Alt-BEAM Archive

Message #03205



To: beam beam@corp.sgi.com
From: Bruce Robinson Bruce_Robinson@bc.sympatico.ca
Date: Tue, 11 May 1999 14:48:09 -0700
Subject: [alt-beam] Re: Incapable robots-Give 'em feet.


> Dennison wrote:
>
> ... I have a few suggestions for building more capable robots.
> Make them larger. With this increased size they will have greater
> weight (A nessesity for nagivating grass and stuff). Give them Paws.
> All our robots are these things balanced on the ends of wire.

One of my long term BEAM goals (gotta learn to crawl first) is to make a
walker than can wander about my back yard. As my back yard (and front
ard, for that matter) consists primarily of forest on the side of a
mountain, your average walker will have problems. I figured something
between the size of a cat and a smallish dog would be about right,
although I don't think extra weight is necessary.

A few reflections on the walking motion of such a critter:

The Tilden patent gives an example of a four-leg/eight motor walker.
Each leg alternately lifts (motor A), moves forward (motor B), lowers
(A), and moves back (B). And each leg is triggered in sequence by a 4Nv
loop. This is similar to the motion of most "legged" mammals. However,
the patent then goes on to state: "... the control circuitry must
synchronize leg movement such that no more than one leg is raised at any
one time; because of this the legs must be somewhat resilient, so that
movement of one leg will draw the torso in the desired direction rather
than being merely resisted by the three stationary legs."

Therein lies a problem -- only one leg moves through its cycle at a
time, so we get a resistance/dragging action.

So what if we did this. Make the lift, forward, lower motions
considerably faster than the backward (walking) motion. A leg would
typically move backward s-l-o-w-l-y, and then lift, swing forward, and
drop very quickly (all three actions completed in 1/3 the time the leg
is in contact with the ground). This means that all three supporting
legs would be moving forward simulatneously. Timing should not be a
problem -- just put a larger resistor on the Nv that moves the leg
backward. But we have to do more than lengthen the delay -- we need to
make the motor run slower as well. And the only way I can think of to do
that efficiently is to lower the voltage. How? Haven't a clue. I'm a
mechanical engineer. Wilf?

I figured it would be pretty hard to do this with less than one motor
per leg, but it occurred to me that Chiu's 5 motor walker might be
adaptable to this type of motion, using different circuitry.

Anyway, that was my thought for the day. By the way, that's how my dog
walks when she's just ambling over for a pet. Her other three gaits have
only two legs in contact with the ground.

My $.03 (Cdn) worth.

Regards,
Bruce

------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/group/alt-beam
http://www.eGroups.com - Simplifying group communications

Home