XPost: rec.arts.sf.written, rec.arts.sf.science
From: droleary@8usenet2013.subsume.com
In article ,
Robert Bannister wrote:
> I do some of my navigation from knowing the road layout, but there are
> some places I go where it's more a question of "When I see it, I'll
> recognise it", where "it" is some arrangement of buildings or trees that
> I can't describe and probably don't even consciously remember, but when
> I see it, I know when to turn. Now, I suppose a self-driving car would
> have a database of all these "landmarks", or would it?
It all comes down to what you mean by "database". Your memory of a
scene to know it when you see it *is* your database. A machine can
certainly be made to "remember" the world as it moves through it. The
bonus for the machine, though, is that it can share those memories with
other machines, such that *shared* database is built. Using a paper map
is what humans used to do to accomplish that, and now it's GPS. It's
still easier for machines to build and share that information directly
rather than going through a human bottleneck.
I navigate much the same way you do, and I wager most humans do because
it's a result of our subjective interaction with the world. After all,
we didn't evolve to get around in a world with street addresses or
geographical coordinates. Even today, although I might map out my route
to somewhere new in advance, future travel to the same place is mainly
done using the world view that is in my head. That makes me garbage
when it comes to giving anyone directions, because I know exactly where
things are but have no good way to translate my mental model into
something they can use.
--
iPhone apps that matter: http://appstore.subsume.com/
My personal UDP list: 127.0.0.1, localhost, googlegroups.com, theremailer.net,
and probably your server, too.
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
