August 13, 2004



Nanobotic Evolution

Kurzweil AI points to two stories this morning that are especially interesting when considered together:

Protein-Based Nanoactuators

Protein-based nanoactuators can now be controlled rapidly and reversibly by thermoelectric signals. In a living creature, contracting or relaxing of muscle tissue is carried out by motor proteins called actomyosin. Scientists designing nano-scale devices would naturally like to emulate the efficiency and compactness of the muscle-moving molecules. A key issue is the controlled rapid activation of the protein motors through simple means.

And that's what researchers at Florida State University have done. They have set up a flow cell in which motor molecules (which can remain viable for days when refrigerated) can be thermally activated into motion in a controllable and reversible way using only input wires which provide a controlled amount of heat.

And…

When Machines Breed

Paul Layzell is a specialist in the budding field of evolvable hardware. Simply put, he helps machines design themselves, using principles borrowed directly from biological evolution.

Layzell once used genetic algorithms to build an oscillator circuit. Some of the solutions were textbook, but one unusual run designed a circuit to take advantage of the radiated hum of the computer he was working on.

In other words, it cheated. The circuit had hacked the system by becoming a radio.

Evolutionary processes have been used in software design since the 1960s.

What is new, however, is the application of evolutionary processes in the hardware realm. Thanks to reconfigurable devices such as the field programmable gate array (FPGA) -- the microchip designer's equivalent of an Etch A Sketch -- and increasing computational power, researchers who once performed simulations of new circuits with an eye on the clock are suddenly free to let their designs evolve for a while just to see what happens. One might not be sure that one understands how a given circuit achieves what it is supposed to, but if it works, is that really a problem?

This is a huge paradigm shift. We don't have to understand our machines anymore. How long until some enterprising researcher programs a simulation of certain tools that we have at our disposal – protein-based nanoactuators could be one such tool – and then lets the computer evolve a plan for a self-assembling nanobot?

Posted by Stephen Gordon at August 13, 2004 09:49 AM | TrackBack
Comments

Self-replicating machines and other complex systems are ideal subjects for genetic algorithms. I think though that perhaps some programmers in this area should practice "eat your own dogfood" by using genetic algorithms to create an improved genetic algorithm. One of the problems I've found is that most such algorithms when applied over a sufficiently general environment generate a lot of unusable programs. Yet somehow, actual biological genes seem to do a good job of handling mutations and generating beneficial mutations. Perhaps, it is because they mutate more efficiently than mere randomness.

Posted by: Karl Hallowell at August 13, 2004 07:02 PM
Post a comment









Remember personal info?