Sunday, December 12, 2010

Self-Replicating Solar Robots

I've wanted to write about self-replicating solar robots for awhile.
I've been interested in the idea of sending self-replicating solar robots to the Moon for awhile, but just recently, I read an article on this topic by professor Klaus Lackner. He calls his self-replicating robots, auxons.
While the idea has been around for awhile, it looks like Klaus Lackner and his co-authors (Darryl Butt and Christopher Wendt) have done the best job of thinking through all of the chemical reaction that must occur to derive the materials needed produce self-replicating robots.

A self-replicating robot has to collect enough electricity from sunlight in order to be able to build a duplicate of itself. I've defined the work return on investment as the ratio of the total net work (electrical work in this case) generated over the lifetime of the machine divided by the total exergy (electricity also in this case) to build the machine. A self-replicating solar robot requires a 'work' return on investment greater than one, and for the robot colony to grow the return on investment must be much greater than one. Also important is the rate of return on investment, which is related to the net electricity generated per unit time divided by the upfront electricity consumed in building the solar auxon. (The "net" in the numerator means the gross electricity produced minus reoccurring electricity expenses associated perhaps with labor, maintenance, fuel costs. For the solar auxons, the reoccurring work is the electricity required to move, repair, and clean.) The rate of return is typical given in units of %/yr. You can invert the rate of return to approximately calculate the pay-back time and then double that value to find the time required to double the population of the solar robots.

Lackner and his co-authors found that the solar colony could double in size every few months. Though, their numbers here seem to rather optimistic because the payback time for solar PV panels is on the order of magnitude of 10 years, and that's using PV solar cells chemistry that probably consumes less electricity than the cell chemistry that would have to be used by the self-replicating robots. On the other hand, the doubling rate for blue-green algae is on the order of 20 hours. I have yet to check their numbers, so I am just speculating right now. (I'd like to do a full analysis, perhaps as a class project in a future course I teach.)

I'm particularly interested in the self-replicating solar robots because they seem to be the best way of populating the Moon. (Unlike Mars, it seems unlikely that water-carbon-based lifeforms, such as algae, could survive on the Moon. We could probably populate Mars with the introduction of the strong greenhouse gases [to melt the ice caps] and some algae from the Earth.) If the Moon could be covered with solar robots, we could possibly use the Moon as a staging ground for further exploration of the solar system.

The chemical composition of the Moon (depending on location) is roughly 45% silica (SiO2), 20% alumina (Al2O3) and 10% iron oxide (FeO). The metals in these three materials would make up the main components of the self-replicating robots (Si, Al, & Fe). As Lackner found, one would have to develop innovative chemical processing techniques in order to make the Si, Al and Fe from the oxides.

While it's possible to produce silicon from some electro-chemical reactions, the robots could also use electricity to run a high temperature plasma arc that heats the silicon dioxide to 4000 K, at which point the oxygen is released in the moon's atmosphere. Unfortunately, the oxygen won't stay up there that long because the moon can't hold on to its gases as long as the Earth can. (I calculated that the probability of an oxygen molecules escaping from the Moon's gravity is 424 million times larger than the probability of an oxygen molecules escaping from the Earth's gravity.) Although, it still may be possible to build up a sizable pressure of O2 in the atmosphere once the self-replicating robots start to cover the entire Moon.

Once the solar robots populate the Moon, the electricity they generate could be used to produce hydrogen from water so that we could fuel hydrogen rockets for further travels into the solar system.

My question is: would we consider self-replicating solar robot colonies to be living organisms? I think that self-replicating solar robots fit the definition of life, but I admit that it'd be a primitive form of life unless they could modify their computer program, i.e. their "DNA".

What would be interesting would be to compare the size of the computer program required to be encoded into each robot vs. the size of DNA for blue-green algae. My guess is the sizes of each would be comparable.

In the future, I plan to calculate the return on investment and the rate of return of the self-replicating solar robots in order to estimate how quickly they could cover the entire surface of the Moon.
A back of the envelope calculation of return on investment can be made from: 1) the exergy requirement for producing Si, Al and Fe; 2) the collection efficiency of each robot; 3) the life time that each robot lives; and 4) the amount of Si, Al, Fe required for each robot.


  1. Eddie,
    Self replication and autonomous systems that can be left alone on remote locations are a great solution to the extreme expense of getting mass into space and to other celestial bodies from Earths gravity well. I feel a straight forward analysis of the problem needs to be looked at from an AI(artificial intelligence) or programming/software perspective. The hardware required to replicate and manufacture systems required to further convert energy and manufacture additional systems already exists. Although adapting for the space situation may require some redesign and testing. So the problem becomes performing tasks remotely and autonomously.

    The concept is an efficient one where robots with remote human guidance replicate, manufacture, smelt, purify, collect energy, develop infrastructure...ect. Such that the last piece is simply for the transportation of humans from Earth to the Moon.

    The problem becomes certain scarce resources that must be transported to the moon system. For instance, fundamental chemicals required for manufacturing processes; any type of integrated chips or circuit components that are too complex for replication remotely; and of course lubricants for any moving parts. Petroleum based lubricants or any good alternative are not available on the Moon. There are likely other critical resources that would be required for this concept that would be discovered during the project development process.

    For return on an investment to the global economy there really isn't anything valuable enough that that could be shipped back to earth. Nothing warrants the development costs and price per unit weight transported back from Moon to earth. Event rare earth elements in a purified state(Neodymium...ect) cost more to transport per unit weight than they are worth. Perhaps hazardous production techniques that are to environmentally costly to be conducted cheaply on Earth are a good candidate investment activity. Some possibilities include; nuclear energy, storage of nuclear waste. If environmental impact costs increases and/or are transferred more to businesses then advanced production of consumer products will be too environmentally costly to be done on Earth. This of course would require about 10 orders of magnitude increase in the current environmental impact costs even if one were to use Mercury, Cadmium, Arsenic and Lead and manufacture electronics in downtown Paris. There is also all of the remote locations on Earth that are much easier to get to. Deep Sea, deserts, North Pole....ect.

    In conclusion, this would be sole for Human exploration, colonization and tourism. In which case build the best hotels on the Moon and start charging the very wealthy guests the cost. Richard Branson has it right, with Virgin Galactic.

    I'm done.

  2. Bobbie,
    Thanks for the comments.
    You've clearly thought about the 'rationales' for going to space and realize that they are bleak right now.
    My goal in writing this post is to point out that the cost of starting a self-replicating power plant on the Moon may be quite small. The Moon has all of the elements required to build solar cells or thermoelectric cells (Si, Fe, Al, C, H, O).

    And if the meaning of life is to consume exergy and generate entropy, then the only way for us to increase our consumption of exergy is to start expanding off the Earth. A good start would be to start placing self-replicating solar auxons on the Moon or Mars. But yes, ultimately, if there were a power plant on the Moon, then it would be easier to build a hotel on the Moon.

    In my mind, the space hotel is not the end, but perhaps it could be a means to an end, which is the expansion of life throughout the universe.

  3. I'm interested in what you find out about the doubling rate on the moon. I'm writing a novel called The Open Skies that involves auxons.

    You mentioned that the return on investment of PV cell is ten years, but this is an insignificant cost in a self-replicating system. I've roughly outlined the "cost" of a self-replicating machine: initial system + materials + energy + labor + land + liability + environmental + supervision + r&d + patent + token + tax + shareholders.

    Materials, energy, and labor are part of the initial system but become zero after the system is initiated. Land prices would skyrocket and realtors would reap the most value of the system unless government land and eminent domain is taken into account. Liability is a huge issue if one too many doublings occur. Environmental costs can be positive (carbon elimination, desalination) and negative but are not assigned a dollar amount unless the system owners are sued hence it would be a liability issue. Of course, they could sell carbon credits and water so it becomes a negative cost.

    Supervision is required because people have to tell the robots what to build and when and where and what else to make besides itself. The owners of the system also must determine what products can be produced exponentially, linearly, or a combination.

    Improvements to the system and products manufactured also need to be determined. R&D costs would be any government research. A US based auxon system would have free access to this research. Patents would involve corporate owned innovations. In my novel, the shareholders are forced to carry out any and all improvements to the system unless there is a significant cost to adding the patent. An AK-47 that costs pennies to manufacture and $5 to distribute and retail does not require a five dollar patent that provides a better grip. Most likely a patent would be bought outright in its entirety if it needed to be produced exponentially.

    A token cost is made up costs. People doing work that is unnecessary. Taxes and shareholder dividends should be self-explanatory.

  4. Hello I am so delighted I located your blog, I really located you by mistake, while I was watching on google for something else, Anyways I am here now and could just like to say thank for a tremendous post and a all round entertaining website. Please do keep up the great work. Zonnepanelen kopen