On 08/12/2007, Mike Tintner <[EMAIL PROTECTED]> wrote:
> Or at any rate, that's my garbled
> impression from cursory reading. There seems to be no way to digitally
> compute/ control & coordinate the movements of complex limbs in real time -
> it's all way too complex. The development of analog models/ body & world
> maps etc is absolutely crucial for AGI. But pure AGI-ers, to judge from our
> discussions, - seem under no pressure or hurry to develop them, (and most
> don't even realise the need) -  whereas roboticists may be.



Arguments over the relative merits of analogue and digital control
have gone on for decades.  It's certainly not true to say that these
things can't be done digitally with computation - in most cases they
are - but it is also possible to achieve similar results for
locomotion and other low level control problems with old-fashioned
analogue electronics.

One of the real geniuses in robotics is Mark Tilden (now working for
WoWee) and for many years he's followed an almost exclusively analogue
approach using very simple inexpensive components cleverly organised
so that the system spontaneously self-organises and for example limbs
move in a coordinated way not because they're directly commanded to do
so but as a dynamic emergent process.  Obviously those wilderness
years as an outspoken advocate of analogue are now paying off for him
in a major way.

However, if you understand what the analogue stuff is doing it is
possible to simulate the same process to arbitrary precision with a
conventional computational implementation.  It just happens that at
present in most cases this more classical route would be too expensive
for a robot toy costing $100, but that won't remain true for long.

-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=8660244&id_secret=73976691-873245

Reply via email to