Forrest Bishop (
Tue, 29 Jul 1997 17:19:34 -0500 (CDT)

Mark Crosby wrote:
>"A swarm of suveillance mites slowly descends from the sky ... they
>assemble and process the information into a detailed picture of the
>state of the atmosphere".
>"the technology necessary for constructing a 'smart dust particle' may
>be within reach in 5 years, says Berlin."
>"By merging sensing and actuation with computation and communication,
>MEMS [microelectriomechanical systems] devices can be distributed
>throughout the environment,

>"'How do you program a cloud of dust?'

Subsumption architectures generate an emergent swarm behavior. Genetic
algorithms can also operate on the subsumptive program itself.

Berlin asks.

>How do units synchronize their activities when they are in motion
>relative to each other and spread over large distances? How can
>communication be established and maintained in such a system?

Via cellular automata-style interaction with nearest neighbors, perhaps
overridden with acoustic SIMD or global instructions. In any case, one
has to forego absolute, deterministic control.

>IMO, these are many of the same hard problems that must be solved for
>AI and nanotechnology. MEMS seem to be a practical, near-term test bed
>for experimenting on some of these problems.

Yes it certainly is, but one problem with MEMS is high defect and
failure rates, with the 'dust' for example.

Forrest Bishop