3D Printing & Self-Replicating Machines

http://singularityhub.com/2009/04/09/3d-printing-and-self-replicating-machines-in-your-living-room-seriously/

One can go further by envisioning a large collection of such self-replicating machines with embedded sensors that collect data on the machine's performance in the field. By that I mean that a set of internal measurements are performed at a constant rate within each machine. That which is being measured is the response of the various components of the machine to field conditions, including its software components. Collectively, these measurements indicate the fitness of the machine's genome to its environment.

This data, in turn, could be transmitted to a centralized location and collected in a data store. Next, genetic and evolutionary algorithms may be invoked to evolve designs for better machines based on the sensor data and the existing blueprints (genome) for the machines. The new designs, in turn, could then be transmitted to the 3D printer for construction.

Note that in many instances, for very complex machines, the evolutionary or genetic algorithms may be invoked only on selected (sub-)parts of the design. In such cases, the entire design may need not be transmitted to the 3D printer but only the differences (departures) from the standard (or pre-existing) blueprint.

As the design changes, the instructions for the construction of the machines, or parts thereof, may also need alteration. Using MS Robotic Studio and the features of .Net Framework such as CodeDOM (see, for example, http://msdn.microsoft.com/en-us/magazine/cc163934.aspx ), one could in principle write software that would automatically generate the new instructions needed for the construction of the machines based on the newly evolved designs.