Re: RP-able robots

From: Jim Mallos (jmallos@wizard.net)
Date: Wed Feb 09 2000 - 14:19:20 EET


Marshall Burns wrote:
>
> On this subject of fabbed (or RP'd) robots, at the 1996 Austin
> conference Ken Hayworth of Ennex Corporation showed a robot finger fabbed in
> the then-new prototype Offset fabber. He took advantage of the fabber's
> ability to stop in mid-process to allow him to lay control cables into
> designed-in channels, which were then covered over and closed by continuing
> the fabbing process over top of them. The paper includes a fairly detailed
> description of how the finger was made, along with a picture of the CAD
> design and a photo of the actual finger. At the conference Ken showed a
> video of the finger operating. The paper from the conference proceedings can
> also be found at http://www.Ennex.com/Technology/paper.sht (look in the
> section, "Applications" towards the end of the paper).
>

Dear Marshall,

Thank you for pointing out to me this igenious and pioneering work in
hybrid fabricating. This in some ways reminds me of the hybrid
(thick-film) integrated circuit which combines some true integration
with micro-assembly methods and has solved many problems that the usual
integrated circuit could not touch.

Nonetheless, hybrid methods would not be included in a concise history
of the integrated circuit revolution. The main storyline was: making the
simplest possible circuits entirely in silicon, incrementally increasing
the complexity and variety of what could be done entirely in silicon,
and, in what is a still on-going and decades-long process, re-inventing
circuit design to exploit the new set of "easies-and-hards" that you
have when you live on a silicon chip.

The last seems too little appreciated. The circuits inside IC's are
fundamentally altered from the old world of discrete design. Inductors,
once considered essential, have undergone a mass extinction. Active
components have come to vastly outnumber passive components. An
amplification stage that might have used one or two transistors before
now uses scores. Error-tolerant switching techniques have replaced
balance-critical linear ones. Most of the techniques used to achieve
this transformation were already known in when IC's came onto the scene.
It is the new set of "easies-and-hards" that now lets them flourish.

Most people do not realize that microtransistors are among the most
crudely shaped artifacts that have ever been made for commerce: they
resemble nothing so much as a tray of homemade buns. The drive to make
them smaller is relentless---it just does not pay to make one
good-looking transistor when you can make ten good-enough transistors
with the same silicon.

Mechanical engineering is going to undergo a similar revolution in this
age of integrated fabrication. Now, when we get a new RP machine, it
won't do to run off a version of the old gears first used in wooden
clocks.

In this age all of the stratagems of the mechanical engineer will be
ranked according to the number of voxels they require. The ones at the
top of the list will get the most use, the ones at the bottom are headed
for the museum of technology.

The changes we can expect are not small:

The wheel, along with its relatives the gear and the full-rotation
bearing, require 100,000's of voxels to turn smoothly. A flex bearing
might be made to get by with less than ten---the Wheel Age is over.

Flexible pneumatics and fluidics can do interesting things with a tiny
ration of voxels---goodbye pistons, cables and mechanical logic.

Electrostatic transducers that are rarely seen today need vastly fewer
voxels than their ubiquitous electromagnetic counterparts---hello new
world.

Regards,

Jim Mallos

For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/



This archive was generated by hypermail 2.1.2 : Tue Jun 05 2001 - 23:02:48 EEST