Alt-BEAM Archive

Message #03271



To: NSX - evenflow88@hotmail.com, beam@corp.sgi.com
From: Dennison dennlill@buffnet.net
Date: Fri, 7 May 1999 22:40:03 -0400
Subject: [alt-beam] Re: Walker effeciency Mischevious Walker Problems



Hey, first of all, I'm gathering english isn't your first langauge, cause
it's really hard to make out exactly what you mean sometimes. But that
doesn't really bother me too much, now about those complete sentences...
:-)

ANyway,

Question: Did you say the springs DID or DID NOT work. I would be shocked if
they didn't, unless you need stronger springs. Anyway, In my book, springs
are probably the way to go. It's simple, and effective. Other methods such
as tuning it a POT are subject to leg drift anyway, as simple Center of
gravity changes in the robot (like changing the leg shape) can require more
recalibration. Leg drift is both a mechanical and electrical problem. It's
electrical in that the component tolerances can actually cause the microcore
to send un-even driver signals. It's mechanical as in the shape, balance,
etc.. can effect leg performance by determining the center of gravity at any
particular point in the cycle. As for your 470K resistor and implex effect,
don't bother yourself. The effects of this are minimal, if any at all. It
certianly won't help with your leg drift.
Other options included limit sensors which detect the leg and then influence
the microcore to 'move on' and swing the leg around back in to place. (I
like this idea, give the microcore info based on it's internal condition)
You can use mechanical stops, Which are just stops which limit the swing of
the robot, (simple effective), You can Use springs, try tuning it some more,
But what I definitly would recomend is that you play with your leg design
some more, it's sounds mostly like you have an unstable layout.

On a interesting note, for people who test their walkers on a flat floor,
walking animals (and robots for this matter) are least effecient and
effective on flat smooth ground. Our (as in walking creatures, like us)
locomotive ability really is impressive on uneven terrain, which is where
legs excell.

good luck!
Dennison


>Hie,
>
> I dont really know why the angle sweep of the rear leg of my walker
are
>turning more than the other it is suppose to be turning left and right at
>the same angle...ya? but mine no matter how i tune it goes to the right
more
>than the left which makes the robot falls on its butt or fall down....i
>tried using bicore and microcore but the end results is still the
>same.....and i tried using springs too it just can make it the same...is it
>the electronics or the mechanis...the leg design is very symetry
>though..and how do i make the bot adaptive like making it feedback feed
>forward or implex......i heard someone saying putting a 470k resistor on
the
>input and output line of the buffering motor driver to make implex...is
that
>it?and how do i put multiple inputs of neural nets to my microcore so that
>it has a emergent behaviour....thanks(sorrie to bother u guys with so many
>questions)
>
>p/s:if i were to build a four motor walker....hows the walking gait like
>does it go forward on the forward right motor and the rear left motor and
>then backward and vise versa or does it go forward with the forward right
>motor and rear left motor and vice versa with the other diagonal pair of
>gearmotor...any hints cause i am quite confused here
>
>
>______________________________________________________
>
------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/group/alt-beam
http://www.eGroups.com - Simplifying group communications

Home