Self-Replicating Robots 305
ABC News is running a story that self-replicating robots are no longer the stuff of science fiction. Scientists at Cornell University have created small robots that can build copies of themselves. Here is a movie demonstrating the self-replication process. And the paper that will be published in Thursdays issue of Nature.
/.'ed; Coral link to Movie (Score:5, Informative)
The Evolution of Leggo? (Score:4, Informative)
In March 2005, we discovered engineers at the University of Bath working on a machine that can rapid prototype and replicate itself [bath.ac.uk].
Researchers Hod Lipson and Jordan B. Pollack at Brandeis University have coupled inkjet technology and software to autonomously design and fabricate robots [brandeis.edu] without human intervention.
Neil Gershenfeld, director of MIT Center for Bits and Atoms, who runs a one-semester smash-hit class called "How to Make Almost Anything", is determined to produce affordable, replicating personal fabricators by 2025 [blogspot.com].
And today Hod Lipson has announced the arrival of simple self replicating robots with enormous potential.
Applications
More complex shapes are possible in principle, such as adding grippers, cameras, new sensors etc. to modules. A robot could assemble itself into a new structure to deal with novel events. Also points a way to self-repairing robots.
Nanomachines: Lipson is interested in making these machines at microscale. That could drive major advances in Nanotechnology because huge numbers of robots are needed to manufacture things at a molecular scale. Self-replication is how biology does it.
Implications
Could change the way almost everything is manufactured. Machines that clone themselves are a key factor in the near horizon revolution of digital fabrication [blogspot.com].
The movie (accelerated 4X) is eerie to watch. It's easy to imagine a clutter of cubes picking themselves up and walking towards you.
Self replication vs grey goo (Score:5, Informative)
But it does mean that self-replicating robots are, unsurprisingly, possible, and that if the robots could be made simpler, they could perhaps replicate using simpler pieces, and so forth.
More importantly, if you gave the robots a whole bunch of pieces (basically, the equivalent of Lego blocks) they could perhaps replicate and reproduce into shapes that best suit their environment - they're modular and expandable, which might have important applications (e.g., rescue, exploration, etc).
Re:Not replication (Score:2, Informative)
L. S. Penrose. ``Self-reproducing machines.'' Scientific American, Vol. 200, No. 6., pages 105-114, June 1959.
Quote:
In fanciful terms, we visualized the process of mechanical self-replication proceeding somewhat as follows: Suppose we have a sack or some other container full of units jostling one another as the sack is shaken and distorted in all manner of ways. In spite of this, the units remain detached from one another. Then we put into the sack a prearranged connected structure made from units exactly similar to those already within the sack... Now we agitate the sack again in the same random and vigorous manner, with the seed structure jostling about among the neutral units. This time we find that replicas of the seed structure have been assembled from the formerly neutral or ``lifeless'' material.''
Videos of the above exist, but I have no sources. They were shown on a 1980's BBC "Tomorrows World"
More info on research (Score:3, Informative)
Page on their self-replication research [cornell.edu] (coral cache [nyud.net])
Their cubes seem pretty cool... basically a physical variant of cellular automata. The Nature paper is neat but necessarily short. Here's an older paper with some more details:
Designed and Evolved Blueprints For Physical Self-Replicating Machines [nyud.net]
Efstathios Mytilinaios, David Marcus, Mark Desnoyer and Hod Lipson, (2004)
Abstract: Self-replication is a process critical to natural and artificial life, but has been investigated to date mostly in simulation and in abstract systems. The near absence of physical demonstrations of self-replication is due primarily to the lack of a physical substrate in which self-replication can be implemented. This paper proposes a substrate composed of simple modular units, in which both simple and complex machines can construct and be constructed by other machines in the same substrate. A number of designs, both hand crafted and evolved, are proposed.
Re:Edward F. Moore's 1959 self-reproducers (Score:3, Informative)
I haven't read the article though, just seen the title, so maybe Moore had one in the same issue.
Re:More! (Score:2, Informative)
Re:Edward F. Moore's 1959 self-reproducers (Score:3, Informative)
She's his wife.
-Gonz
Re:More! (Score:5, Informative)
See Barry McMullin's paper [eeng.dcu.ie] or Tim Taylor's thesis [ed.ac.uk].
The simple way to do that is to have a "plan" (the genome) that can be read by a "constructor" (the rest of the machine) which follows the plan for building a copy of itself, including the plan. Modifications in the plan lead to modifications in the result. That sounds obvious to us, but Von Neumann wrote about those things more than a decade before the structure of DNA was elucidated.
It also means that the constructor must be, or contain, a Turing machine - a universal computer, making it able to construct anything that can be mechanically constructed out of a program. In living beings, the Turing machine is the result of the complex interactions between proteins that regulate each other's transcriptions and activity. Again, this is obvious to us, but only because Monod and Jacob discovered it in the 70s.
That's why Von Neumann had to invent a very complex structure in a very complex cellular automaton to obtain a really "self-replicating" system (in the interesting sense). That's also why Chris Langton's self-replicating loops are not really "interestingly" self-replicating. And that's why the structures in TFA are even less interestingly self-replicating. Hell, they have to rely on ready-made modules ! They are not even on the same level as simple self-replicating patterns in the Game of Life, wince in the Game of Life new "modules" are constantly created.
The defining factor of life is not self-replication on the global scale. It is the fact that this self-replication occurs by constant self-building. Living systems can build themselves, not out of ready-made modules (babies aren't built by patching together bits of arms, legs, brains, etc) but by breaking down external materials, extracting energy from their environment, then using it to build themselves, in apparent complete contempt the 2nd law of thermodynamics (the key word here is apparent - every single reaction in living beings is completely compatible with the laws of physics, otherwise it wouldn't take place - duh!). Even though the resulting compounds are thermodynamically very unfavorable, they persist because they are constantly replenished by the set of chemical reactions known as "life", which can essentially be defined as autocatalysis resulting in structures with a capacity for evolution.
Hod Lipson is a really great researcher. His work on developmental systems for evolutionary design of structure is so cool it hurts. But I think he and his guys might want to tone down the comparisons with biological self-replication. Right now the structures they have are not even on the same level as the simple patterns that you can see in the Game of Life !