## Re: 12 motor microcore

Expand Messages
• This isn t exactly what you are looking for, but it might prove useful. In particular, the section toward the end subtitled An Example . Hope this helps. So
Message 1 of 12 , Jun 3, 2001
This isn't exactly what you are looking for,
but it might prove useful. In particular, the section
toward the end subtitled "An Example".

Hope this helps.

So Long,
Cliff
• ... Hi, John. If you were to use the design concept in Mark Tilden s original patent, you would need 30 neurons. Each leg would have a chain of 4 neurons, and
Message 2 of 12 , Jun 3, 2001
jdpuddy@... wrote:
>
> Hi all
>
> I am trying to design a microcore control circuit for a 12 motor
> walker - just a simple one to walk forwards to start with. I have
> had the idea of using two 6 neuron microcores with one link between
> the two to synchronise them. then one core would control horizontal
> movement and another would control vertical.
>
> However, I read on one of the BEAM sites that about 30 neurons would
> be needed for 12 motors. Can anyone explain this or provide a link
> to a site about 12 motor hexapods?

Hi, John.

If you were to use the design concept in Mark Tilden's original patent,
you would need 30 neurons. Each leg would have a chain of 4 neurons, and
there would be a central hexcore to trigger the legs in sequence. So, 4
x 6 + 6 = 30. If you wanted to use the typical static-balance gait with
3 legs moving forward while the other 3 shift the robot, then you could
use a saturated hexcore for the central controller (with delays = 2 x
the leg circuit delays), so there would be no need for extra startup
circuitry.

Tilden's later robots tended to use simpler configurations, often
driving more than one motor off a single pair of neurons. You could, for
example, use just 2 chains of 4 neurons, each driving 3 sets of 2 motors
(each set operates one leg). This would cause each trio of legs to
operate together, giving you the same gait as the 30-neuron arrangement.
Also, instead of a central "core" to trigger each chain, they could
trigger each other when the signal had travelled halfway down the chain.
(The circuit actual maps out to a microcore with 2 branches of 2
neurons, but it's easier to visualize as 2 chains of 4). So there you
have the same 12 motor walker with the same gait, and only 8 neurons.

What you lose in this configuration is the independence of each leg.
With each motor controlled by its own dedicated pair of neurons, you can
influence the neurons to do such things as lift one leg higher, or have
it stretch forward a little further than it's companions. You lose that
sort of flexibility when neurons drive more than one motor. But on the
other hand, if you aren't going to add the extra circuitry to take
advantage of the more flexible layout, why bother? Keep it simple. And
don't forget, Tilden was very surprised at the adaptability of his
simpler robots.

In any event, have fun :)

Bruce
• although not beamed, here s one site that has a couple of 12 motor hpods http://www.tarry.de/index_us.html ,
Message 3 of 12 , Jun 3, 2001
although not beamed, here's one site that has a couple of 12 motor hpods
http://www.tarry.de/index_us.html
, mechanical structure architecture is
always interesting to look at.

and this general robotics link seems to be working again
http://ranier.hq.nasa.gov/telerobotics_page/coolrobots.html

dave.

In a message dated 06/03/2001 5:16:43 AM Pacific Daylight Time,
jdpuddy@... writes:

to a site about 12 motor hexapods?

Message 4 of 12 , Jun 3, 2001
whoops, i guess the tarry machines had 18 motors, seems like alot have 3dof
legs...

http://www.frasco.demon.co.uk/

dave.
• my current interest is in building a hexapod with two motors and a few bellcranks to distribute the leg motion to the 6 legs. I plan on using a 4 Nv microcore
Message 5 of 12 , Jun 3, 2001
my current interest is in building a hexapod with two motors and a few
bellcranks to distribute the leg motion to the 6 legs. I plan on using a 4 Nv
microcore ("standard" 2- motor, 4 leg walker, miller tutorial, etc. type of
circuit) 2 Nv's to provide the middle leg/motor up/down oscillation, with the
other two Nv's (driving the second motor) providing the forward/back
oscillation for all six legs.

this same circuit should be able to drive a twelve servo hexpod (without
bellcranks, just a few more motor driver circuits instead).

circuit complexity depends on how complex a behavior you're after.

dave.
• Thanks for the input. The end goal is that my hexapod will be able to move each leg independently so will probably end up using the 30 neuron method. as an
Message 6 of 12 , Jun 3, 2001
Thanks for the input. The end goal is that my hexapod will be able
to move each leg independently so will probably end up using the 30
neuron method.

as an aside does anyone in the UK know where I can get some springs
to centre the legs on my 2 motor walker since I haven't got anything
to take apart at the moment - damn my video player still works!!

I have build the microcore nad motor drivers which work first time
but the mechanics are letting me down a bit since the legs and motors
dont quite work - not sure if there is enough torque in the motors
but think springs might work

thanks
• ... have 3dof ... Not at all 1 moter for each degree of freedom, up/down, forward/back and in/out or toward the body/away from the body 3 degrees of freedom
Message 7 of 12 , Jun 3, 2001
dkbovaird@a... wrote:
> whoops, i guess the tarry machines had 18 motors, seems like alot
have 3dof
> legs...
> dave.

Not at all 1 moter for each degree of freedom,
up/down, forward/back and in/out or
toward the body/away from the body
3 degrees of freedom per leg = 3 motors per leg
6 legs X 3 motores per leg = 18 motors

Most hexipods Ive seen have only 2 degrees of
freedom per leg: up/down and forward/back
In this case you would only need 12 motors.
• A ten second search on Google turned up Ashfield in the UK. http://www.ashfield-springs.com/ Give them a try! Regards, Dave Simmons ... From:
Message 8 of 12 , Jun 3, 2001
A ten second search on Google turned up Ashfield in the UK.

http://www.ashfield-springs.com/

Give them a try!

Regards,
Dave Simmons

----- Original Message -----
From: <jdpuddy@...>
To: <beam@yahoogroups.com>
Sent: Sunday, June 03, 2001 2:06 PM
Subject: [beam] Re: 12 motor microcore

> Thanks for the input. The end goal is that my hexapod will be able
> to move each leg independently so will probably end up using the 30
> neuron method.
>
> as an aside does anyone in the UK know where I can get some springs
> to centre the legs on my 2 motor walker since I haven't got anything
> to take apart at the moment - damn my video player still works!!
>
> I have build the microcore nad motor drivers which work first time
> but the mechanics are letting me down a bit since the legs and motors
> dont quite work - not sure if there is enough torque in the motors
> but think springs might work
>
> thanks
>
>
> To unsubscribe from this group, send an email to:
> beam-unsubscribe@egroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
• ... The electronics is the easy part. Seriously. Getting the mechanics just right is very difficult. One problem is that you want your motors just powerful
Message 9 of 12 , Jun 3, 2001
jdpuddy@... wrote:
>
> ... I have build the microcore nad motor drivers which work
> first time but the mechanics are letting me down a bit since
> the legs and motors dont quite work - not sure if there is
> enough torque in the motors but think springs might work ...

The electronics is the easy part. Seriously. Getting the mechanics just
right is very difficult. One problem is that you want your motors just
powerful enough to do their job properly under "normal" conditions, but
you want them slighlty overloaded when they encounter resistance in
their environment. This allows them to adapt more easily to changes in
the enviroment (and I'm refering to three types of feedback here -
mechanical, electro-mechanical, and electronic). The problem is, you
don't know just how much power those motors are going to need to supply

You can get around this to some extent by changing the length of the
legs/limbs. Shorter legs = less torque required to supply the same
linear effort. If your motors have barely enough power to move your
robot now, then you will want very light centering springs.

You can also boost the power of your motors by increasing the voltage
slightly (some H-bridges allow a different voltages for the motors and
the control circuits). But when you do this, your motors turn faster, so
the timing in the control circuits has to be changed. And of course, the
motors consume more energy, so you may need larger batteries. So now the
weight is changing as well, and you have to essentially redesign
everything.

Nature does this very nicely in the larger scales (e.g. mammals) by
allowing the structural elements -- bones & muscles -- to change their
sizes in response to the loads they normally carry. Unfortunately the
only equivalent to growing bones & muscles in BEAM is for us to build,
and rebuild, and rebuild. But as long as it's fun ... :)

Bruce
• sorry, i guess I wasn t specific enough. while looking for links to hexapods, the majority of the sponsored bots seemed to have 3dof legs. Although there may
Message 10 of 12 , Jun 3, 2001
sorry, i guess I wasn't specific enough.

while looking for links to hexapods, the majority of the sponsored bots
seemed to have 3dof legs. Although there may be some redeeming value in
having an in/out dof I don't feel it justifies the cost of the additional
hardware. Of course, if the money is not coming out of your own pocket...

On the other hand, I spent this past weekend in Sedona, Az. at my buddies
place playing his Mech Warrior 4 game. This is inspiring me to go for a
biped, with three motors per leg, similar to a Troody dinobot type of leg.
But I seem to be swaying back and forth over the issue of how to maintain the
balancing act without resorting to a microprocessor controller (and I don't
want one of those things with the overlapping sideways "toes"). RC helicopter
(and sailplane) gyros should be able to do the job...I think. I'll have to
think about it some more. Those mech warriors were pretty neat contraptions
though.

dave.

In a message dated 06/03/2001 1:19:57 PM Pacific Daylight Time,
droidmakr@... writes:

Not at all  1 moter for each degree of freedom,
up/down, forward/back and in/out or

• I wonder if it might be simpler to use legs that are not super-legs and save the 3 DoF legs for the front. Or maybe have a one motor hexapod with bellcranks
Message 11 of 12 , Jun 4, 2001
I wonder if it might be simpler to use legs that are not super-legs and save
the 3 DoF legs for the front.
Or maybe have a one motor hexapod with bellcranks just to move forward and
have an extra motor on each front leg for reaching higher or turning or some
such thing.
RECALL:
1) There is some bot that has 6 legs that are just wires that whip around
based on the idea that a cockroach just moves fast and does not necessarily
place each foot in some optimum position. I think a pix is in "RoboSapiens".
2) Recently someone posted a video of a critterbot climbing through
underbrush with legs that whip around.
3) Rodney Brooks had a bot shown on Scientific American on PBS where each
Hexapod foot carefully decided where to put its foot. That complexity was
really neat-o to watch.
Food for thought. John S.

>From: "Cliff" <droidmakr@...>
>To: beam@yahoogroups.com
>Subject: [beam] Re: 12 motor microcore
>Date: Sun, 03 Jun 2001 20:14:16 -0000
>
>dkbovaird@a... wrote:
> > whoops, i guess the tarry machines had 18 motors, seems like alot
>have 3dof
> > legs...
> > dave.
>
>Not at all 1 moter for each degree of freedom,
>up/down, forward/back and in/out or
>toward the body/away from the body
>3 degrees of freedom per leg = 3 motors per leg
>6 legs X 3 motores per leg = 18 motors
>
>Most hexipods Ive seen have only 2 degrees of
>freedom per leg: up/down and forward/back
>In this case you would only need 12 motors.
>
>
>

_________________________________________________________________