Alt-BEAM Archive
Message #10848
To: beam@sgiblab.sgi.com
From: George Rix rix.g@bmts.com
Date: Fri, 25 Feb 2000 11:07:14 -0500
Subject: [alt-beam] Re: BEAM: Tendency toward miniaturization
> I don't know about that. Spyder was just too complicated. If you look at the
> robot's construction, it's some serious peice of enginnering. If you read
> DAve's little bio about the robot, he say's it's suspension system was so
> nice that you could actually bounce the robot up and down on it's legs.
That's what I mean!
Especially in the control circuitry, it's too comlex.
The bicore is a very simple (especially compared to Spyder's brain) circuit,
and can yield some nifty results.
> Ouch! in addition, Mark T likes simplicity. The fact is that a two motor
> walker can just walk better than spyder can. Such spyder looks cooler, but
> that's not exactly what Mr. T is looking for. Anyway, as Mark say's himself,
> it's unlikely that another such spyder will ever be build. I know rich was
> trying to build one, I haven't heard about his project but I suspect he will
> run into his own difficulties with his design. One of the points I try to
> make sometimes is that if your going to start building larger robots, your
> going to start to have to do some enginnering. We seriously can't just
> 'scale up' our robots. They aren't going to work like that, we can overlook
> so many things when we build our small insect like robots. Maybe huge
> nervous nets will provide an answer, but the inbetween stages of moderaly
> large nets, probably won't show any evidence of being usefull, and more than
> likely just discourage people. It will be the extremes that show any
> usefullness. (my predicition)
However, if you built a frame with the eight (or however many) motors in it,
there's no scaling involved.
Yes, you have to add some sort of suspension so it won't kill itself, but
with some springs, you can make at least a temporary fix.
And while you may be right about extremes, how are we to know when we
haven't tested?
Richard, get thy Spyder in gear!
It's time for a brain (or backbone) transplant! :)
My personal theory is that by adding 'free' neurons, you'll increase the
walker's capability, but also its complexity.
And at a certain point, the complexity will become the overriding factor
(whether due to weight or to space or whatever else we may not foresee), and
we will have to figure out other ways of making nifty tip-toeing bots.
Anyhow, I recently discovered that I have four motors that are close enough
to each other that I can use them in a walker design.
Now I need four more.
Anybody want a set of used parents?
Good mileage on them... :)
Trading value is 4 duplicate gearmotors :)
Peace out!
Rob Rix
This has been a recording.
Home