Re: Moravec, "Intelligent" Robots, and a Central Robot Database

Alexander 'Sasha' Chislenko (sasha1@netcom.com)
Tue, 04 Nov 1997 18:31:27 -0500


At 05:30 PM 11/4/97 -0500, DOUG.BAILEY@EY.COM wrote:
>I just finished reading Moravec's "Mind Children".  I was intrigued by
>Moravec's idea of robots gradually building a"real world simulation database"
>to help them adapt to the real world without constantly having to reinvent the
>wheel every time they are confronted with a situation.  A good idea, which did
>not seem to occur to Moravec, would be to have each robot in the world to
>download its "simulation database" to a central robot database every night (or
>week or whatever). 

That's my favorite point, Doug. There will be a network of knowledge servers
collecting experiences, generalizing algorithms, beta-testing them on some robots,
and either downloading them into robots on as-needed basis, or just performing
"remote thinking" for them. An individual robot would be just a semi-intelligent
and semi-autonomous front-end (sensor and actuator) to the global system, with
some individual adjustments. People are already working on different aspects
of such systems, as always, without understanding what they are doing.

This thought occurred to me about 3 years ago, after reading Moravec's next
book on Mind Age, and I wrote a response to it, starting from the above
considerations, and taking them in a few different directions.
You can find the essay on my website.
I have more interesting considerations now, but have not written them down,
largely because I do not get much criticism for my papers, so sharing my ideas
doesn't do much to help me advance. I would be quite happy to discuss this
stuff if you are interested though.

---------------------------------------------------------------
Alexander Chislenko <http://www.lucifer.com/~sasha/home.html>
---------------------------------------------------------------