Every morning I sit down at my computer and work on an idea that could change the world: Automated factories that can grow, like an acorn can grow into an oak tree.
Like any factory, these factories would produce useful products, but they would also make parts and materials for their own expansion. From a starter kit, which I call a "Seed Factory," the factory takes design files and turns them into more equipment. It does this by copying existing parts (replication), making parts for new equipment not in the starter set (diversification), and making larger versions of what it already has (scaling).
As the factory grows and has more equipment, it can automate more of the steps from raw materials to finished product, and make a higher percentage of its own parts. The disruptive change is from linear production—where a factory produces a given product at a given rate, to exponentially expanding production, including making more Seed Factories. Some people worry that robots and automation are going to take their jobs. I say let them. If you own the automation, you have nothing to worry about. Self-expanding factories that grow from a relatively small and inexpensive starter kit can make that possible.
So how did a rocket scientist end up working on this? Not surprisingly, it started with space projects. I have always been interested in space. The Space Age began just 3 weeks before I was conceived in 1957, and one of my earliest memories is watching a rocket launch on TV. Science fiction and the original Star Trek series in the 1960's fed that interest, and by the time I was in college I knew that is what I wanted to do for a living.
Space was, and still is, too expensive for ordinary people to do on their own. So fresh physics degree in hand, I did what any young engineer in those days did. I went to work for a big government contractor, because NASA and their contractors were the only game in town if you wanted to work on space. But I didn't just want to work on space in general, or become an astronaut in the NASA hero mold. My imagination was fired in the 1970's by the first Star Wars movie, and the studies of space colonies started by Gerard O'Neill. In those, space was where lots of ordinary people lived and worked. For that to happen, space had to get way cheaper.
You see, the cost of doing anything in space is still dominated by shipping costs. That cost is made up of the weight of equipment and people you want to send, and how much it costs to send each kilogram. Space systems engineers always try to make things as light as possible, so the big leverage seemed to be in the shipping cost. The Space Shuttle needed $18,500 to launch one kilogram to orbit, or 41 percent of the current price of gold at $45,000 per kilogram. When you literally measure cost in terms of its weight in gold, regular people just can't afford it.
Lots of people are working on ways to lower the shipping cost, but hovering in the background was the idea of reducing how much you need to send. NASA calls it ISRU, which is short for In-Situ Resource Utilization, but the rest of us can call it space mining, or "using what’s already there." The idea in fiction dates back at least to 1898's Edison's Conquest of Mars, and serious consideration goes back at least to a 1975 NASA study of a space colony.
Mining equipment, and the factories to process the raw materials into finished products, are heavy though. So another NASA study, in 1980 added the idea of a self-replicating factory. Instead of taking your heavy industrial equipment to space, you bring a small, lightweight seed factory, which proceeds to copy itself a number of times. When you have enough copies, they switch to building your big mining machines and factories. The idea was good, but 1980-era computers and remote control from Earth were not up to the task of running such a factory, so the idea was shelved. I had copies of both NASA studies, and in 2012 was working on a book about current and future space systems engineering. I realized two things. The first is that automation, robotics, and remote control technology is vastly better now than 1980. We now have the technology to operate self-replicating factories. The second is you can use that idea on Earth too, which NASA never considered because Earth isn't their job.
Once it became obvious that that self-expanding factories would work just as well on Earth as in space, my focus changed to Earth. This is where everything is made today, even space hardware, and this is where many people still don't yet have a decent quality of life.
Though I am still interested in space, I have set that aside for now, and am starting a "Seed Factory Project" to design and build such factories for Earth use. The project builds on existing industrial and manufacturing engineering, and will use robots, automated machine tools, 3D printers, and similar devices in the starter kit. It is too complicated for one person to do all of it, though. So I intend it to be an open-source collaboration, where different people can contribute their knowledge and skills, and end up with a share of the resulting factory.
The first step, which started earlier this year, is writing a book that explains how you design such factories, and goes into some detail on example versions. When the designs are far enough along, I want to start building prototypes of the machines and software. Ideally that would start later this year, but we won't know for sure until we know exactly what to build.
The first design is what I call a "Personal Factory," which would produce food, shelter, and utilities directly for the owners. "Food" means things like a robot tractor and greenhouse, shelter means building materials, and utilities means solar or wind generators. From those outputs, you can then work back to the starter kit which leads to them.
Such a factory would supply the basics you need to live, so losing a paid job is no longer a catastrophe. Since it's mostly automated, you can still work elsewhere if you want, or just take it easy. Future versions of such factories may eventually be used in space, and the ones built on Earth will give valuable experience. For now, though, my feet are firmly planted on the ground, helping design something that meets the most basic needs that people have. That's what gets me up every morning and keeps me motivated.
I'm a systems engineer by profession. I did that kind of work for Boeing for many years. Normally you divide a complex design project into three stages of increasing detail: conceptual, preliminary, and detail design. The Seed Factory work is in the conceptual stage. The end result of this stage is a design concept that identifies the major elements of the factory, their performance and outputs, and a description of how the factory would be operated and maintained.
This is such an enormous undertaking that the FastCo.Labs editors suggested we explain it through question-and-answer. Skepticism about the feasibility of this project is natural, and it’s important to emphasize that these space-age factories aren’t going to be popping up around town anytime soon. The Seed Factory won't reach high levels of automation and integration at the start—at first, it will be just a step above conventional factory automation. The idea is over time you can increase those levels.
Manufacturing is at the beginning of a sea change owed to 3-D printing. Here’s hoping it can help us produce more with less. Now, on to the FAQ.
Why use the metaphor of a "seed" for self-constructing factories?
Biological seeds grow into larger organisms using local matter and energy, and eventually produce copies of the original seed. By analogy a seed factory grows from a small starter set to a larger factory using local matter and energy and high levels of automation. The factory is flexible and general purpose, and produces a variety of useful products. It's also intended to be self-replicating, producing more seed factories.
We talk in terms of a factory rather than a single machine because (1) a number of different production processes are required which are best carried out separately, and (2) for the size and quantity of products we want to make, the final set of equipment is closer to commercial building size than garage or desktop size.
Why hasn’t this been done before?
For any number of reasons, historical, social, political, and economic, we tend to divide production into separate factories—the oil refinery is separate from the steel mill, which is separate from the auto plant. This separation requires tremendous cooperation and coordination among the various manufacturers to bring all of the raw materials and fabricated parts together to assemble the complex *things* of our modern lives. In short, the old manufacturing model makes 21st century optimization impossible.
What's the biggest obstacle to Seed Factories?
There is no simple answer, but it's easy to illustrate the problem. In 1908, Ford revolutionized manufacturing and the automobile market with the introduction of the Model T. There was a popular joke at the time that you could get one in any color, as long as it was black. These days, black may be considered the height of fashion, but only because everything comes in such a wide variety of colors, sizes, shapes, flavors, and so on ad nauseam.
The point being, manufacturing is a much more complex process 100 years later. Our old-style factories have been abandoned, because it seemed impossible to retrofit and update them (though labor issues and profit margins played a large part in this). This has left much of the West with an industrial vacuum; an enormous decrease in manufacturing activity, because we've innovated ourselves out of the business.
How do we reverse the decline in manufacturing?
Automation and optimization! It's only recently that computers have had the raw power to automate such large and complex tasks. Similarly, advances in robotics make old robots look like junkyard wrecks, though it's really advances in programmability and miniaturization that have made the difference. We've got the power and flexible tools to make this happen, and more importantly, to make the changes necessary to keep it moving forward.
Has anyone attempted to realize a Seed Factory?
Partial versions of the Seed Factory exist, though they haven't been recognized as such. For a while now, the manufacturers of industrial robots use their own robots to make new robots. More recently, some advanced builders of automated machine tools use their own products to make more of themselves.
Why should we try this?
We feel current industrial and social organization has deficiencies that self-owned, community-oriented, flexible production can address. Current business models rely on large, specialized factories, in diverse locations, under separate ownership from each other and their employees, with maximizing profit as the primary motivation. This business model creates the following issues:
- Reverse Economies of Scale: Bigger is no longer better. Large, specialized factories tie up enormous capital during construction, aren't easily retrofitted to respond when markets or technology change, and have a disproportionate impact on communities when they become obsolete.
- Large, specialized, factories tie up a lot of capital in their construction, and do not respond well when technology or markets change. If the factory becomes uneconomic, it can have a disproportionate impact on a local community.
Has anyone ever built a Seed Factory that you're aware of?
Civilization as a whole built out its capacity from nothing. Blacksmiths and their descendants, machine shops and foundries, have always been able to reproduce their own tools. But a self-contained automated factory of the type I am working on has not, that I know of. NASA studied the concept in 1982, but computers and automation were not up to the task at the time, so it was shelved. The NASA study also missed applying the concept to Earth. This is understandable, since Earth isn't NASA's job, but this is where the most people and applications will be. The closest examples I know of are companies like HAASCNC, who use their own product, automated machine tools, to make more automated machine tools, and companies that build robots using other robots.
Why do Seed Factories promise such enormous cost savings over regular factories?
The diversity of locations at which we sell things imposes long shipping distances for raw materials and finished products. This consumes energy, takes time, and requires multiple handling at each end of the shipping routes to load and unload. Physical separation makes it more difficult to automate tasks, or integrate flows between operations to gain efficiency. Making matters worse, the separation of suppliers and customers, and owners and employees, tends to make them "other entities" to be taken advantage of, rather than collaborating for mutual gain. Thus the incentives for the owners are to reduce or eliminate wages in order to keep more for themselves, leading to job insecurity for the people who were doing the actual work.
Where exactly do the inefficiencies in present-day manufacturing lie?
Conventional factories tend to be specialized, and you have to ship products from one factory to another at each stage. By bringing the steps together, you can automate the transfer between steps, and also have the potential to integrate processes. For example, making cement for concrete involves high temperature furnaces, and emits CO2. In theory you can use the waste heat and CO2 in other processes instead of just dumping them into the air, as happens today.
How could this change business as we know it?
While increasing profit is well known to be a strong motivator, the modern world is more complex, and other factors are more important than in the past. The single-purpose corporate model of profit above all else is ill-suited to the modern environment. Considered as corporate entities, they must be forced to behave towards other goals against their basic nature. This creates inefficiencies in compliance overhead and inevitable attempts to avoid meeting societal goals. It would be better if goals besides profit were designed into the structure of an organization in the first place.
Is there any historical precedent for this project?
Humans have used tools to make more tools since the Paleolithic (Old Stone Age) development of sharpened stones and controlled fire. The whole of our technological civilization has been built up from simpler starter elements over time. Developing nations since the Industrial Revolution have used the virtuous loop of coal, steel, and steam to increase quantities of all three, and spin off other products powered by them. So in this sense the concept of a growing industrial capacity that makes more of itself is not new. As a continuation of this historical process we plan to use existing tools and machines to build the prototype machines for seed and final factories.
The new aspects to this project are treating the production and end users as an integrated system, purposely designing the elements so they can copy their parts, basing them on widely available and sustainable material and energy resources, and using modern automation and robotics to make them highly productive.
Don’t 3-D printers only make things from plastic?
I expect to include a 3D printer or related additive manufacturing device as part of the Seed Factory elements. But current 3D printers can't make their metal parts, or the plastic filament you feed their print head. For that you need metal-working machine tools and some kind of chemical processing. In turn, metal working machines (mill and lathe) can't make whatever plastic parts they need. The more different machines you have, the more parts they can make for each other. For the starter set, I am looking to spec flexible equipment that can make a wide range of products. That way you can maximize the percentage of self-production.
Will Seed Factories have any human workers at all?
You will never reach 100 percent self-production. Things like computer chips will be cheaper to just buy, and some raw materials just won't be available locally. So as the factory grows, you will have to feed it with the item's it cannot make itself. But from a business standpoint, if your starter set + added parts and materials are 25% of the cost of a full size conventional factory, that is a heck of an incentive.
What’s the software stack look like for a Seed Factory?
An automated factory will require complex software to operate it. At the individual equipment level there exists software for controlling things like machine tools and 3D printers. At the overall factory level there will need to be software and a control network to coordinate things. I have not gotten to the detailed functions the software needs to do, so I cannot say if any existing software can handle it yet. My expectation is there will need to be modified or custom software developed.
How can I follow this project?
The work on this project is being documented in a Wikibook, with the design notes as the last section of the book. I'm using Wikibooks because it makes it easy for other people to collaborate. The book is far from complete at this point, but it puts the data out there for other people to look at. My intent is once the design is sufficiently detailed, to start building prototypes of the factory elements. I've been in touch with an Atlanta area hackerspace about collaborating when it gets to the hardware stage, and gotten a positive response.
The Wikibook is self-documenting, you can click the "view history" tab on the upper right of any page and see all the changes since the page was created. I set up a placeholder website but it does not have much content yet. I posted an unfinished flow diagram in the "files" page. That will show the connections between the various parts of the factory once I finish it. I had not bothered to set up a blog or flesh out the website much, because there just isn't that much to report yet.
How are you funded?
My Bitcoinstarter fund raiser just completed successfully. In addition to the 10+ BTC from that site, someone who found the fundraiser sent me 10.02 coins directly, so I actually got twice the goal. That is not enough to build a factory, but a couple of thousand dollars is a nice start towards an equipment fund.
Excerpts from this article appear in this project’s Wiki, which you can read here.
[Image: Flickr user Steve Jurvetson]