Machine Prints 3D Copies Of Itself

TaeKwonDood writes "Automated machines have been around for decades. They have basically been dumb devices that do simple assembly tasks. But RepRap takes that a step further because, instead of assembling pre-fabricated parts, it creates 3-D objects by printing them — squirting molten plastic in layers — and then building them up as the plastic solidifies. It works on coat hooks, door handles and now it can even make working copies ... of itself. The miracle of additive fabrication, coming soon to a robotic overlord near you."

Re:I...

[Sebastien_Bailard]

Fussing that RepRap is not 'perfectly self-replicating' yet is an extremely common criticism. This pedantic but factually true statement glosses over the fact that it's a machine that cheaply and easily makes its own parts*, using inexpensive feedstock. And it can make other useful things. That's the important stuff, which your criticism fails to address.

*Aside from common stuff from a hardware store and an electronics store.

(Yes, I'm a RepRap developer, and yes, that's a cut-and-paste.)

Star Trek Replicator

[sm62704]

The self-printing machine is another step to Star Trek's "matter replicator". Society will have some sweeping changes when physical property is as easy and cheap (or beer-free) as intellectual "property" (imaginary property) is to replicate.

Someone in an RIAA/MPAA thread said that since physical property was getting cheaper and cheaper to manufacture and took less and less people to make that we need to stake our future to IP. I say this is hogwash - I may be creative, but most people aren't.The record labels are already quaint anachronisms, and the movie studios will soon follow as the cost and necessary technical expertise drop. It no longer takes lots of gruntwork to make an album; the band and a guy running the studio is all you need now. What will those who have no creativity do for a living?

Heaven on earth is on its way and technology is bringing it here. And the greedy rich are fighting its arrival tooth and nail. Their sense of entitlement and feelings that they are better than the rest of us is sickening.

Self-replicating? Not by a long shot

[Alwin Henseler]

from what I gather it's only able to print the plastic connection parts, so I'm not sure how this counts as "self-replicating"

It doesn't. I'd say you can call something self-replicating if it can reproduce itself using only the essential raw materials. In this case: plastics, metal(s), energy. Perhaps a lot more ingredients, but at least those.

A good comparison is reproducing an OS in a Linux-From-Scratch style (using only source code, disk space and CPU cycles). *THE* thing you need is a C compiler. But to run that, you need a kernel, and a C library below. Then you need shell scripts to automate it, thus a shell. Most sources include makefiles, therefore you need 'make'. And bigger components use all sorts of preprocessing utilities like awk, lex, sed, grep, and so on. All these programs use a variety of standard utilities for copying/removing files, creating directories, etc. So before you get 'full circle', you need a pretty big set of things to reproduce what you start with (think of a compressed Gentoo stage 1).

Maybe this would be a good idea for an X-Prize kind of challenge: create a factory that makes *any commodity of choice*, and keeps itself working indefinitely using just the raw materials, and energy. That is, repairs/rebuilds machines if they break, does maintenance, etc. Say that the only role of humans would be to hit the 'on' switch, stock up supplies/energy, and to keep roof & walls of the building in place. I suspect that even for the simplest kind of product, the minimum size for such a factory would be *huge* if you include stuff like electronics (create new IC's from raw silicium to replace failed ones).

Perhaps all the required technology to do this already exists, but we're still a long way from putting all those parts together.

Re:Close but no cigar...

[ortholattice]

Whether a machine is considered self-reproducing or not is somewhat subject to interpretation I suppose. Similar issues arise with quines [nyx.net] (self-reproducing programs). For example, consider the classic C quine,

char*f="char*f=%c%s%c;main(){printf(f,34,f,34,10);}%c";main(){printf(f,34,f,34,10);}

For me, this is not a true quine, because there is no "#include<stdio.h>". It will not compile on typical C compilers. (There are longer quines that do have the include.)

Basically, you have to agree on a starting environment and what "self-reproducing" means. Computer viruses might be argued to be better quines than a program that simply prints itself and requires a human (or another program) to take the output and run it again.

Similarly, one might demand that a true self-reproducing machine be able to reproduce itself in the middle of the desert with only the sand as raw material and sunlight for energy. But most people would accept something in between that and the machine described in TA.

Self-reproducing lifeforms have similar issues. It is possible for a very simple "lifeform [ucr.edu]" with only 54 base pairs to be self-reproducing, but only if it parasitic. On the other hand, the simplest known lifeform that can reproduce independently is the Mycoplasma genitalium [wikipedia.org] bacteria with 582970 base pairs. This probably isn't the simplest one that can theoretically exist - it is hard to imagine the right combination out of 4^582970 appearing at random in the pre-life organic soup - but whatever simpler thing existed before it is a mystery.

Re:Close but...

[smaddox]

The picture makes the child device look like a cheap piece of crap. But, i guess if it works, who cares.

I would galdly pay $300 to build on of these if it could build new plastic caps for the back of remote controls.

There are so many little pieces of plastic that break and make a product useless. If I could replace them after an hours work, I would be sooooo happy.

Toxic?

[GreatRedShark]

What sort of molten plastic does it use? Does it start in a liquid state and then dry when it's exposed to the air? Or does it melt solid plastic and then reshape it? I had thought that molten plastic would result in some nasty, maybe toxic vapours. Anyone here know for sure?

Re:Bad business model

[smellsofbikes]

You joke, but in the woodworking industry companies make enormously sophisticated jigs for production of wooden forms with repeating features, and at least one company has sold their jigs with EULA language specifically prohibiting using it to duplicate itself or make other jigs that allow its functionality to be duplicated, AND prohibit resale of the jig. Completely bogus, obviously, but they've gone after people with civil suits for violating the EULA and gotten settlements, I've heard.
So, yeah, people are already trying to figure out how to force profitability on things that can self-replicate in a limited sense.

Re:I...

[fugue]

If it can really print itself, the cost of replication goes down to the cost of parts. This is starting to sound a lot like the GPL's provision of distribution for the cost of media. Are we about to see GPLed hardware?

All it makes is its own brackets

[Animats]

It doesn't make a "copy of itself". It doesn't make the plastic output device. It doesn't make the servomotors, the cables, the metal rods, or the control computer. All it actually makes, in fact, are the brackets used to assemble the other parts. The easy parts.

A manual Bridgeport milling machine, on the other hand, used to be considered "self-replicating". If you have a milling machine, a small foundry, a supply of good quality steel scrap, sand, and fuel, and a skilled machinist, you can eventually make another milling machine and all the foundry equipment. Factories that made Bridgeport milling machines (the design was widely copied) did in fact make them using Bridgeport milling machines. A good 1930s machine shop really can replicate itself with only a supply of good people and raw materials.

This machine is more hype than substance. It's just a mediocre stereolithography machine. If you want to use a good one, and you're in Silicon Valley, sign up with TechShop in Menlo Park. They have one, and it's not used much.

Re:All it makes is its own brackets

[smellsofbikes]

You bring up an interesting point. Lathes and mills, like compilers, can be self-replicating. In a belt-driven machine shop, working with somewhat crude tolerances, you can make a whole new belt-driven machine shop to the same (or maybe even slightly better) tolerances. But a '30's machine shop couldn't make modern vacuum-degassed, oil-impregnated bronze bearing stock, for instance: they could only make machines that had basically the same materials they themselves were made of.
Our ability to work with bulk materials has always lagged somewhat behind our ability to make specific, custom materials, in other words -- consider what I think is the highest point of materials science, directional solidification casting of turbine blades, where we have figured out how to control not only what goes in, but how the molecules structurally relate to one another in three dimensions. To build a universal 3D printer, we have to learn how to print more than just atom-by-atom: we actually have to figure out how to distort atom-by-atom printing to establish strain within materials -- and that's just to replicate things we're already building.
Anyone interested in further reading on 3d printers could stand to start by reading Saul Griffith's master thesis (pdf) [saulgriffith.com] on the subject. I'm building a larger version of the LEGO chocolate printer he discusses/documents in there, and I've gotten a couple of jobs by explaining to crabby old machinists how I managed to cut a new, true lathe spindle on my old lathe with a bent headstock spindle. The idea of self-healing and self-replicating machines has always fascinated me.

Re:An Old Programming Problem

[Teancum]

I remember the contest... and somebody turned in a zero byte length source file. Really! Nothing at all as the source.

A standard C compiler (like GCC) will produce precisely an exact copy of the source code when the object code is executed at run-time.

The one thing that kept the contest from getting flooded with additional variations of this software was the requirement to be original and that nobody could use a previously published algorithm. This does make you think, however.