Alt-BEAM Archive
Message #04539
To: beam@sgiblab.sgi.com
From: Mark Lerman mlerman@ix.netcom.com
Date: Sat, 12 Jun 1999 14:33:04 +0000
Subject: [alt-beam] Re: Turning a walker
Just a note re Steven's Spider. If you look in the Edmond Scientific (and
other) catalogs, you will see what they call a 'soccerbot'. This is a small
6 legged, 2 motor robot that works much the way Steven describes. It can
turn on a dime by reversing one motor, and uses a very clever mechanical
linkage.
Mark Lerman
At 08:11 AM 6/12/99 +0200, you wrote:
>On Sat, 12 Jun 1999, Ben Hitchcock wrote:
>
>> Steven Bolt did this a while ago, using two motors, and EIGHT legs.
>>
>> I'm pretty sure you could make a frame with four legs that would
>> work just as well, provided you used the same basic design.
>
>That's not correct. Ken Huntington has tried with six legs and
>appears to be doing well, but some `brain' control of the gait
>becomes necessary, and manoeuvrability may suffer somewhat. The
>wrong gait causes it to flip backwards rather easily. Four legs
>would really not be enough.
>With eight legs, the Spider doesn't need a brain to walk and
>manages quite well in difficult terrain. Lifting its feet about
>2cm, it walks all over the tools and keyboard on my desk, and
>handles tallish gras by kind of swimming over it. There are two
>recognizable gaits:
>
>Big Step - 180 or 0 degree phase difference between any
> left/right pair of legs
>Wave - 90 degree phase difference between any
> left/right pair of legs.
>
>The `wave' looks most natural and avoids peaks in current
>consumption; the average is a little lower than in `big step'.
>The Spider brain will be taught to maintain `wave' when walking
>straight.
>
>The phase of legs on one side has to alternate (the first at 0, the
>next at 180 degrees, then 0 again, followed by 180 for the fourth)
>for walking to be good.
>
>> The interesting thing about this design is that it is a walker,
>> that doesn't need a microcore. All you need is one battery, two
>> motors, and that's it electrically. I would hazard a guess that
>> the gait would 'evolve' over a period of several seconds into
>> something that approaches the alternating diagonal gait.
>
>If left brainless, small differences between the motors will cause
>all gaits to happen in repeating sequence.
>
>> Deceptively simple, this one! You just treat the walker like you
>> would treat a wheeled 'bot - to walk forward, you just run both
>> motors forward. To reverse, you run both in reverse. To turn,
>> you make one run, while the other either stops or goes in
>> reverse.
>
>In a way it is a wheeled `bot, with all `wheels' powered. Most
>people don't consider this approach to be true walking, but it fits
>my requirements:
>
> 1. Everything touching the ground has to be powered. Otherwise
> mobility will be impaired in all but easiest environments.
> 2. More than two motors are out of the question. Adding motors
> increases complexity and thus makes construction more difficult.
> Cost also goes up fast, and not only because good motors are
> expensive; you need more power to move the extra weight, and that
> means larger batteries and/or solar cells.
> 3. The robot must be able to turn around its top axis, that is on
> the spot. Otherwise horizontal mobility just won't be good
> enough.
>
>Best,
>
>Steve
>
> ----------------------------------------------------------------------
> # sbolt@xs4all.nl # Steven Bolt # popular science monthly KIJK #
> ----------------------------------------------------------------------
>
>
>
------------------------------------------------------------------------
eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com
- Simplifying group communications
Home