Alt-BEAM Archive
Message #02019
To: James Wilson jameswilson@globalserve.net, Sean crowbar@top.monad.net
From: Dennison dennlill@buffnet.net
Date: Sat, 3 Apr 1999 18:23:02 -0500
Subject: [alt-beam] Re: a simple challenge
I agree, with hardwiring you have to, to some degree, predict nearly all
possible situations. And plan for them. While you do need to plan fro
situations with a CPU it's much easyer to create a robot that can learn
mostly on it's own.
Dennison
>True. If you had read the original message you would have read exactly
>what you wrote. The point is there is a point at which it makes more
>sense to use a CPU to complete tasks instead of trying to wire them.
>The complexity of the tasks will dictate the necessity of a CPU.
>
>I am interested in your opinion and mine are always offered freely,
>however, "is simply saying that one lacks the necessary knowledge" is
>neither fair or productive. If you need to insult me or others who are
>simply expressing ideas and opinions then you are not welcome in this
>forum.
>
>The main focus of using a CPU is to use a language like LISP to program
>in. LISP is a recursive language that is able to alter its own code as
>it is executed. This allows a machine to learn and adapt and have a
>memory, all functions that can not be hard wired in. For example if I
>were to place two light sources on a table one that is significantly
>brighter than the other and place a photovore on the table the behaviour
>would be quite predictable, the vore would simply track the bright
>light. Now if I were to place a predator beside the bright light the
>vore would simply be eaten and that would be the end no matter how many
>times I tried it the same thing would occur. Now if I were to place a
>vore with a CPU and the ability adapt and learn it would at first do the
>same thing but after realizing that at the brightest light source there
>was a bad monster waiting it would not again (maybe after several tries)
>revisit the bright light. This type of behaviour can not be simply hard
>wired and I believe it to be impossible as it requires changes to its
>initial "program" to avoid being eaten.
>
>James
>
>Sean wrote:
>>
>> To say that a robot cannot do something without a CPU is simply saying
that one lacks the necessary knowledge to build a robot without a CPU that
performs the desired task. After all, all a CPU is composed of is simple
logic gates. The advantage a CPU has is that it can be easily modified, and
its function can be changed. BUT! Even the most complicated computer
program can be hard wired, that is, built out of logic components, without a
CPU. As long as there is no necessity to load a different program, there is
no necessity of a CPU.
------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/list/alt-beam
Free Web-based e-mail groups by eGroups.com
Home