterogeneous and homogenous robots. Space solar power systems will involve large numbers of robots (tens of thousands) all of which must act in a semi-coordinated fashion…Lunar production systems can be teleoperated/supervised from Earth. As materials extraction, fabrication, and assembly processes become more complex, the autonomous robotic systems should provide greater efficiencies.”

Workshop participants further concluded: “Potential self-replication designs can support the space solar power (SSP) objectives in at least three ways: lowered construction costs, lowered transportation costs by using extraterrestrial material and, most significantly, acceleration of the pace of space development so that humanity can benefit from space power over a time frame of decades rather than centuries.”

A lunar power system proposed in the 1990s by Criswell [1007, 1008] at the Lunar and Planetary Institute would require antipodal arrays of photovoltaic solar collectors with microwave transmitter panels on the lunar surface to be constructed from lunar raw materials using a local mining, processing, and manufacturing infrastructure with assembly and maintenance to be provided by “von Neumann machines”* (i.e., bootstrapping or self-replicating systems). An evaluation by Boeing engineers [1161] of five approaches to space solar power found that Criswell’s proposal “has the highest costs and difficulty, but also has the highest potential to supply a significant part of the world’s energy over the long term.” A partially self-replicating (energy closure only; Section 5.6) lunar-based solar cell factory proposed in 1998 by Ignatiev et al [981] would be “a mobile, lightweight rover equipped with a series of solar concentrators for the volatilization and deposition of regolith-extracted elemental materials....The lunar solar cell production facility would be highly mechanized and remote controlled occupying about 4 m³ at a mass of 300-400 kg. This compares extremely favorably to the current production mass of about 650 kg for 75 kW of silicon solar panels. It would require ~300 W-hr of electrical energy to grow 25 W of solar cells, and hence would have an energy payback time of 10-12 hrs. This makes the in-situ lunar solar cell fabrication technique essentially self-generating in that the cells grown would after 10-12 hr generate excess energy that could be used to grow more cells.

* See related footnote, Section 3.11.

Last updated on 1 August 2005