DPRG
DPRG List  



[DPRG] The ideal robot controller

Subject: [DPRG] The ideal robot controller
From: Steve Rainwater srainwater at ncc.com
Date: Tue Nov 29 19:19:10 CST 2011

On Tue, 2011-11-29 at 17:40 -0600, Randy M. Dumse wrote:
> and why doesn't a "better" instrument
> doesn't give better results?

How about an analogy from the world of photography. 

Photographers have historically said that experience, talent and skill
are more important than what camera you use. The canonical quote is
attributed to photographer Henri Cartier-Bresson: "The best photographer
in the world is not as good as the worst camera". My personal experience
confirms this. I've taken photos with a 50 year old, $2 camera found at
an estate sale that rival shots I've taken with modern DSLRs and
multi-thousand dollar lenses.

I think David is suggesting a similar principle applies here. Pretty
much any controller already exceeds the needs of making a robot with
interesting behaviors. Most of our robots don't do much more than Grey
Walter's did in the 1950s with a few vacuum tubes. 

A nematode has about 1 MIPS of computational power; something really
evolved and smart like a dragonfly has about 1,000 MIPS (about 80% of
which is estimated to be for vision). Are our AVR/arduino robots as
smart as a nematodes or our Beagleboard robots demonstrating behaviors
as intelligent as dragonflies? Maybe it's not newer, faster hardware we
need, but a better idea of what to do with the existing hardware (i.e.
learning to write smarter software).


-Steve


More information about the DPRG mailing list