From: Dani Eder (email@example.com)
Date: Tue Jan 08 2002 - 08:14:42 MST
The obvious answer is that the robot brains are
still too expensive. For example, a robot
vacuum cleaner might need to be as smart as an ant.
An ant has 250,000 neurons in it's brain, which
imples a processing capacity of 25 gigabits per
second (# neurons x 1000 synapses/neuron x 100 Hz
firing rate). To map and track the state of
the brain at a conservative 1 byte per synapse
imples 25 MB of storage. The whole processing
and sensing package for the vacuum cleaner needs
to cost something less than $100 in parts wholesale,
so as not to add more than $200 to the retail price
of the machine. We aren't quite there yet, although
we might be within an order of magnitude.
In industrial jobs, such as painting and welding
car bodies, robots are in fact used. The processing
loads are less for these tasks (the auto body is
a predictable shape vs. a changeable furniture
layout in a house), and the value is higher (replacing
a full time human in a factory vs. replacing an
hour a week for a home vacuum).
The real, useful, Intelligence Amplifier that we
have now is search engines like Google, which
allowed me to find the number of neurons in an
ant brain in about 10 sec.
Do You Yahoo!?
Send FREE video emails in Yahoo! Mail!
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT