| |
Re: The Clague - Sophie Project, has now become,
The Clague - Sophie - Hassenplug Combine
Get out your Pen and Paper kiddies, this is gonna be good.
Gents, here is what I propose.
I've been sorting my White and Dark Grey.
I'm itching to make a Scorpion.
I'll work with you both to make this pretty nifty.
I'll rig the mechanics and framework.
Install the pistons, switches, motors and sensors.
RCX and Spybots on hand.
It will require compressors be built in the body somewhere.
You'll be able to do that Kevin. Plus link the pistons and switches in the Arms,
and Tail to a pnuematic nervous system. You can alternate or combine the flow of
air to make the 3 limbs move about in whatever sequence you choose. That is your
Department.
Steve will be posted on what mechanics and hardware is involved.
Preparing you to coordinate Motor and sensor controls Steve. You'll need to
inspect the biometrical range of motion for the Legs and other motor controls.
Correlate the sensors at each joint as you see fit. You'll need to make
suggestions on working with devices such as the Spybots for remote capabilities.
That is your challange, to correlate the devices I'll use to attain the
biomechanics.
Each of us will spend time with the actual model. We must agree to limit the
time we each have the model. So we can shoot for a Brick Fest Demo.
Take heart. I could have a Deca-pod frame in Kevin's hands quick, don't temp me.
Though sadly, money for shipping is my only weak spot. The Lego I can cover.
Perhaps we can come to an arrangement.
Gents. I can produce a Scorpion Frame, loose at the Arms and Tail ready for
pneumatic circut logic to be installed, and with Data and values for sensors,
motors and device port controls. Each of you will work on the model utilizing
your expertise.
Kevin will Pnuemo King it, Steve will coordinate the articulation and sensor
work, I'll flick some SNOT on it to finish it off, and if it doesn't work, Rob
Limbaugh will fix it.
Then when we are done we can send it to Chris Masclet.
;)
I'm gonna go do some scale studies.
e
www.biomechanicalbricks.com
|
|
| |
In lugnet.technic, Eric Sophie wrote:
>
> Then when we are done we can send it to Chris Masclet.
>
> ;)
Hi Eric
Best American builders, no parts quantity limit.... It seems I cannot fight !!
;-)
Chris.
|
|
| |
This will be interesting because it is my first partner build..... and I know of
no pattern to follow. We each have our own approaches and I'm not sure how
similar they are. I figured we might as well do the designing process here on
LUGNET, rather than behind the scenes. If people object, we can take this
conversation offline.
For me, because the pneumatics control plays such a large part, the overall
walking mechanism can have a huge impact on the body design. Now with the
Sophie effect coming into play, we get to consider much more. I like it.
Here is a site that has lots of scorpion pictures. I'll leave it to Eric's
artistic eye to pick the species we're going for.
http://www.ub.ntnu.no/scorpion-files/gallery.php
I start with the overall concept of the walking mechanism..... Inarguably a
scorpion has eight legs, and does not use its own body as a leg, so the body
must stay off the ground during the walking process. I tried unsuccessfully to
find movie clips of scorpions walking, but no lock, so I'll make some
assumptions.
1. Typically they walk forward with two groups of four legs.
2. Turning is done by using a different sweep, primarily with the front two
and back two legs.
If we want to avoid gravity wells, we can use the hex242 circuit with extensions
to support forward and backward movement. We might have to have some special
help for turning (maybe bias the sweep forward or backwards).
The leg sweep will be similar to the way that an oar on a rowboat moves.
Down,back,up,forward.
This requires two pistons (plus two switches each) per leg. One for
forward/backward, and the other for up/down. Each piston has to control two
switches that basically instrument the piston state for the rest of the circuit.
The switches can be anywhere as long as they are mechanically linked to the
expansion and contraction of the piston.
If SSCorpion gets too heavy we can double or triple up pistons for a given
degree of freedon. The multiple pinstons will be mechanically linked so they
all expand at the same rate.
Two groups of four legs is a simple extension to hex242 and hex363.
Scorpion's body looks oval, close to rectangular, so the leg sweeps should be
all about the same mechansim.
The walking mechanism will be completely wired into pneumatics. For walking
backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
The RCX may need to know that we are in the right state before it changes the
muxes, so that will take some kind of sensor.
Other kinds of articulation could include the claw arms, and/or the tail.
The tail is pretty rigid unless a scorpion is threatened, then the scorpion
tries to make itself bug by standing taller, bringing the tail far forward, and
raising its claws. This is not part of the regular walking mechanism, so we'll
have to figure out if this is pneumatic (my guess for now is no.)
I think that we have a lot of freedom as to the shape of the body, so we need to
work on the hip and leg joints and leg structure.
Eric, do you have enough to work up body and leg stuff? My only constraints are
that there are two switches per piston, and two pistons per leg. Feel free to
add more pistons if you think they are needed.
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> This will be interesting
-snip-
This is facinating, I am preparing a more in depth response at this moment.
e
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> This will be interesting because it is my first partner build.....
I assume you're not counting the Team Magno fiasco. I guess there really wasn't
much building involved there. :)
> For me, because the pneumatics control plays such a large part, the overall
> walking mechanism can have a huge impact on the body design. Now with the
> Sophie effect coming into play, we get to consider much more. I like it.
...
>
> 1. Typically they walk forward with two groups of four legs.
> 2. Turning is done by using a different sweep, primarily with the front two
> and back two legs.
...
> Two groups of four legs is a simple extension to hex242 and hex363.
Will this type of design allow it to turn in place, or can it only turn while
walking forward/backward?
I think it should be able to turn in place. That could mean dis-connecting the
legs on each side.
Also, moving sideways would be cool. That would mean adding a side-to-side
piston, to go along with the forward/back piston on each leg. Right?
Thoughts?
>
> The walking mechanism will be completely wired into pneumatics. For walking
> backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
> The RCX may need to know that we are in the right state before it changes the
> muxes, so that will take some kind of sensor.
The control could be Motor->pneumatic switch for:
A) Forward/Backward/Stop (x2) (one for each side, as in diff drive)
or
B) Forward/Backward and a second switch for Stop/Go (x2)
or
C) Forward/Backward, Stop/Go, Same/Opposide (legs on both sides go same
direction or opposite direction)
Any thoughts about what works well, Kevin?
> Other kinds of articulation could include the claw arms, and/or the tail.
The claw is a must. As is curling the tail.
But, do the arms move?
As I said, I'd like to see sensors on the arms, somewhat like antenna, so when
it's being autonomous, it can detect objects. And on the front feet, so it can
avoid falling in holes, or walking off tables.
The first challenge I see from our brief discussions, is that Eric builds
backward. :) I think both Kevin and I start from the inside, and build out.
While Eric starts on the outside, and builds in.
I usually worry about how my robot look, but that's secondary to how they work.
This is fun.
Steve
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> This will be interesting because it is my first partner build..... and I know of
> no pattern to follow. We each have our own approaches and I'm not sure how
> similar they are. I figured we might as well do the designing process here on
> LUGNET, rather than behind the scenes. If people object, we can take this
> conversation offline.
This is my first Co-build as well. But not my first multi-person project.
I have also witness first hand the success you both have had in working with
others and in your own projects.
I'm interested in unlocking my potential. I'm interested in synergy.
> For me, because the pneumatics control plays such a large part, the overall
> walking mechanism can have a huge impact on the body design. Now with the
> Sophie effect coming into play, we get to consider much more. I like it.
blush.
> Here is a site that has lots of scorpion pictures. I'll leave it to Eric's
> artistic eye to pick the species we're going for.
>
> http://www.ub.ntnu.no/scorpion-files/gallery.php
Ahh reference.
> I start with the overall concept of the walking mechanism..... Inarguably a
> scorpion has eight legs, and does not use its own body as a leg, so the body
> must stay off the ground during the walking process. I tried unsuccessfully to
> find movie clips of scorpions walking, but no lock, so I'll make some
> assumptions.
>
> 1. Typically they walk forward with two groups of four legs.
> 2. Turning is done by using a different sweep, primarily with the front two
> and back two legs.
>
> If we want to avoid gravity wells, we can use the hex242 circuit with extensions
> to support forward and backward movement. We might have to have some special
> help for turning (maybe bias the sweep forward or backwards).
>
> The leg sweep will be similar to the way that an oar on a rowboat moves.
> Down,back,up,forward.
>
> This requires two pistons (plus two switches each) per leg. One for
> forward/backward, and the other for up/down. Each piston has to control two
> switches that basically instrument the piston state for the rest of the circuit.
> The switches can be anywhere as long as they are mechanically linked to the
> expansion and contraction of the piston.
>
> If SSCorpion gets too heavy we can double or triple up pistons for a given
> degree of freedon. The multiple pinstons will be mechanically linked so they
> all expand at the same rate.
>
> Two groups of four legs is a simple extension to hex242 and hex363.
excellent.
> Scorpion's body looks oval, close to rectangular, so the leg sweeps should be
> all about the same mechansim.
>
> The walking mechanism will be completely wired into pneumatics. For walking
> backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
> The RCX may need to know that we are in the right state before it changes the
> muxes, so that will take some kind of sensor.
we can work a couple of tricks for this. it will be cool. and intensive.
> Other kinds of articulation could include the claw arms, and/or the tail.
>
> The tail is pretty rigid unless a scorpion is threatened, then the scorpion
> tries to make itself bug by standing taller, bringing the tail far forward, and
> raising its claws. This is not part of the regular walking mechanism, so we'll
> have to figure out if this is pneumatic (my guess for now is no.)
>
> I think that we have a lot of freedom as to the shape of the body, so we need to
> work on the hip and leg joints and leg structure.
>
> Eric, do you have enough to work up body and leg stuff? My only constraints are
> that there are two switches per piston, and two pistons per leg. Feel free to
> add more pistons if you think they are needed.
>
> Kevin
Ok, the way a Scorpion rears up and curls or extends its tail forward to sting
or threaten, wow, that requires some travel at the upper leg connection at the
body. It would be cool.
I have many thoughts that can be translated into brick form. I see this Grand
Scorpion, over 2 dozen pistons. Technic. Yellow cyclinders. sensors on switches.
hoses. trained all over. bundles routed all over in and out of compressors.
Motors and sensor leads to RCXs'. more technic frame work.
Arms. hoses. pistons. Switches. more then one sensor. Claws. gripy. tactile.
Tail. ready. poised. pistons. master sensor/locator/suite in place of stinger.
Can I build a moc up. or can I help you replicate the legs and help provide
parts for the truss of frame work. What techinc bits do you need. I'll count my
pnuematics.
e
|
|
| |
Kevin L. Clague wrote:
> I figured we might as well do the designing process here on
> LUGNET, rather than behind the scenes. If people object,
> we can take this conversation offline.
Well, since nobody has objected, let me just say up front... PLEASE keep it
on-line. I agree with Steve, this is fun (and that's just as a spectator). With
regards to Steves question about "decoupling" the left and right sides, Kevin,
could you use an RCX as a higher level of control? Subsumption design, with one
layer being pneumatic, and a supervisory level (that could de-couple and
re-couple the pneumatic circuits on either side, for instance... have to sense
what state those circuits are in) that is electronic in the RCX / sensor net? I
would think mechanical switches either on critical legs, or, (perhaps better
from the standpoint of locating the sensor packages), pressure sensors on
critical hoses, could give the RCX feedback. Use the RCX IR/proximity detection,
and it's a cyclops SSCorpian with very simple vision (detect moving objects for
attack? hehehe...).
--
Brian Davis
|
|
| |
In lugnet.technic, Eric Sophie wrote:
>
> This is my first Co-build as well. But not my first multi-person project.
> I have also witness first hand the success you both have had in working with
> others and in your own projects.
>
> I'm interested in unlocking my potential. I'm interested in synergy.
synergy. cool. :)
> Ok, the way a Scorpion rears up and curls or extends its tail forward to sting
> or threaten, wow, that requires some travel at the upper leg connection at the
> body. It would be cool.
I hadn't really thougth about standing up on it's hind legs. I don't know how
well it will work, but what if it were to sit down (front legs straight, back
legs fully retracted) then, start straightning legs from the front to the back.
The net effect would (could) be to stand up on it's rear legs & tail.
Nice trick for a dog. Great trick for a scorpion. :)
The balance would need to be just right, and the back legs would need to be
strong enough to lift the whole robot. If we can do it, it would be great.
Steve
|
|
| |
In lugnet.technic, Brian Davis wrote:
> Kevin L. Clague wrote:
> > I figured we might as well do the designing process here on
> > LUGNET, rather than behind the scenes. If people object,
> > we can take this conversation offline.
>
> Well, since nobody has objected, let me just say up front... PLEASE keep it
> on-line. I agree with Steve, this is fun (and that's just as a spectator). With
> regards to Steves question about "decoupling" the left and right sides, Kevin,
> could you use an RCX as a higher level of control? Subsumption design, with one
> layer being pneumatic, and a supervisory level (that could de-couple and
> re-couple the pneumatic circuits on either side, for instance... have to sense
> what state those circuits are in) that is electronic in the RCX / sensor net? I
> would think mechanical switches either on critical legs, or, (perhaps better
> from the standpoint of locating the sensor packages), pressure sensors on
> critical hoses, could give the RCX feedback. Use the RCX IR/proximity detection,
> and it's a cyclops SSCorpian with very simple vision (detect moving objects for
> attack? hehehe...).
Hi Brian,
I figured there would be plenty of lurkers and non-so lurkers on this topic.
One of our goals is to have the scorpion turn in place. Hex242 could do this,
and it will be easy to make SSClagorpion (Eric's choice for name) do this as
well, without the need for the RCX in between the two sides.
Using two pneumatic polarity reversors, we can make SSClagorpion turn right.
WIth two more polarity reversors, we can make it turn left, or walk backwards.
The RCX can control the polarity reversors to modify walking behavior. I've
always thought of complex pneumatic sequencers as a way to work around the RCX's
limited I/O, and have always wanted to use the RCX for the higher brain
functions.
I think that the first step (pardon the pun), which is to prototype out a
strong pneumatic leg with two or three (side walking) degrees of freedom.
We may want to be able to articulate the hips to help SSClagorpion posture for
the attach.
Kevin
|
|
| |
Are you planning to publish building instructions and/or an ldraw version?
Hmm, over 2 dozen cylinders.... Backhoes are on special... Mebbe I'd better get
3.
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > This will be interesting because it is my first partner build.....
>
> I assume you're not counting the Team Magno fiasco. I guess there really wasn't
> much building involved there. :)
>
> > For me, because the pneumatics control plays such a large part, the overall
> > walking mechanism can have a huge impact on the body design. Now with the
> > Sophie effect coming into play, we get to consider much more. I like it.
>
> ...
> >
> > 1. Typically they walk forward with two groups of four legs.
> > 2. Turning is done by using a different sweep, primarily with the front two
> > and back two legs.
With four polarity reversors (16 pneumo switches), we can make it walk
forward, backward, turn right or left. If we used more polarity reversors we
could make one side walk forward, while the other side turns, leading to a
curved walking path.
>
> ...
> > Two groups of four legs is a simple extension to hex242 and hex363.
>
>
> Will this type of design allow it to turn in place, or can it only turn while
> walking forward/backward?
All of the above.
>
> I think it should be able to turn in place. That could mean dis-connecting the
> legs on each side.
Nope, just dilligent reversal of horizontal leg sweep.
>
> Also, moving sideways would be cool. That would mean adding a side-to-side
> piston, to go along with the forward/back piston on each leg. Right?
Yes, I think this would take a third degree of freedom in the legs.
>
> Thoughts?
Plenty of them.
>
>
> >
> > The walking mechanism will be completely wired into pneumatics. For walking
> > backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
> > The RCX may need to know that we are in the right state before it changes the
> > muxes, so that will take some kind of sensor.
>
>
> The control could be Motor->pneumatic switch for:
> A) Forward/Backward/Stop (x2) (one for each side, as in diff drive)
> or
> B) Forward/Backward and a second switch for Stop/Go (x2)
> or
> C) Forward/Backward, Stop/Go, Same/Opposide (legs on both sides go same
> direction or opposite direction)
We have to use polarity reversors in pairs. If we reverse the leg sweep, by
swapping pressure hoses, then we need to swap the AND gate outputs (two
pneumatic switches) also.
For walking backward, we reverse *all* the leg sweeps, and reverse the output of
the sweep complete AND chains. This requires one motor to flip these two
reversors.
For turning, we reverse *half* the leg sweeps, and reverse the AND gate outputs
for the same pistons. We can hose things together so we can use only one
reversor for this. So turning requires one motor to flip these two reversors.
Since you want to stop, you probably want to use a motor and a single pneumo
switch to the *off* position.
>
> Any thoughts about what works well, Kevin?
Yep.
>
>
> > Other kinds of articulation could include the claw arms, and/or the tail.
>
> The claw is a must. As is curling the tail.
>
> But, do the arms move?
>
> As I said, I'd like to see sensors on the arms, somewhat like antenna, so when
> it's being autonomous, it can detect objects. And on the front feet, so it can
> avoid falling in holes, or walking off tables.
>
>
> The first challenge I see from our brief discussions, is that Eric builds
> backward. :) I think both Kevin and I start from the inside, and build out.
> While Eric starts on the outside, and builds in.
>
> I usually worry about how my robot look, but that's secondary to how they work.
Yes, I noticed this too.
I'm going to prototype out some leg designs, but suspect/hope that Eric will
just use these to get started, and will redesign the joints from the outside in.
;^)
>
> This is fun.
>
> Steve
Yes, this is fun!
Kevin
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > This will be interesting because it is my first partner build.....
>
> I assume you're not counting the Team Magno fiasco. I guess there really wasn't
> much building involved there. :)
I built nothing there, I just programmed. For me these two experiences are
*very* different.
Kevin
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Eric Sophie wrote:
> >
> > This is my first Co-build as well. But not my first multi-person project.
> > I have also witness first hand the success you both have had in working with
> > others and in your own projects.
> >
> > I'm interested in unlocking my potential. I'm interested in synergy.
>
> synergy. cool. :)
>
>
> > Ok, the way a Scorpion rears up and curls or extends its tail forward to sting
> > or threaten, wow, that requires some travel at the upper leg connection at the
> > body. It would be cool.
>
> I hadn't really thougth about standing up on it's hind legs. I don't know how
> well it will work, but what if it were to sit down (front legs straight, back
> legs fully retracted) then, start straightning legs from the front to the back.
> The net effect would (could) be to stand up on it's rear legs & tail.
>
> Nice trick for a dog. Great trick for a scorpion. :)
Actually we want it to stand on its front legs. This brings the tail forward.
Scorpions sting forward with their tails, so we need to lower the front end and
raise the backend, plus curl the tail forward more than normal.
>
> The balance would need to be just right, and the back legs would need to be
> strong enough to lift the whole robot. If we can do it, it would be great.
If we add a degree of freedom that makes the hip thrust the leg downward, along
an axis from front to back, this will thrust the back end up.
>
> Steve
Quad242/Hex363 class walkers have a simple formula for the number of pistons and
switches needed. One piston and switch per leg group plus two pistons and four
switches per leg.
Since it looks like SSClagorpion will be a Quad242/Hex363 class walker, it will
take:
Pistons = 2 + 2*8 = 18
Switches = 2 + 4*8 = 34
We need four polarity reversors for forward/backward/right/left controls. Each
reversor takes four switches, so we have to add another 16 switches.
Now we're up to 50 switches and 18 pistons. I know I have this much but I have
a rediculous amount of pneumatics. Thank goodness for bricklink and the 8455!
Side walking will cost us even more......
SSClagorpion (at least the first copy) is coming out of Eric's inventory, so I
think we might be limited by the number of switches he has.
Kevin
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
| |
In lugnet.technic, Kevin L. Clague wrote:
| |
This will be interesting because it is my first partner build.....
|
|
Steve, I think its cool too because you have experience at obtaining objectives
like with the Robotics Challanges you have participated in. As well as you own
work.
| |
| |
1. Typically they walk forward with two groups of four legs.
2. Turning is done by using a different sweep, primarily with the front
two and back two legs.
|
...
| |
Two groups of four legs is a simple extension to hex242 and hex363.
|
Will this type of design allow it to turn in place, or can it only turn while
walking forward/backward?
I think it should be able to turn in place. That could mean dis-connecting
the legs on each side.
|
A nice quad pivot pattern with opposesing sweep movements alternating to change
the body’s direction. See what Kevin can do with the air circuts. I know he has
worked hard to created the alternating leg motion, the tough part is to have a
seperate sub sequence that can be “Switched” into place to begin the turn cycle.
A toughie. Am I right? Oh escpecially since many alternating cycles will have to
take place. The body will be able to spin in it’s own foot print.
Or there are other creative ways to slowly alter the gate to start heading it in
different directions, but that is not practical in terms of trying to attain
agility.
We’ll just see what Kevin comes up with and we’ll work with that.
I also offer the old addage that if you plan for 10 cool things or functions to
be included in a creation, and if you only get 8 or 9 of them nailed down, then
you did ok. There is always compromise with practicallity. Just tossing that
out.
| |
Also, moving sideways would be cool. That would mean adding a side-to-side
piston, to go along with the forward/back piston on each leg. Right?
|
That would help it trun better too? right? if you could that would be cool.
My inital thought as I eleuded too before, is to what level do we want to pursue
this. Do we want to make it super badass and include all kinds of great stuff or
do we want to see if we can just manage letting Kevin create the muscles, I make
it look good and you Steve guides us in sensor placement and articulation
management.
We can try a few paths, its fun to think about how we determine the best route.
My first priority is to allow the muscle movement to be set by Kevin and have
him detail to you Steve the kinds of links that will needed to manage the Leg
Systems.
| |
| |
The walking mechanism will be completely wired into pneumatics. For walking
backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
The RCX may need to know that we are in the right state before it changes the
muxes, so that will take some kind of sensor.
|
The control could be Motor->pneumatic switch for:
A) Forward/Backward/Stop (x2) (one for each side, as in diff drive)
or
B) Forward/Backward and a second switch for Stop/Go (x2)
or
C) Forward/Backward, Stop/Go, Same/Opposide (legs on both sides go same
direction or opposite direction)
Any thoughts about what works well, Kevin?
|
Nice!
| |
| |
Other kinds of articulation could include the claw arms, and/or the tail.
|
The claw is a must. As is curling the tail.
But, do the arms move?
As I said, I’d like to see sensors on the arms, somewhat like antenna, so when
it’s being autonomous, it can detect objects. And on the front feet, so it
can avoid falling in holes, or walking off tables.
|
We’ll totally do that. We can mount Sensor on the front or backside of the Claw
to help detect things. Also, on the tips of the Legs as you suggested.
Could we consider using touch sensors on the tips of the feet so we can
determine when each foot is on the ground, or react if a given leg is not where
it is expected to be? Like edge of the table detection like you suggested?
Also there is a special set of organs at the bottom of the head that sweep the
group that gives the Scorpion more info about its surroundings, we could emulate
that too as an additional set of protection/avoidance sensors.
| |
The first challenge I see from our brief discussions, is that Eric builds
backward. :) I think both Kevin and I start from the inside, and build out.
While Eric starts on the outside, and builds in.
|
ok, sorry but you got that mixed up. I gotta correct you on that one.
You gotta go to school in my Brickshelf Gallery.
All my creations start our with functions and frame work built first. What you
see is the illusion crafted over the mecha underneath. If you thought otherwise
then I have succeeded in my craft.
In other words by that logic do you really think I made Jama beautiful then made
the guts, no way, form and function are considered upon eachother in a combined
way until a sweet song is produced I listen for the song when searching for the
best most beautiful solution I can. IMHO.
Look!
| |
I usually worry about how my robot look, but that’s secondary to how they
work.
|
Make the leap. This is what this project is about. Form and Function synergized.
I never scarifice looks for function or vise versa, so long as scale permits.
If not, I just build bigger and enjoy the enpansion in rules.
cool, this will be fun.
e
p.s. btw, I should predicate this upon Kevin’s desire to bring an additional
organic feel to the systems he has created. He approached me and asked if we
could create something together that used the best of our own merits.
And now I’m so glad you Steve are willing to enter into this triumvirate to help
manage the subsytems and advise. Your ability to help engineer sensor placement
and RCX control will help greatly.
Heck maybe we can get a JCX out of it.
;)
e
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > In lugnet.technic, Eric Sophie wrote:
> > > Ok, the way a Scorpion rears up and curls or extends its tail forward to sting
> > > or threaten, wow, that requires some travel at the upper leg connection at the
> > > body. It would be cool.
> >
> > I hadn't really thougth about standing up on it's hind legs. I don't know how
> > well it will work, but what if it were to sit down (front legs straight, back
> > legs fully retracted) then, start straightning legs from the front to the back.
> > The net effect would (could) be to stand up on it's rear legs & tail.
> >
> > Nice trick for a dog. Great trick for a scorpion. :)
>
> Actually we want it to stand on its front legs. This brings the tail forward.
>
> Scorpions sting forward with their tails, so we need to lower the front end and
> raise the backend, plus curl the tail forward more than normal.
>
> >
> > The balance would need to be just right, and the back legs would need to be
> > strong enough to lift the whole robot. If we can do it, it would be great.
>
> If we add a degree of freedom that makes the hip thrust the leg downward, along
> an axis from front to back, this will thrust the back end up.
We could make this eaily with a Technic Turntables. Depends on how you will
create the legs. It will be your decision.
> >
> > Steve
>
> Quad242/Hex363 class walkers have a simple formula for the number of pistons and
> switches needed. One piston and switch per leg group plus two pistons and four
> switches per leg.
>
> Since it looks like SSClagorpion will be a Quad242/Hex363 class walker, it will
> take:
>
> Pistons = 2 + 2*8 = 18
> Switches = 2 + 4*8 = 34
>
> We need four polarity reversors for forward/backward/right/left controls. Each
> reversor takes four switches, so we have to add another 16 switches.
>
> Now we're up to 50 switches and 18 pistons. I know I have this much but I have
> a rediculous amount of pneumatics. Thank goodness for bricklink and the 8455!
>
> Side walking will cost us even more......
>
> SSClagorpion (at least the first copy) is coming out of Eric's inventory, so I
> think we might be limited by the number of switches he has.
>
> Kevin
Uggg, I don't have that many!
:(
I know I'm talking like I have got it all, but I still gotta give Rob Limbaugh
some. LOL!
heh heh Match ya? heh heh, if we fully consider the process and have commitments
and solid plans, we could, ask for help with parts.
When done in a way to promote building and understanding, we may qualify.
Hint Hint, nudge nudge, people in high places. Cough Cough.
e
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Brian Davis wrote:
> > Kevin L. Clague wrote:
> I think that the first step (pardon the pun), which is to prototype out a
> strong pneumatic leg with two or three (side walking) degrees of freedom.
Yes, that's it! Kevin, Like you said, create the bestest strongest 2 or three
degree of movment prototype leg. Perfect it. We'll determine what parts you need
to build all the legs. As you work on the sequence routing and structures beyond
the leg so we can begin to join them centrally.
I want there to be a centralizing of compressors and others necc. switching
compenents located to the rear of the last set of legs in the back of the body.
This will serve to counter act the weight of the Arms.
So feel free to keep in mind the thorax and tail area are where your valve and
motor processes take place. Works out nice.
> We may want to be able to articulate the hips to help SSClagorpion posture for
> the attach.
If you want to do that, create the legs to have two degrees of movment and I
link the upper part of the Legs you create to the main connection point of the
body with Technic Turntables. So the 3 degrees required for each leg will be
divided into 2 parts pnuematic and one part mechanically articulated.
I mean you could have all 3 degrees per leg, wow, that's insane! But that's the
point right? This is what is possible.
Consider the upper leg connection to be mechanical if you so choose.
your choice.
> Kevin
e
|
|
| |
In lugnet.technic, Benton Jackson wrote:
> Are you planning to publish building instructions and/or an ldraw version?
All we know is that it is part of our goal to share the experience.
It is also tradition for all 3 of us to publish our work in one way or another.
We look forward to seeing what becomes of it.
> Hmm, over 2 dozen cylinders.... Backhoes are on special... Mebbe I'd better get
> 3.
Can we borrow your Lego Benton? Please?
;)
e
|
|
| |
> Can we borrow your Lego Benton? Please?
What, and TAKE APART my backhoe? :P
|
|
| |
In lugnet.technic, Chris Masclet wrote:
> In lugnet.technic, Eric Sophie wrote:
> >
> > Then when we are done we can send it to Chris Masclet.
> >
> > ;)
>
>
>
>
> Hi Eric
>
> Best American builders, no parts quantity limit.... It seems I cannot fight !!
>
> ;-)
>
> Chris.
We will have a toast to you!
Perhaps we can consider the Hand design you developed for LEGOTIC!
It is now Kevin's pride!
What to do!
btw, you have made some interesting electronics. When the time comes feel free
to advise.
e
|
|
| |
In lugnet.technic, Benton Jackson wrote:
> > Can we borrow your Lego Benton? Please?
>
> What, and TAKE APART my backhoe? :P
I'll have Tim Courtney buy you one, nay, two Pints at Brick Fest.
How's that?
e
|
|
| |
In lugnet.technic, Eric Sophie wrote:
> Re: The Clague - Sophie Project, has now become,
> The Clague - Sophie - Hassenplug Combine
Holy $%#*!!! This has instantly become one of the coolest Lego ideas I've seen
in recent memory! What a collaboration! A thread I'll be watching like a hawk.
With the talent of you three guys, makes me wonder where it will go!!!?
After building your deadly scorpion, you'll have to build a tuxedo wearing bot
to play Henry Lim's harpsichord! (hehe) What size tux do you think Jama wears?
Rock on guys! This will be fun to watch..........
-Tommy Rainwater
"Cogito Ergo Lego"
|
|
| |
In lugnet.technic, Eric Sophie wrote:
| |
I also offer the old addage that if you plan for 10 cool things or functions
to be included in a creation, and if you only get 8 or 9 of them nailed down,
then you did ok. There is always compromise with practicallity. Just tossing
that out.
|
Yes. Right now, I think we’re still brainstorming, so we need to come up with
all the cool ideas about what we want to do, then we can figure out what is
feasable, and what we want to skip.
| |
My inital thought as I eleuded too before, is to what level do we want to
pursue this. Do we want to make it super badass and include all kinds of
great stuff
|
Yes. Start with super badass, and trim from there.
| |
| |
As I said, I’d like to see sensors on the arms, somewhat like antenna, so
when it’s being autonomous, it can detect objects. And on the front feet,
so it can avoid falling in holes, or walking off tables.
|
We’ll totally do that. We can mount Sensor on the front or backside of the
Claw to help detect things. Also, on the tips of the Legs as you suggested.
Could we consider using touch sensors on the tips of the feet so we can
determine when each foot is on the ground, or react if a given leg is not
where it is expected to be? Like edge of the table detection like you
suggested?
|
We’ll need touch sensors on the tip of the front leg (foot) and also some way to
detect when the foot should be touching the ground (when it’s at the bottom of
it’s movement)
| |
Also there is a special set of organs at the bottom of the head that sweep
the group that gives the Scorpion more info about its surroundings, we could
emulate that too as an additional set of protection/avoidance sensors.
|
I like Kevin’s idea of going with an all LEGO build. (for now) That keeps us
limited to contact sensors for object detection.
If we do cross over that line, there are many things we can get ahold of from
HiTechnic, like this: http://www.hitechnic.com/developmentlab.htm
| |
All my creations start our with functions and frame work built first. What
you see is the illusion crafted over the mecha underneath. If you thought
otherwise then I have succeeded in my craft.
|
so, you have. :)
Steve
|
|
| |
In lugnet.technic, Eric Sophie wrote:
<snip>
>
> > We may want to be able to articulate the hips to help SSClagorpion posture for
> > the attach.
>
> If you want to do that, create the legs to have two degrees of movment and I
> link the upper part of the Legs you create to the main connection point of the
> body with Technic Turntables. So the 3 degrees required for each leg will be
> divided into 2 parts pnuematic and one part mechanically articulated.
>
> I mean you could have all 3 degrees per leg, wow, that's insane! But that's the
> point right? This is what is possible.
>
> Consider the upper leg connection to be mechanical if you so choose.
> your choice.
I assumed that this third degree of freedom would me motorized.
Kevin
|
|
| |
In lugnet.technic, Tommy Rainwater wrote:
| |
In lugnet.technic, Eric Sophie wrote:
| |
Re: The Clague - Sophie Project, has now become,
The Clague - Sophie - Hassenplug Combine
|
Holy $%#*!!! This has instantly become one of the coolest Lego ideas I’ve seen
in recent memory! What a collaboration! A thread I’ll be watching like a hawk.
With the talent of you three guys, makes me wonder where it will go!!!?
|
Thank you. Each of us I think knows the power of the building system.
I think we know it can happen. It is worthy. Consistant with our interests.
| |
After building your deadly scorpion, you’ll have to build a tuxedo wearing bot
to play Henry Lim’s harpsichord! (hehe) What size tux do you think Jama wears?
|
I don’t know let’s find out!
oh Jama will take a petite size 2 please.
| |
Rock on guys! This will be fun to watch..........
-Tommy Rainwater
“Cogito Ergo Lego”
|
We will make it happen. First the leg motion discovery.
e
|
|
| |
In lugnet.technic, Eric Sophie wrote:
> In lugnet.technic, Benton Jackson wrote:
> > > Can we borrow your Lego Benton? Please?
> >
> > What, and TAKE APART my backhoe? :P
>
> I'll have Tim Courtney buy you one, nay, two Pints at Brick Fest.
You're being quite friendly to someone who's usually just a lurker on lugnet.
Are you mistaking me for the other Ben Jackson, who might have actually been to
LugNet before?
|
|
| |
In lugnet.technic, Eric Sophie wrote:
| |
In lugnet.technic, Tommy Rainwater wrote:
| |
In lugnet.technic, Eric Sophie wrote:
| |
Re: The Clague - Sophie Project, has now become,
The Clague - Sophie - Hassenplug Combine
|
Holy $%#*!!! This has instantly become one of the coolest Lego ideas I’ve
seen in recent memory! What a collaboration! A thread I’ll be watching like
a hawk. With the talent of you three guys, makes me wonder where it will
go!!!?
|
Thank you. Each of us I think knows the power of the building system.
I think we know it can happen. It is worthy. Consistant with our interests.
-snip-
We will make it happen. First the leg motion discovery.
e
|
OKAY, I have not read this thread from the beginning but from the topic snippets
I have to say this is toooo cool! I too will be avidly watching this
manifestation.
Oh, and HELLO e! Long time, no see!
-Andrew
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Eric Sophie wrote:
>
> <snip>
>
> >
> > > We may want to be able to articulate the hips to help SSClagorpion posture for
> > > the attach.
> >
> > If you want to do that, create the legs to have two degrees of movment and I
> > link the upper part of the Legs you create to the main connection point of the
> > body with Technic Turntables. So the 3 degrees required for each leg will be
> > divided into 2 parts pnuematic and one part mechanically articulated.
> >
> > I mean you could have all 3 degrees per leg, wow, that's insane! But that's the
> > point right? This is what is possible.
> >
> > Consider the upper leg connection to be mechanical if you so choose.
> > your choice.
>
> I assumed that this third degree of freedom would me motorized.
>
> Kevin
Oh excellent, then do you feel comfortable...or rather can you create everything
from the Technic Turntables down? Just give me something to grab from off of the
Legs to begin connections to the Turntables. Or you do it, or we could make a
standard. So they'll mate up to the main body.
btw, as you know, using Technic Turntables at this connection point on the body
is a sound way of creating this crucial joint. Work the turntable into the upper
joint of the Leg desgin, if you don't mind me suggesting.
{[On the other hand you could turn the Turntable into a big L-shaped lever and
articulate it pneumatically, with no motors at all. Then it could be part of
your circut. ?}
I'll take it from there. Or rather I can complete all of the internal gearing
and worm drives, along the center axis of the body.
Seriously, you might want to consider this, you know how the Technic Turntable
slipts in half? Well you should build the legs up to the "Top part" of the
Turntable and have it clip into the "Lower part" of the turntable on the main
body truss I create? Same with the Arms and Tail. I just provide the heviest
duty joint at the main body.
This 2 step pnuematic (lower half of leg) , 1 step motorized (upper leg joint)
can be managed by the RCX, it includes the pnuematics valve sensor controls
anyway. It'll all work together.
What do you suggest?
wait until your happy with a leg design ready to attachment to the central body.
You'l need your own Central body truss work to contain the Legs and and do
Testing! Boy this is sounding like alot of work on your end. Sure you don't want
me to build the body frame or perhaps arms and tail? What do you want?
Either way you'll know more once your creating the structure.
whatever works best for you.
e
|
|
| |
In lugnet.technic, Benton Jackson wrote:
> In lugnet.technic, Eric Sophie wrote:
> > In lugnet.technic, Benton Jackson wrote: • -snip-
> Are you mistaking me for the other Ben Jackson, who might have actually been to
> LugNet before?
ahh, Tim will buy you one too.
e
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Brian Davis wrote:
> > Kevin L. Clague wrote:
> > > I figured we might as well do the designing process here on
> > > LUGNET, rather than behind the scenes. If people object,
> > > we can take this conversation offline.
> >
> > Well, since nobody has objected, let me just say up front... PLEASE keep it
> > on-line. I agree with Steve, this is fun (and that's just as a spectator).
>
> I figured there would be plenty of lurkers and non-so lurkers on this topic.
>
> One of our goals is to have the scorpion turn in place. Hex242 could do this,
> and it will be easy to make SSClagorpion (Eric's choice for name) do this as
> well, without the need for the RCX in between the two sides.
I just made up a web page, so I have a place to put MY notes and thoughts.
http://www.teamhassenplug.org/robots/ssclagorpion/
Don't expect much there, yet, we're still pretty early in the project. Feel
free to bookmark the page. It's not going away anytime soon.
Now would be a good time for you two to do the same, if you want to track
progress.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
>
> I just made up a web page, so I have a place to put MY notes and thoughts.
>
> http://www.teamhassenplug.org/robots/ssclagorpion/
>
> Don't expect much there, yet, we're still pretty early in the project. Feel
> free to bookmark the page. It's not going away anytime soon.
>
> Now would be a good time for you two to do the same, if you want to track
> progress.
I like SCorPeon..... It has a nice ring to it.
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> >
> > I just made up a web page, so I have a place to put MY notes and thoughts.
> >
> > http://www.teamhassenplug.org/robots/ssclagorpion/
>
>
> I like SCorPeon..... It has a nice ring to it.
Unfortunately, whenever I see SSClagorpion I think of a boat.
Maybe, we should give Eric the E and call it SCorp-Eon. :)
Or, it could be SCHorpion.
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
Gents, I am going to upstate NY to prepare for an art show I have this weekend.
I will be out of the loop for a few days.
I look forward to picking this up when I return.
e
http://www.lugnet.com/~1051
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > In lugnet.technic, Kevin L. Clague wrote:
> > > This will be interesting because it is my first partner build.....
> >
> > > 1. Typically they walk forward with two groups of four legs.
> > > 2. Turning is done by using a different sweep, primarily with the front two
> > > and back two legs.
>
> With four polarity reversors (16 pneumo switches), we can make it walk
> forward, backward, turn right or left. If we used more polarity reversors we
> could make one side walk forward, while the other side turns, leading to a
> curved walking path.
>
> > ...
> > > Two groups of four legs is a simple extension to hex242 and hex363.
> >
> >
> > Will this type of design allow it to turn in place, or can it only turn while
> > walking forward/backward?
>
> All of the above.
>
> >
> > I think it should be able to turn in place. That could mean dis-connecting the
> > legs on each side.
>
> Nope, just dilligent reversal of horizontal leg sweep.
>
> >
> > Also, moving sideways would be cool. That would mean adding a side-to-side
> > piston, to go along with the forward/back piston on each leg. Right?
>
> Yes, I think this would take a third degree of freedom in the legs.
>
> >
> > Thoughts?
>
> Plenty of them.
>
> > >
> > > The walking mechanism will be completely wired into pneumatics. For walking
> > > backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
> > > The RCX may need to know that we are in the right state before it changes the
> > > muxes, so that will take some kind of sensor.
> >
> >
> > The control could be Motor->pneumatic switch for:
> > C) Forward/Backward, Stop/Go, Same/Opposide (legs on both sides go same
> > direction or opposite direction)
>
>
> We have to use polarity reversors in pairs. If we reverse the leg sweep, by
> swapping pressure hoses, then we need to swap the AND gate outputs (two
> pneumatic switches) also.
>
> For walking backward, we reverse *all* the leg sweeps, and reverse the output of
> the sweep complete AND chains. This requires one motor to flip these two
> reversors.
>
> For turning, we reverse *half* the leg sweeps, and reverse the AND gate outputs
> for the same pistons. We can hose things together so we can use only one
> reversor for this. So turning requires one motor to flip these two reversors.
>
> Since you want to stop, you probably want to use a motor and a single pneumo
> switch to the *off* position.
Ok, so I'm thinking about motors to flip pneumo-switches. Each of these may be
one or more switches, all activated at the same time.
1) Switch for Start/Stop (on/off) When this switch is "on" the scorpion will be
walking. (legs going up/down + some other direction). This switch will activate
all eight legs.
2) Switch to select forward or backward for left legs.
3) Switch to select forward or backward for right legs.
4) Switch to select forward/backward OR side-to-side for left legs.
5) Same as 4 for the right legs.
4&5 are optional, depending on if it CAN go sideways.
Is that right for basic operation?
I'm assuming a bank of legs (right or left) can change sweep direction, but they
will still move up/down when ever it's walking.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > In lugnet.technic, Kevin L. Clague wrote:
> > > > This will be interesting because it is my first partner build.....
> > >
> > > > 1. Typically they walk forward with two groups of four legs.
> > > > 2. Turning is done by using a different sweep, primarily with the front two
> > > > and back two legs.
> >
> > With four polarity reversors (16 pneumo switches), we can make it walk
> > forward, backward, turn right or left. If we used more polarity reversors we
> > could make one side walk forward, while the other side turns, leading to a
> > curved walking path.
> >
> > > ...
> > > > Two groups of four legs is a simple extension to hex242 and hex363.
> > >
> > >
> > > Will this type of design allow it to turn in place, or can it only turn while
> > > walking forward/backward?
> >
> > All of the above.
> >
> > >
> > > I think it should be able to turn in place. That could mean dis-connecting the
> > > legs on each side.
> >
> > Nope, just dilligent reversal of horizontal leg sweep.
> >
> > >
> > > Also, moving sideways would be cool. That would mean adding a side-to-side
> > > piston, to go along with the forward/back piston on each leg. Right?
> >
> > Yes, I think this would take a third degree of freedom in the legs.
> >
> > >
> > > Thoughts?
> >
> > Plenty of them.
> >
> > > >
> > > > The walking mechanism will be completely wired into pneumatics. For walking
> > > > backwards, we can use pneumatic polarity reversors controlled by RCX/Motors.
> > > > The RCX may need to know that we are in the right state before it changes the
> > > > muxes, so that will take some kind of sensor.
> > >
> > >
> > > The control could be Motor->pneumatic switch for:
> > > C) Forward/Backward, Stop/Go, Same/Opposide (legs on both sides go same
> > > direction or opposite direction)
> >
> >
> > We have to use polarity reversors in pairs. If we reverse the leg sweep, by
> > swapping pressure hoses, then we need to swap the AND gate outputs (two
> > pneumatic switches) also.
> >
> > For walking backward, we reverse *all* the leg sweeps, and reverse the output of
> > the sweep complete AND chains. This requires one motor to flip these two
> > reversors.
> >
> > For turning, we reverse *half* the leg sweeps, and reverse the AND gate outputs
> > for the same pistons. We can hose things together so we can use only one
> > reversor for this. So turning requires one motor to flip these two reversors.
> >
> > Since you want to stop, you probably want to use a motor and a single pneumo
> > switch to the *off* position.
>
>
>
> Ok, so I'm thinking about motors to flip pneumo-switches. Each of these may be
> one or more switches, all activated at the same time.
>
> 1) Switch for Start/Stop (on/off) When this switch is "on" the scorpion will be
> walking. (legs going up/down + some other direction). This switch will activate
> all eight legs.
Yes.
>
> 2) Switch to select forward or backward for left legs.
No.
>
> 3) Switch to select forward or backward for right legs.
No
>
> 4) Switch to select forward/backward OR side-to-side for left legs.
>
> 5) Same as 4 for the right legs.
>
> 4&5 are optional, depending on if it CAN go sideways.
No.
Ignoring side walking for a moment.....
1) One motor to turn a single switch on/off. This allows or prevents walking.
2) One motor to flip 8 switches. This one controls walking vs. turning.
3) One motor to flip 8 switches. When walking, this one controls forward vs.
backwards. When tuning (in place), this one controls right vs. left turning.
I see no reason to turn off one side or the other, when we already have a
mechanism that will turn in place.
If we wanted to get fancier, we could add two more sets of controls so that one
set of four legs can walk forward/backard or turn right/left, while the other
set of legs can walk forward/backward or turn right/left.
With seperate leg group controls we could have it walk in a curved path by
having one leg group (four legs) turning, while the other leg group is walking.
You might want to study this:
http://www.kclague.net/Quad242/index.htm
In particular the movies of walking/turning to see what I have in mind.
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
>
> Ignoring side walking for a moment.....
>
> 1) One motor to turn a single switch on/off. This allows or prevents walking.
>
> 2) One motor to flip 8 switches. This one controls walking vs. turning.
>
> 3) One motor to flip 8 switches. When walking, this one controls forward vs.
> backwards. When tuning (in place), this one controls right vs. left turning.
That was my second choice.
Ok, three motors.
If it can be worked out, we COULD combine #1 and #3, right?
Leaving:
1) Function (walk/turn)
2) Direction (forward/backward/stop)
I assume both of these groups are controlled only by motors, and at no time are
they moved by pistons.
I still think walking sideways is cool, but optional.
> I see no reason to turn off one side or the other, when we already have a
> mechanism that will turn in place.
Right.
>
> You might want to study this:
>
> http://www.kclague.net/Quad242/index.htm
looks good. But slow.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> >
> > Ignoring side walking for a moment.....
> >
> > 1) One motor to turn a single switch on/off. This allows or prevents walking.
> >
> > 2) One motor to flip 8 switches. This one controls walking vs. turning.
> >
> > 3) One motor to flip 8 switches. When walking, this one controls forward vs.
> > backwards. When tuning (in place), this one controls right vs. left turning.
>
>
> That was my second choice.
>
> Ok, three motors.
>
> If it can be worked out, we COULD combine #1 and #3, right?
Nope, but we could just eliminate #1 by just turning of the compressors.
> Leaving:
> 1) Function (walk/turn)
> 2) Direction (forward/backward/stop)
>
> I assume both of these groups are controlled only by motors, and at no time are
> they moved by pistons.
>
>
> I still think walking sideways is cool, but optional.
It is still on the list, I was just leaving it out of the equation for a moment.
>
>
> > I see no reason to turn off one side or the other, when we already have a
> > mechanism that will turn in place.
>
> Right.
>
> >
> > You might want to study this:
> >
> > http://www.kclague.net/Quad242/index.htm
>
>
> looks good. But slow.
Yes, slow, more pressure, or more force. Shall I assume four pistons per leg?
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Eric Sophie wrote:
> In lugnet.technic, Benton Jackson wrote:
> > In lugnet.technic, Eric Sophie wrote:
> > > In lugnet.technic, Benton Jackson wrote: -snip-
> > Are you mistaking me for the other Ben Jackson, who might have actually been to
> > LugNet before?
>
> ahh, Tim will buy you one too.
That would probably convince me, if I knew who Tim was, and if I ever went to
BrickFest. I'm not that dedicated an AFOL I guess.
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> >
> > Ignoring side walking for a moment.....
> > You might want to study this:
> >
> > http://www.kclague.net/Quad242/index.htm
>
>
> looks good. But slow.
This is a reality given large LEGO pneumatic pistons, but I knew this when I
started this project by inviting Eric to join me.
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > In lugnet.technic, Kevin L. Clague wrote:
> > > You might want to study this:
> > >
> > > http://www.kclague.net/Quad242/index.htm
> >
> >
> > looks good. But slow.
>
> Yes, slow, more pressure, or more force. Shall I assume four pistons per leg?
That's up to you. I'd like to see it move as fast as possible, but I think even
if we make it as small as possible, it's going to weigh at least 10 pounds. 3
or 4 pounds will be the RCXs, battey box, Spybot, and motors.
So, what will it take to move that?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > In lugnet.technic, Kevin L. Clague wrote:
> > > > You might want to study this:
> > > >
> > > > http://www.kclague.net/Quad242/index.htm
> > >
> > >
> > > looks good. But slow.
> >
> > Yes, slow, more pressure, or more force. Shall I assume four pistons per leg?
>
>
>
> That's up to you. I'd like to see it move as fast as possible, but I think even
> if we make it as small as possible, it's going to weigh at least 10 pounds. 3
> or 4 pounds will be the RCXs, battey box, Spybot, and motors.
>
> So, what will it take to move that?
Hmmmm.... interesting question.
I don't know much pneumatic theory, but.....
Pressure is in pounds per square inch. So if we know the area of the face of a
piston, and we know the pressure we can determine how many pounds a single
piston can lift.
Force = Pressure * Area
A pneumatic piston face is about a half an inch in diameter. The area then is
pi times diameter, or about 1.6. Four pistons then have the combined area of
6.4.
So the minimal pressure needed is 10 pounds / 6.4 sq inch = 1.5 pounds/square
inch. Unless my math is wrong (and it could be) four large pistons should be
easily able to lift 10 pounds within the upper limit of 32 PSI inherent in LEGO
pneumatics.
One nice thing is that the vertical pistons do not have to *lift* the body,
because the body always stays the same distance off the ground. So the speed at
which legs the legs lift and drop is simply related to the speed at which the
pistons can expand/contract, combined with the resistance provided by the two
switches the piston flips. If the resistance provided by the switches causes a
big slow down in piston expansion, then doubling the pistons will reduce the
expansion time. If the switches don't slow things down much, adding a second
piston won't help much. I'll have to test this.
Pistons expansion/contraction rate depends on inherent resistance and pressure.
The way I control the pneumatics, when multiple pistons are
expanding/contracting at the same time I make sure that all have completed their
change before the next change starts. This means that the slowest pistons
control the rate. This is intentional to make sure that they all stay
coordinated.
I'll try to make sure my leg/hip design does not slow piston
expansion/contraction.
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> Pressure is in pounds per square inch. So if we know the area of the face of a
> piston, and we know the pressure we can determine how many pounds a single
> piston can lift.
>
> Force = Pressure * Area
>
> A pneumatic piston face is about a half an inch in diameter. The area then is
> pi times diameter, or about 1.6. Four pistons then have the combined area of
> 6.4.
Mark Riley was nice enought to point out the errors in my math. The area of the
piston care is pi * r^2 = 0.2
Four times that is 0.8.
So the minimal pressure needed is 10 pounds / 0.8 square inches or, 12.5 pounds
per square inch. Still well below the 32 PSI upper limit for LEGO pneumatics.
The next question is, how easy is it to get to 13 PSI with LEGO based
compressors.
If we double the number of pistons, we'd only need to get to 6 PSI, but will
this really make much difference on piston transition time.
The pistons that do the most work (force through a distance) are the horizontal
pistons. They actually make the distance term non-zero.
Back to compressors for a second...... There has always been this debate about
which pumps are better, small pumps or large pumps. Small pumps provide less
piston face are than larger pumps, and can drive the pressure in a closed system
to max pressure quicker than the large pumps (even with spring removed)... but a
pneumatic sequencer (motor for example) is not a closed system. It is always
releasing pressure when the pistons are switching direction. A second test that
measures the RPM of a pneumatic engine (motor?) using a small pump vs. a large
pump showed that the large pump did better.
One of the issued with large pumps is that they are large... large enough to
affect the minimum scale of a compressor, and therefore the creation being
made......
I'd like us to make Scorpion completely out of LEGO parts. I'm not a full time
purist, but typically this is my goal so that others can make copies. I guess
also that I'd like to make it out of currently available parts. This rules out
the use of the original single ported pistons. Unfortunatly this also rules out
the use of:
http://www.bricklink.com/catalogItemPic.asp?P=4692
Using two of these and a double ported piston would give us a double acting
pump, instead of the single action pumps lego makes. Presumably this would
reduce the inherent amount of friction in the compressor.
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
>
>
> The next question is, how easy is it to get to 13 PSI with LEGO based
> compressors. • ...
>
> If we double the number of pistons, we'd only need to get to 6 PSI, but will
> this really make much difference on piston transition time.
With twice the pistons, the pressure is not as great, but the volume of air
would be doubled, right?
Using logic, not physics, I think once the pressure is built up (to 13 psi) the
compressor may have to do less work to keep it there, than to maintain twice the
volume, at 6 psi.
Or, it's the same amount of work. But I doubt it will be less work on the
compressor to maintain the 6 psi with twice the pistons.
>
> I'd like us to make Scorpion completely out of LEGO parts. I'm not a full time
> purist, but typically this is my goal so that others can make copies. I guess
> also that I'd like to make it out of currently available parts. This rules out
> the use of the original single ported pistons. Unfortunatly this also rules out
> the use of:
I agree with trying to use LEGO parts (as a goal), but I'm not sure about
limiting to currently available parts. I mean, do you REALLY think anyone would
try to replicate this? :)
And, if they do, the parts are available on Bricklink, right?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
<snip>
> I agree with trying to use LEGO parts (as a goal), but I'm not sure about
> limiting to currently available parts. I mean, do you REALLY think anyone would
> try to replicate this? :)
>
> And, if they do, the parts are available on Bricklink, right?
Well, I know of situations where LEGO has asked some people to recreate their
MOCs for them. You know you have arrived if that happens.
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Kevin L. Clague wrote:
>
> > Pressure is in pounds per square inch. So if we know the area of the face of a
> > piston, and we know the pressure we can determine how many pounds a single
> > piston can lift.
> >
> > Force = Pressure * Area
> >
> > A pneumatic piston face is about a half an inch in diameter. The area then is
> > pi times diameter, or about 1.6. Four pistons then have the combined area of
> > 6.4.
>
>
> Mark Riley was nice enought to point out the errors in my math. The area of the
> piston care is pi * r^2 = 0.2
>
> Four times that is 0.8.
>
> So the minimal pressure needed is 10 pounds / 0.8 square inches or, 12.5 pounds
> per square inch. Still well below the 32 PSI upper limit for LEGO pneumatics.
>
> The next question is, how easy is it to get to 13 PSI with LEGO based
> compressors.
>
> If we double the number of pistons, we'd only need to get to 6 PSI, but will
> this really make much difference on piston transition time.
>
> The pistons that do the most work (force through a distance) are the horizontal
> pistons. They actually make the distance term non-zero.
>
> Back to compressors for a second...... There has always been this debate about
> which pumps are better, small pumps or large pumps. Small pumps provide less
> piston face are than larger pumps, and can drive the pressure in a closed system
> to max pressure quicker than the large pumps (even with spring removed)... but a
> pneumatic sequencer (motor for example) is not a closed system. It is always
> releasing pressure when the pistons are switching direction. A second test that
> measures the RPM of a pneumatic engine (motor?) using a small pump vs. a large
> pump showed that the large pump did better.
>
> One of the issued with large pumps is that they are large... large enough to
> affect the minimum scale of a compressor, and therefore the creation being
> made......
>
> I'd like us to make Scorpion completely out of LEGO parts. I'm not a full time
> purist, but typically this is my goal so that others can make copies. I guess
> also that I'd like to make it out of currently available parts. This rules out
> the use of the original single ported pistons. Unfortunatly this also rules out
> the use of:
>
> http://www.bricklink.com/catalogItemPic.asp?P=4692
>
> Using two of these and a double ported piston would give us a double acting
> pump, instead of the single action pumps lego makes. Presumably this would
> reduce the inherent amount of friction in the compressor.
>
> Kevin
A few things spring to mind about this:
1. I usually run robots from a car tyre air compressor at 25PSI. This gives
good performance with 1-2 large cylinders moving at a time. Using 6 large pump
cylinders as the air source, performance is half as good, probably around 13psi,
which is what you're aiming for. I suggest that 6 large pump cylinders with
springs removed might provide enough air.
2. However, the model will have to support its own weight without a gravity well
(due to losing pressure in supporting leg cylinders when other cylinders use the
air in the pipes). You'll have to either increase the source pressure so that
the minimum pressure stays above about 10psi (gaining a bit of leeway for the
friction helping to prevent movement) or put more isolation logic in place on
each leg supporting cylinder.
3. The model is going to walk, so the force on each leg is potentially a dynamic
one, if it walks fast enough. When you jump repeatedly, the force on your legs
is not equal to your weight, but significantly more - at least double, depending
on how fast you jump. Therefore the load on the legs will be at least twice the
weight of the robot divided by the number of legs that stay on the ground when a
step is taken.
Since you're going for a completely current Lego parts solution, that rules out
my 2-mini-compressors-driven-by-a-12-volt-train-motor solution. The most
powerful motors you've got are the 8475 RC buggy ones with the red plugs on.
They'll use 7.2W of electrical power but put out 2.38W each.
See http://www.philohome.com/motors/motorcomp.htm
I suggest you make a trial compressor engine using a pair of 8475 RC buggy
motors geared down and four pump cylinders with springs removed, with cranks
offset at each 90 degree angle. Offsetting the pumps by 90 degrees should make
the load on the motors smoother. The engine will need to be very strong due to
the torque required. I suggest two shafts geared together, supported by double
width beams throughout. Use 40 tooth cogs to make the cranks as they're more
sturdy than other crank parts and will allow you to gear the two shafts together
easily.
Provided the non-return valves in the pump cylinders are good, you only need to
use T-junctions to connect them together. I suggest an initial shaft speed of
1rev/second. That said, using 6 pump cylinders in my hands I pump them at a
rate of 1 pump/second, which is 6 cylinders of air/second. The engine will do
just 4, but see how well it works with the amount of motor power. You might
need several engines!
Just a thought - what's the mechanical load required to compress a large pump
cylinder at 25psi? That's the load the motors have to cope with. You could
measure the volume of a pump cylinder by using a graded syringe to collect the
air.
Overall, the load on the compressor engine will be greater than the load on the
legs, since the system can't be 100% efficient. My car tyre air compressor
needs 6 Amps at 12 Volts, so 72 Watts - that's ten RC buggy motors or 5 engines!
This is getting expensive! Surely a £20 compressor and 12V 6Ah lead acid
battery are fair game? I abandoned Lego controllers for my railway in favour of
a large PSU as soon as I got beyond 2 train motors!
Mark
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> looks good. But slow.
Thinking about this more.......
One of the reasons that makes this slow is the inherent piston friction which
combined with pressure defines the rate of expansion and contraction. So the
rate is not very negotiable.
But using
rate * time = distance
and turning it around for time we get
time = distance/rate
Since we cannot really control rate well, we must reduce distance to decrease
time.
In the past, to avoid stresses to structure, I've always made the piston/switch
linkage work so that the piston never pushes or pulls the switch to its limits.
i.e. The throw due to piston expansion contraction never exceeds the max throw
of the switch handle.
But...... if I use as little piston expansion/contract as I can to flip the
switch, I end up with the shortest time for piston expansion/contraction. Cut
the piston throw in half, and double your walk rate.
I'll start on new designs using this concept. Because I'm building something
that wants to tear itself apart, I'm going to use studded beams for extra
strength.
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
<snip>
> But...... if I use as little piston expansion/contract as I can to flip the
> switch, I end up with the shortest time for piston expansion/contraction. Cut
> the piston throw in half, and double your walk rate.
>
> I'll start on new designs using this concept. Because I'm building something
> that wants to tear itself apart, I'm going to use studded beams for extra
> strength.
So using #1 axle connectors as the linkage, we get a reduced piston expansion,
down to 86% of maximum.
Using technic axle joiner perpendicular, we can reduce the range to 70% of
maximum. I can't think of a way to make this any smaller.
Reducing the expansion range of the piston also reduces the leverage the piston
has on the switch, making it harder to throw, increasing pressure requirements.
Ignoring the pressure issue, a simple reduction of 30% of throw, decreases time
by 30%, so the increase in step rate is less than doubled.
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> <snip>
> > But...... if I use as little piston expansion/contract as I can to flip the
> > switch, I end up with the shortest time for piston expansion/contraction. Cut
> > the piston throw in half, and double your walk rate.
> >
> > I'll start on new designs using this concept. Because I'm building something
> > that wants to tear itself apart, I'm going to use studded beams for extra
> > strength.
Yes, I'd rather see a solid, strong, slow robot over a fast robot that succeeds
in tearing itself apart. :)
> Ignoring the pressure issue, a simple reduction of 30% of throw, decreases time
> by 30%, so the increase in step rate is less than doubled.
That's the throw of the switches, right? Not the throw of the cylinders on the
legs. So the legs will "move" at the same speed, but the time between the start
of one move, and the start of another move could be less.
In any case, yes, it does sound good.
I've been playing around with the Spybot & it's controller. I can make the
Spybot follow the remote, and I can process remote commands, which allows me to
"control" the Spybot, using it's nice little remote control.
I just had another idea. In the movie RoboCop, there was a big mech-looking
robot, which ended up going out of control. When we first saw that robot, they
powered it up, and the whole thing raised up a bit, and came to life.
That would be another cool trick for SSClagorion to do. Actually, I think by
letting all the pressure out of the system, it may sit down on the ground.
Then, when it's repressurized, it will "power-up". Is that right?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > <snip>
> > > But...... if I use as little piston expansion/contract as I can to flip the
> > > switch, I end up with the shortest time for piston expansion/contraction. Cut
> > > the piston throw in half, and double your walk rate.
> > >
> > > I'll start on new designs using this concept. Because I'm building something
> > > that wants to tear itself apart, I'm going to use studded beams for extra
> > > strength.
>
>
> Yes, I'd rather see a solid, strong, slow robot over a fast robot that succeeds
> in tearing itself apart. :)
>
>
> > Ignoring the pressure issue, a simple reduction of 30% of throw, decreases time
> > by 30%, so the increase in step rate is less than doubled.
>
> That's the throw of the switches, right? Not the throw of the cylinders on the
> legs. So the legs will "move" at the same speed, but the time between the start
> of one move, and the start of another move could be less.
>
> In any case, yes, it does sound good.
>
>
>
> I've been playing around with the Spybot & it's controller. I can make the
> Spybot follow the remote, and I can process remote commands, which allows me to
> "control" the Spybot, using it's nice little remote control.
>
>
>
> I just had another idea. In the movie RoboCop, there was a big mech-looking
> robot, which ended up going out of control. When we first saw that robot, they
> powered it up, and the whole thing raised up a bit, and came to life.
>
> That would be another cool trick for SSClagorion to do. Actually, I think by
> letting all the pressure out of the system, it may sit down on the ground.
> Then, when it's repressurized, it will "power-up". Is that right?
Well Steve, it might and it might not. Given 18 bits of state (two bits/pistons
per leg plus two extra), we could have a total of 256K possible combinations of
leg positions. We only want about about 10 of them. If we start out the
pistons in one of the desired state combinations it will progress through the
magic sequence and get back to the initial state.
If we start it out in an unexpected state, it is hard to know what it will do,
but it probably won't get itself into a desired state. It will probably just
hang.
It would take a *lot* more logic (switches and maybe pistons) to make it self
correcting and get itself back into a good state.
So the short answer is *probably* not.
>
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Mark Bellis wrote:
> A few things spring to mind about this:
>
> 3. The model is going to walk, so the force on each leg is potentially a dynamic
> one, if it walks fast enough. When you jump repeatedly, the force on your legs
> is not equal to your weight, but significantly more - at least double,
That makes sense, but you may want to take a look at the video of Kevin's
walkers:
http://www.kclague.net/Quad242/q242-forward.mpg
As you see, at the end of each stride, all four legs are at the bottom of their
travel. So while it is true that half the legs will have to support all the
weight, I don't think the force on the legs will ever be above the weight of the
walker.
It's pretty cool the way it walks. I'm not exactly sure what it's doing, it
must be some kind of black magic...
> I suggest you make a trial compressor engine using a pair of 8475 RC buggy
> motors geared down and four pump cylinders with springs removed, with cranks
> offset at each 90 degree angle. Offsetting the pumps by 90 degrees should make
> the load on the motors smoother. The engine will need to be very strong due to
> the torque required. I suggest two shafts geared together, supported by double
> width beams throughout. Use 40 tooth cogs to make the cranks as they're more
> sturdy than other crank parts and will allow you to gear the two shafts together
> easily.
Yes, Kevin was just talking about using beams for strength throughout the model,
instead of the lighter liftarms he often uses.
I'm hoping he has some experience with compressors. If not, I'll have to throw
together something using the buggy motors. Thanks for the idea.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Mark Bellis wrote:
> > A few things spring to mind about this:
> >
> > 3. The model is going to walk, so the force on each leg is potentially a dynamic
> > one, if it walks fast enough. When you jump repeatedly, the force on your legs
> > is not equal to your weight, but significantly more - at least double,
>
>
> That makes sense, but you may want to take a look at the video of Kevin's
> walkers:
>
> http://www.kclague.net/Quad242/q242-forward.mpg
>
> As you see, at the end of each stride, all four legs are at the bottom of their
> travel. So while it is true that half the legs will have to support all the
> weight, I don't think the force on the legs will ever be above the weight of the
> walker.
>
> It's pretty cool the way it walks. I'm not exactly sure what it's doing, it
> must be some kind of black magic...
Funny! My presentation at Brickest (classic?) was named Pneumatic Magic.
When I started on quad242 I thought the cool thing was going to be the ability
to make it modify its behavior to go backwards and turn, but the hardest part
was the "no gravity well" concept you point out above Steve.
> > I suggest you make a trial compressor engine using a pair of 8475 RC buggy
> > motors geared down and four pump cylinders with springs removed, with cranks
> > offset at each 90 degree angle. Offsetting the pumps by 90 degrees should make
> > the load on the motors smoother. The engine will need to be very strong due to
> > the torque required. I suggest two shafts geared together, supported by double
> > width beams throughout. Use 40 tooth cogs to make the cranks as they're more
> > sturdy than other crank parts and will allow you to gear the two shafts together
> > easily.
>
>
>
> Yes, Kevin was just talking about using beams for strength throughout the model,
> instead of the lighter liftarms he often uses.
>
> I'm hoping he has some experience with compressors. If not, I'll have to throw
> together something using the buggy motors. Thanks for the idea.
I have some experience with compressors.
I've worked out a perfect geometry for large pumps so that they can be pumped by
a motor and use the entire piston head travel range. This was one of those holy
grail kinds of things.
There has been a long standing debate about which are better small pumps vs.
large pumps. C.S. Soh is set on small pumps, and he actively scorns those who
use them (well at least me anyway.) I've seen good arguments for large pumps
also.
If we go with four pistons per leg (more volume with less pressure), I for sure
think that large pumps are the way to go. It is hard to acheive high pressure
with LEGO pneumatics, but it is not nearly as hard to get the volume up. I'm
still pondering the 2 vs. 4 scenario.
I've got a pretty strong 2 piston hip/leg assembly going. The hip pivot is very
sturdy (one of the weak points on hex363, but the linkage from that to the leg
piston is pretty weak.
Here is are some pics.
http://www.kclague.net/SSCorpion/P4130047.JPG
http://www.kclague.net/SSCorpion/P4130048.JPG
http://www.kclague.net/SSCorpion/P4130045.JPG
http://www.kclague.net/SSCorpion/P4130046.JPG
http://www.kclague.net/SSCorpion/P4130049.JPG
The force of the sweep is effectively transfered through the hip joint, but the
force is then transferred to the leg through the leg piston and the leg
switches. Don't think the 3x5 quarter oval and switch handles are strong enough
to transfer the sweep force.
Kevin
>
> Steve
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > In lugnet.technic, Mark Bellis wrote:
> > > A few things spring to mind about this:
> > >
> > > 3. The model is going to walk, so the force on each leg is potentially a dynamic
> > > one, if it walks fast enough. When you jump repeatedly, the force on your legs
> > > is not equal to your weight, but significantly more - at least double,
> >
> >
> > That makes sense, but you may want to take a look at the video of Kevin's
> > walkers:
> >
> > http://www.kclague.net/Quad242/q242-forward.mpg
> >
> > As you see, at the end of each stride, all four legs are at the bottom of their
> > travel. So while it is true that half the legs will have to support all the
> > weight, I don't think the force on the legs will ever be above the weight of the
> > walker.
> >
> > It's pretty cool the way it walks. I'm not exactly sure what it's doing, it
> > must be some kind of black magic...
>
> Funny! My presentation at Brickest (classic?) was named Pneumatic Magic.
>
> When I started on quad242 I thought the cool thing was going to be the ability
> to make it modify its behavior to go backwards and turn, but the hardest part
> was the "no gravity well" concept you point out above Steve.
>
> > > I suggest you make a trial compressor engine using a pair of 8475 RC buggy
> > > motors geared down and four pump cylinders with springs removed, with cranks
> > > offset at each 90 degree angle. Offsetting the pumps by 90 degrees should make
> > > the load on the motors smoother. The engine will need to be very strong due to
> > > the torque required. I suggest two shafts geared together, supported by double
> > > width beams throughout. Use 40 tooth cogs to make the cranks as they're more
> > > sturdy than other crank parts and will allow you to gear the two shafts together
> > > easily.
> >
> >
> >
> > Yes, Kevin was just talking about using beams for strength throughout the model,
> > instead of the lighter liftarms he often uses.
> >
> > I'm hoping he has some experience with compressors. If not, I'll have to throw
> > together something using the buggy motors. Thanks for the idea.
>
> I have some experience with compressors.
>
> I've worked out a perfect geometry for large pumps so that they can be pumped by
> a motor and use the entire piston head travel range. This was one of those holy
> grail kinds of things.
>
> There has been a long standing debate about which are better small pumps vs.
> large pumps. C.S. Soh is set on small pumps, and he actively scorns those who
> use them (well at least me anyway.) I've seen good arguments for large pumps
> also.
>
> If we go with four pistons per leg (more volume with less pressure), I for sure
> think that large pumps are the way to go. It is hard to acheive high pressure
> with LEGO pneumatics, but it is not nearly as hard to get the volume up. I'm
> still pondering the 2 vs. 4 scenario.
>
> I've got a pretty strong 2 piston hip/leg assembly going. The hip pivot is very
> sturdy (one of the weak points on hex363, but the linkage from that to the leg
> piston is pretty weak.
>
> Here is are some pics.
>
> http://www.kclague.net/SSCorpion/P4130047.JPG
> http://www.kclague.net/SSCorpion/P4130048.JPG
>
> http://www.kclague.net/SSCorpion/P4130045.JPG
> http://www.kclague.net/SSCorpion/P4130046.JPG
> http://www.kclague.net/SSCorpion/P4130049.JPG
>
> The force of the sweep is effectively transfered through the hip joint, but the
> force is then transferred to the leg through the leg piston and the leg
> switches. Don't think the 3x5 quarter oval and switch handles are strong enough
> to transfer the sweep force.
>
> Kevin
I can see the good sense in putting the hip switches on their own sets of
levers, as your pics show. That way, the force required for them and for the
legs is independent, so there's no extra load on the switches. I am used to
mounting the switches all in a line on the chassis, especially if there are more
than two.
Have you seen this compressor and regulator (near the bottom of the page)?:
http://www.genuinemodels.com/jcb.htm
This compressor has the pumps at 90 degrees offset by using a V-engine
orientation. It uses 71427 or 43362 motors, which are widely available, and a
good gear ratio has already been achieved. You could crib the design if you use
small compressors or uprate it for large ones, with lower gearing.
You might also see some nice techniques used in the control panel.
(Wow! I wrote less than a page on pneumatics for once!)
Mark
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > In lugnet.technic, Kevin L. Clague wrote:
> > >
> > > Ignoring side walking for a moment.....
> > >
> > > 1) One motor to turn a single switch on/off. This allows or prevents walking.
> > >
> > > 2) One motor to flip 8 switches. This one controls walking vs. turning.
> > >
> > > 3) One motor to flip 8 switches. When walking, this one controls forward vs.
> > > backwards. When tuning (in place), this one controls right vs. left turning.
> >
> >
> > That was my second choice.
> >
> > Ok, three motors.
> >
> >
> > I still think walking sideways is cool, but optional.
>
> It is still on the list, I was just leaving it out of the equation for a moment.
Ok, so what's required for walking sideways? One switch (= single set of
switches, all activated at once) to switch from forward/back/turn to
side-to-side? I'm sure that comes to about 16 reverser switches, or something.
:)
I just happened to think, the legs should be in a known position when switching
from forward/back to side-to-side.
Do the legs need to be in a know position when switching from forward to
backward? Or walking to turning? If so, we need sensors for that...
Here's my current bill of materials:
1 Spybot, for IR direction control. This is mostly used as a sensor.
2 RCXs. I really haven't counted I/O, but two sounds like enough.
1 Battery box for compressor(s)
All those should be installed in the body of the beast. The Spybot/RCXs will
need to communicate via IR. (there are other options, but we'll start there)
The Spybot has two motors, but very limited sensor feedback. It would be good
to use the Spybot motors to activate something like claws (open & close), where
it won't be critical if they work exactly right. We can use the Spybot remote
to send it commands, like "follow me" or "attack anything in front of you"
All the outputs from 1 RCX control movement. 1=Activate, 2=function
(turn/straight), 3=forward/backward
Two inputs would be used for sensing the edge-of-table. That's one input for
each front leg. Each leg would have one sensor to detect when the leg is in
it's "down" position, and one sensor to detect the ground. The two sensors in
each leg can be connected together, to use one input. This could be two touch
sensors, or a touch+light sensor combination.
Two inputs can be used for the arm "feelers". Maybe several touch sensors per
arm, or one rotation sensor for each.
I'm not sure the RCXs really need to "know" when the SSClagorpion is actually
walking. It can assume if it runs the motors, then the switches are set, and it
is going.
Beyond that, there is one motor to set the switch for curling the tail.
That leaves 2 motor outputs, and 2 sensor inputs, for further expansion. Plus a
touch sensor & light sensor built into the Spybot. The outputs from the second
RCX can be connected to a gearbox, like Eric was suggesting, if we want to
control more things, but I think that may just make things more complex than
they need to be.
It would be nice to mount the Spybot in the end of the tail, but it needs to
communicate with the RCXs via IR, so I don't think that will work well, unless
we connect a fiber-optic line. No, I don't think that will work, either.
So, SSClagorpion will walk forward & backward, turn right & left, open & close
two claws, and move it's tail. It can stay on a table/stage, detect when it
bumps into something (going forward), follow the "controller" or be driven
around like an RC car.
Is that enough cool functionality?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > In lugnet.technic, Kevin L. Clague wrote:
> > > >
> > > > Ignoring side walking for a moment.....
> > > >
> > > > 1) One motor to turn a single switch on/off. This allows or prevents walking.
> > > >
> > > > 2) One motor to flip 8 switches. This one controls walking vs. turning.
> > > >
> > > > 3) One motor to flip 8 switches. When walking, this one controls forward vs.
> > > > backwards. When tuning (in place), this one controls right vs. left turning.
> > >
> > >
> > > That was my second choice.
> > >
> > > Ok, three motors.
> > >
> > >
> > > I still think walking sideways is cool, but optional.
> >
> > It is still on the list, I was just leaving it out of the equation for a moment.
>
>
> Ok, so what's required for walking sideways? One switch (= single set of
> switches, all activated at once) to switch from forward/back/turn to
> side-to-side? I'm sure that comes to about 16 reverser switches, or something.
> :)
It is my belief that sideways walking will require a third degree of
articulation in the leg/hip structure. It will cause the legs to swing towards
and away from the body. By replacing the forward/backward part of the leg sweep
with the side to side leg sweep we should be able to walk.
I'd guess 16 also..... We'd need 8 to swap the side to side motions for the
forward back sweep, this gives us one direction of side walking. We'd need
another 8 to swap the direction sweep so that we can walk toward the other side.
Looks like the sideways muxes will need a motor. We can probably use the
forward/backward motor to also control the right/left muxes as well.
>
> I just happened to think, the legs should be in a known position when switching
> from forward/back to side-to-side.
>
> Do the legs need to be in a know position when switching from forward to
> backward? Or walking to turning? If so, we need sensors for that...
Yes. Changing the forward/backward/turning muxes needs to be done when the feet
are in the air. Observing any one foot is probably sufficient.
The above costs us four motors, and one sensor.
>
>
> Here's my current bill of materials:
> 1 Spybot, for IR direction control. This is mostly used as a sensor.
> 2 RCXs. I really haven't counted I/O, but two sounds like enough.
> 1 Battery box for compressor(s)
>
> All those should be installed in the body of the beast. The Spybot/RCXs will
> need to communicate via IR. (there are other options, but we'll start there)
>
> The Spybot has two motors, but very limited sensor feedback. It would be good
> to use the Spybot motors to activate something like claws (open & close), where
> it won't be critical if they work exactly right. We can use the Spybot remote
> to send it commands, like "follow me" or "attack anything in front of you"
>
> All the outputs from 1 RCX control movement. 1=Activate, 2=function
> (turn/straight), 3=forward/backward
Until you try side to side walking.
>
> Two inputs would be used for sensing the edge-of-table. That's one input for
> each front leg. Each leg would have one sensor to detect when the leg is in
> it's "down" position, and one sensor to detect the ground. The two sensors in
> each leg can be connected together, to use one input. This could be two touch
> sensors, or a touch+light sensor combination.
>
> Two inputs can be used for the arm "feelers". Maybe several touch sensors per
> arm, or one rotation sensor for each.
>
>
> I'm not sure the RCXs really need to "know" when the SSClagorpion is actually
> walking. It can assume if it runs the motors, then the switches are set, and it
> is going.
Yes, I think it can assume that if it has set the switches correctly it is
walking.
>
> Beyond that, there is one motor to set the switch for curling the tail.
>
> That leaves 2 motor outputs, and 2 sensor inputs, for further expansion. Plus a
> touch sensor & light sensor built into the Spybot. The outputs from the second
> RCX can be connected to a gearbox, like Eric was suggesting, if we want to
> control more things, but I think that may just make things more complex than
> they need to be.
>
>
> It would be nice to mount the Spybot in the end of the tail, but it needs to
> communicate with the RCXs via IR, so I don't think that will work well, unless
> we connect a fiber-optic line. No, I don't think that will work, either.
>
>
> So, SSClagorpion will walk forward & backward, turn right & left, open & close
> two claws, and move it's tail. It can stay on a table/stage, detect when it
> bumps into something (going forward), follow the "controller" or be driven
> around like an RC car.
>
> Is that enough cool functionality?
It works for me!
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > Ok, so what's required for walking sideways?
> > I just happened to think, the legs should be in a known position when switching
> > from forward/back to side-to-side.
> >
> > Do the legs need to be in a know position when switching from forward to
> > backward? Or walking to turning? If so, we need sensors for that...
>
> Yes. Changing the forward/backward/turning muxes needs to be done when the feet
> are in the air. Observing any one foot is probably sufficient.
>
> The above costs us four motors, and one sensor.
Ok, we [plan to] have a sensor on the front foot, and can detect when that is
down. So, no more sensors are required.
So you said, the forward/backward must be changed while the feet are in the air.
We've been thinking of having a touch sensor on the front feet, so it can detect
when it goes off an edge (like a table). However, it would not know the surface
is not there, until it trys to put the foot down. So, all the feet would be
down, and it needs to backup. Can it do that? Or will it need to keep walking
a bit?
Overall, I THINK that just adds one motor to my previous list:
1) Activate (move!)
2) Direction (forward/back, turn right/left, slide right/left)
3a) Function 1 (walk or turn)
3b) Function 2 (function 1 or slide)
> > So, SSClagorpion will walk forward & backward, turn right & left,
step right & left
> > open & close
> > two claws, and move it's tail. It can stay on a table/stage, detect when it
> > bumps into something (going forward), follow the "controller" or be driven
> > around like an RC car.
> >
> > Is that enough cool functionality?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > Ok, so what's required for walking sideways?
> > > I just happened to think, the legs should be in a known position when switching
> > > from forward/back to side-to-side.
> > >
> > > Do the legs need to be in a know position when switching from forward to
> > > backward? Or walking to turning? If so, we need sensors for that...
> >
> > Yes. Changing the forward/backward/turning muxes needs to be done when the feet
> > are in the air. Observing any one foot is probably sufficient.
> >
> > The above costs us four motors, and one sensor.
>
> Ok, we [plan to] have a sensor on the front foot, and can detect when that is
> down. So, no more sensors are required.
Actualy we need to sense when it is *up*.
>
> So you said, the forward/backward must be changed while the feet are in the air.
> We've been thinking of having a touch sensor on the front feet, so it can detect
> when it goes off an edge (like a table). However, it would not know the surface
> is not there, until it trys to put the foot down. So, all the feet would be
> down, and it needs to backup. Can it do that? Or will it need to keep walking
> a bit?
As soon as you figure out that a foot is up, you can reverse directions. I'll
have to study the state sequence, because there are two non-weight bearing
pistons that control the feet, so it might be sufficient to see that they are
telling the feet to lift. You might need two touch sensors wire-ORed together
(one for each leg group).
>
> Overall, I THINK that just adds one motor to my previous list:
> 1) Activate (move!)
> 2) Direction (forward/back, turn right/left, slide right/left)
> 3a) Function 1 (walk or turn)
> 3b) Function 2 (function 1 or slide)
I think so too.
>
> > > So, SSClagorpion will walk forward & backward, turn right & left,
>
> step right & left
>
> > > open & close
> > > two claws, and move it's tail. It can stay on a table/stage, detect when it
> > > bumps into something (going forward), follow the "controller" or be driven
> > > around like an RC car.
> > >
> > > Is that enough cool functionality?
>
>
>
> Steve
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> >
> > Ok, we [plan to] have a sensor on the front foot, and can detect when that is
> > down. So, no more sensors are required.
>
> Actualy we need to sense when it is *up*.
Is *up* the same as "not down"? I was thinking it was, but that may not be the
case.
> >
> > So you said, the forward/backward must be changed while the feet are in the air.
> > We've been thinking of having a touch sensor on the front feet, so it can detect
> > when it goes off an edge (like a table). However, it would not know the surface
> > is not there, until it trys to put the foot down. So, all the feet would be
> > down, and it needs to backup. Can it do that? Or will it need to keep walking
> > a bit?
>
> As soon as you figure out that a foot is up, you can reverse directions. I'll
> have to study the state sequence, because there are two non-weight bearing
> pistons that control the feet, so it might be sufficient to see that they are
> telling the feet to lift. You might need two touch sensors wire-ORed together
> (one for each leg group).
This sounds like we can NOT do the table edge detection stuff, using it's feet.
Imagine you're crawling around on your hands and knees, with your eyes closed.
When you get to the top of a stairway, you put your hand down, and it doesn't
touch the ground.
If I understand you correctly (which doesn't always happen via e-mail) you must
continue to crawl forward, until your other hand and knee are off the ground.
That will be a problem.
It's fine if it can't turn, or switch to moving sideways while all feet are on
the ground, but it really needs to be able to reverse direction.
I suspect the front legs can be a small amount longer, so it will detect the
lack of ground, before the weight is really shifted to the new legs.
Will that work?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > >
> > > Ok, we [plan to] have a sensor on the front foot, and can detect when that is
> > > down. So, no more sensors are required.
> >
> > Actualy we need to sense when it is *up*.
>
>
> Is *up* the same as "not down"? I was thinking it was, but that may not be the
> case.
I think that you are right, that they are the same thing.
>
>
> > >
> > > So you said, the forward/backward must be changed while the feet are in the air.
> > > We've been thinking of having a touch sensor on the front feet, so it can detect
> > > when it goes off an edge (like a table). However, it would not know the surface
> > > is not there, until it trys to put the foot down. So, all the feet would be
> > > down, and it needs to backup. Can it do that? Or will it need to keep walking
> > > a bit?
> >
> > As soon as you figure out that a foot is up, you can reverse directions. I'll
> > have to study the state sequence, because there are two non-weight bearing
> > pistons that control the feet, so it might be sufficient to see that they are
> > telling the feet to lift. You might need two touch sensors wire-ORed together
> > (one for each leg group).
>
>
> This sounds like we can NOT do the table edge detection stuff, using it's feet.
Hmmm.... We would need to know two things to determine we'd hit the edge of the
table.
1. The foot is down as far as it will go.
2. The down foot is not touching anything.
We would need one of these sensor pairs for each front foot.
When the foot is down and not touching anything, we wait for the foot start to
rise, and flip the muxes to backwards. At the time the legs get to the top,
they will go backwards, so I don't think we'll walk off the table.
>
> Imagine you're crawling around on your hands and knees, with your eyes closed.
> When you get to the top of a stairway, you put your hand down, and it doesn't
> touch the ground.
>
> If I understand you correctly (which doesn't always happen via e-mail) you must
> continue to crawl forward, until your other hand and knee are off the ground.
>
> That will be a problem.
>
> It's fine if it can't turn, or switch to moving sideways while all feet are on
> the ground, but it really needs to be able to reverse direction.
>
> I suspect the front legs can be a small amount longer, so it will detect the
> lack of ground, before the weight is really shifted to the new legs.
Remember, 7 of the 8 legs will be down at the time we realize that we're at the
edge. Given that pneumatics are slow, we should have plenty of time to slam it
into reverse.
>
> Will that work?
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > This sounds like we can NOT do the table edge detection stuff, using it's feet.
>
> Hmmm.... We would need to know two things to determine we'd hit the edge of the
> table.
>
> 1. The foot is down as far as it will go.
> 2. The down foot is not touching anything.
>
> We would need one of these sensor pairs for each front foot.
Right. That's listed somewhere in my prior ramblings. It should work fine with
2 inputs (one for each foot)
> When the foot is down and not touching anything, we wait for the foot start to
> rise, and flip the muxes to backwards. At the time the legs get to the top,
> they will go backwards, so I don't think we'll walk off the table. • ...
>
> Remember, 7 of the 8 legs will be down at the time we realize that we're at the
> edge. Given that pneumatics are slow, we should have plenty of time to slam it
> into reverse.
We may need to figure something out, here. I think it WILL be a problem.
At any given time, he will (could) have only four feet on the ground, right?
If I understand this correctly, when he puts his other four feet down, and
senses one foot is off the table, he must continue walking forward, raising four
feet from the table, and leaving him standing on three feet, including NO front
feet.
That makes me think we'll have to consider weight distribution, and how it will
balance on three feet.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > This sounds like we can NOT do the table edge detection stuff, using it's feet.
> >
> > Hmmm.... We would need to know two things to determine we'd hit the edge of the
> > table.
> >
> > 1. The foot is down as far as it will go.
> > 2. The down foot is not touching anything.
> >
> > We would need one of these sensor pairs for each front foot.
>
>
> Right. That's listed somewhere in my prior ramblings. It should work fine with
> 2 inputs (one for each foot)
OK. You are going to put two touch sensors on one port?
I thought you needed to know that the leg is fully extended, but not touching.
How will you do that with one sensor port?
>
>
> > When the foot is down and not touching anything, we wait for the foot start to
> > rise, and flip the muxes to backwards. At the time the legs get to the top,
> > they will go backwards, so I don't think we'll walk off the table. ...
> >
> > Remember, 7 of the 8 legs will be down at the time we realize that we're at the
> > edge. Given that pneumatics are slow, we should have plenty of time to slam it
> > into reverse.
>
> We may need to figure something out, here. I think it WILL be a problem.
>
> At any given time, he will (could) have only four feet on the ground, right?
>
> If I understand this correctly, when he puts his other four feet down, and
> senses one foot is off the table, he must continue walking forward, raising four
> feet from the table, and leaving him standing on three feet, including NO front
> feet.
>
> That makes me think we'll have to consider weight distribution, and how it will
> balance on three feet.
You are right, he needs to balance on the three feet that are further back, but
it will not have to walk forward. It can walk backwards.
Kevin
>
> Steve
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> You are going to put two touch sensors on one port?
>
> I thought you needed to know that the leg is fully extended,
> but not touching. How will you do that with one sensor port?
I can think of two was: use two touch sensors, and be able to read the state
of both of them (you can do this with a 3rd party multiplex, or cybermaster
sensors (which are kinda rare... right, Steve?), or use touch sensors stacked
with standard LEGO lights. Even using this last method (the roughest of the
bunch), I can easily get two touch sensors per port (and could get more, but it
gets bulky).
I'm trying to think of a good way to use two touch sensors ANDed together,
but a "false" reading doesn't tell you the legs have completed their motion.
--
Brian Davis
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > > Hmmm.... We would need to know two things to determine we'd hit the edge of the
> > > table.
> > >
> > > 1. The foot is down as far as it will go.
> > > 2. The down foot is not touching anything.
> > >
> > > We would need one of these sensor pairs for each front foot.
> >
> >
> > Right. That's listed somewhere in my prior ramblings. It should work fine with
> > 2 inputs (one for each foot)
>
> OK. You are going to put two touch sensors on one port?
>
> I thought you needed to know that the leg is fully extended, but not touching.
> How will you do that with one sensor port?
There are a couple ways that come to mind. If you don't want to know the
details, skip ahead...
We want to know when the leg is extended, and the foot is not touching the
ground.
There will be four possible states:
1) Leg up, foot not touching ground
2) Leg extended, foot not touching ground
3) Leg up, foot touching ground
4) Leg extended, foot touching ground
The first thing to remember is that when the foot IS touching the ground, it
doesn't matter if the leg is extended. (that's the key to making it easy)
So, there's only three usable states. The "foot" sensor can override the
reading from the "leg" sensor.
If you connect a light sensor to look at the leg/knee joint, so it reads white
when the leg is extended, and grey or black when it's in other places, then you
can also connect the foot touch sensor to the same port.
When the foot sensor is "open" the RCX will read values from the light sensor,
so it can actually calculate the exact location of the leg/joint. When the foot
sensor is "closed" it can assume the foot is on the ground, so the leg must be
extended.
You could also do something like that with two touch sensors, and a couple
lights. I have even more ideas, but I think the light/touch combination would
be the best, and most "stock LEGO" approach.
> > If I understand this correctly, when he puts his other four feet down, and
> > senses one foot is off the table, he must continue walking forward, raising four
> > feet from the table, and leaving him standing on three feet, including NO front
> > feet.
> >
> > That makes me think we'll have to consider weight distribution, and how it will
> > balance on three feet.
>
> You are right, he needs to balance on the three feet that are further back, but
> it will not have to walk forward. It can walk backwards.
Well, it will walk forward enough to lift the other four feet off the ground,
before it starts backward, again.
I'm going to have to look at the walking logic more to understand why it can't
start walking backward, while all the feet are on the ground.
So, we'll need to make sure the center of gravity is closer to the third set of
legs, and it CAN'T be in front of the second set.
I hope Eric's listening. :)
Steve
|
|
| |
In lugnet.technic, Brian Davis wrote:
> I'm trying to think of a good way to use two touch sensors ANDed together,
> but a "false" reading doesn't tell you the legs have completed their motion.
Again, the main goal is to know when the leg is extended, and the foot isn't
touching the ground.
So, you can mount the foot sensor so it's normally closed (pressed) except when
the foot is on the ground and mount the leg sensor so it is normally open,
except when the leg is down. Then, wire them in series (ANDed) so if they are
both closed, the leg is extended, and the foot is not touching.
Using this method, you could connect both front feet to a single sensor port.
When the port is "closed", one of the feet has a problem, and it must
stop/backup. But, the RCX can't detect if he's successfully walking, like it
can with some other methods.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
<snip>
> > > If I understand this correctly, when he puts his other four feet down, and
> > > senses one foot is off the table, he must continue walking forward, raising four
> > > feet from the table, and leaving him standing on three feet, including NO front
> > > feet.
> > >
> > > That makes me think we'll have to consider weight distribution, and how it will
> > > balance on three feet.
> >
> > You are right, he needs to balance on the three feet that are further back, but
> > it will not have to walk forward. It can walk backwards.
>
>
> Well, it will walk forward enough to lift the other four feet off the ground,
> before it starts backward, again.
Nope. The sequence is:
1. All feet down.
2. Lift one foot group
3. Sweep legs
4. put feet down
5. Lift other foot group
6. Sweep legs
So it does not need to walk forward to lift the feet.
>
> I'm going to have to look at the walking logic more to understand why it can't
> start walking backward, while all the feet are on the ground.
The description is here:
http://www.kclague.net/Quad242/quad242-design.jpg
Piston A is a timing only piston (i.e. it is not part of any leg assembly).
A' and A'' are weight bearing leg pistons that simply mimic A.
Piston B is a timing only piston.
B' and B'' are weight bearing leg pistons that simply mimic B.
C,D,E, and F are hip pistons that force the legs to sweep.
A' is paired with C
A'' is paired with D
B' is paired with E
B'' is paired with F
A contract = B' & B'' & ~C & ~D & E & F
A expand = C & D & ~E & ~F
A' = A
A'' = A
B contract = A' & A'' & C & D & ~E & ~F
B expand = C & D & ~E & ~F
B' = B
B'' = B
C expand = D expand = E contract = F contract =
A' & A''
C contract = D contract = E expand = F expand =
B' & B''
To flip from forward to reverse involves swapping the inputs to C, and the
outputs from F.
The feet go to the ground at synhronization points 0 and 3. Those depend on B',
B'' and A', A'' respectfully. Throwing it into reverse at this time would try
to make the feet sweep while the feet are on the ground.
At synchronization points S1 and S4, two of the four feet are on the ground, and
C, D, E, and F will be trying to change, so this is the fine time to flip it
from forward to reverse or visa versa.
At syncrhonization points S2 and S5, you are depending on the fact that C, D, E
and F have stabilized to expand A or B respectively. Now is not a great time to
reverse polarities because you have a potential race between polarity reversal
and A or B completing expansion, so the results are hit or miss.
>
> So, we'll need to make sure the center of gravity is closer to the third set of
> legs, and it CAN'T be in front of the second set.
>
> I hope Eric's listening. :)
Is he back yet?
>
> Steve
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
<snip>
>
> A contract = B' & B'' & ~C & ~D & E & F
> A expand = C & D & ~E & ~F
>
> A' = A
> A'' = A
>
> B contract = A' & A'' & C & D & ~E & ~F
> B expand = C & D & ~E & ~F
Oops:
B expand = ~C & ~D & E & F
>
> B' = B
> B'' = B
>
> C expand = D expand = E contract = F contract =
> A' & A''
>
> C contract = D contract = E expand = F expand =
> B' & B''
>
> To flip from forward to reverse involves swapping the inputs to C, and the
> outputs from F.
Kevin
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > Well, it will walk forward enough to lift the other four feet off the ground,
> > before it starts backward, again.
>
> Nope. The sequence is:
>
> 1. All feet down.
> 2. Lift one foot group
> 3. Sweep legs
> 4. put feet down
> 5. Lift other foot group
> 6. Sweep legs
>
> So it does not need to walk forward to lift the feet.
I keep thinking about a rotary type walker, where the feet more or less go in a
circle and the body is always moving forward. Please forgive me. :)
So, when it lifts one leg group, and someone switches it into reverse, will it
put those legs back down, right away, and lift the other leg group?
|
|
| |
The latest robotic update:
As it stands now, SSClagorpion will:
1) Move: (one of these at a time)
A) Walk forward & backward,
B) Turn right & left
C) Step right & left (move sideways)
D) Stop
2) Open & close two claws on it's arms (via pneumatics)
3) Move it's tail. (curl? via pneumatics?)
4) It can stay on a table/stage,
5) Detect when it bumps into something (when going forward),
6) Follow the "controller"
7) Be driven around like an RC car.
Is that enough cool functionality?
To do this we need:
1 Spybot w/controller
2 RCXs
4 motors for movement
2 motors for claws (built into Spybot)
1 motor for tail stuff
2 touch sensors for front feet
2 light sensors for leg position (of front feet)
2 touch sensors (or rotation sensors) for arm-bump detection
This leaves us ports for 1 more motor, and 2 more sensors, and there is
currently no function for the touch and light sensors built into the Spybot.
The above totals don't count motors and batteries for a compressor.
That's where I stand.
I'm interested in messing around with an arm design that can use a rotation
sensor to detect when it comes into contact with an object.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > In lugnet.technic, Kevin L. Clague wrote:
-snip-
Good Read.
> Do the legs need to be in a known position when switching from forward to
> backward? Or walking to turning? If so, we need sensors for that...
>
>
> Here's my current bill of materials:
> 1 Spybot, for IR direction control. This is mostly used as a sensor.
> 2 RCXs. I really haven't counted I/O, but two sounds like enough.
> 1 Battery box for compressor(s)
Question please.
The Spybot. That's for Directional Control.
Can you eloborate? I find this interesting.
I have conceived of possible contrivences, I'm interested in dicussing a method.
What would the Spybot's two motors trigger exactly?...to aid in directional
control?
It also needs to be in a line of sight with the RCX right?
What possible arrangement set ups could there be? /are best?
Where in/on the body is best?
What main factors need to be in play in order to utilize the Spybot?
ie, what main valve switches, sensors, motor/polarity/pnuematic reversors need
to be in alignment so the functions can be triggered or guided by the Spybot.
Does it come down to having the said controls ending up at the outputs of the
Sypbot? So ultimately, what should those master controls be?
Strictly in the sense of, oh wait, could it just be two simple sensors that send
feedback to the RCX to trigger the Master Gait seqences? For turning, and moving
about?
Thoughts?
> All those should be installed in the body of the beast. The Spybot/RCXs will
> need to communicate via IR. (there are other options, but we'll start there)
>
> The Spybot has two motors, but very limited sensor feedback. It would be good
> to use the Spybot motors to activate something like claws (open & close), where
> it won't be critical if they work exactly right.
Just a thought. Having them in the Claws would be cool. It does alter the scale.
Or more so, it sets a certain scale range minimum.
btw, I have three Spybots.
> We can use the Spybot remote
> to send it commands, like "follow me" or "attack anything in front of you"
Refer to above, like I was asking, if you'll indulge me, what would you say in
your opinion is/are these funtions. What do they exactly refer to in the actual
design set up?
> All the outputs from 1 RCX control movement. 1=Activate, 2=function
> (turn/straight), 3=forward/backward
>
> Two inputs would be used for sensing the edge-of-table. That's one input for
> each front leg. Each leg would have one sensor to detect when the leg is in
> it's "down" position, and one sensor to detect the ground. The two sensors in
> each leg can be connected together, to use one input. This could be two touch
> sensors, or a touch+light sensor combination.
>
> Two inputs can be used for the arm "feelers". Maybe several touch sensors per
> arm, or one rotation sensor for each.
>
>
> I'm not sure the RCXs really need to "know" when the SSClagorpion is actually
> walking. It can assume if it runs the motors, then the switches are set, and it
> is going.
>
> Beyond that, there is one motor to set the switch for curling the tail.
>
> That leaves 2 motor outputs, and 2 sensor inputs, for further expansion. Plus a
> touch sensor & light sensor built into the Spybot. The outputs from the second
> RCX can be connected to a gearbox, like Eric was suggesting, if we want to
> control more things, but I think that may just make things more complex than
> they need to be.
Hey, that's up to you. If we need strength, either multiplex the pistons force
output with levers or other fulcrum advantages or go mechanical an gear it. If
the digits of this project prove wimpy. I'll mechanize it.
Ain't nothin a couple of 8 and 40 tooth gears and motors with my homie; wormie
can't fix.
>
> It would be nice to mount the Spybot in the end of the tail, but it needs to
> communicate with the RCXs via IR, so I don't think that will work well, unless
> we connect a fiber-optic line. No, I don't think that will work, either.
>
>
> So, SSClagorpion will walk forward & backward, turn right & left, open & close
> two claws, and move it's tail. It can stay on a table/stage, detect when it
> bumps into something (going forward), follow the "controller" or be driven
> around like an RC car.
>
> Is that enough cool functionality?
>
> Steve
Excellent. Somebody dare us to add Clikits.
we da meign.
e
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > Ok, so what's required for walking sideways?
> > > I just happened to think, the legs should be in a known position when switching
> > > from forward/back to side-to-side.
> > >
> > > Do the legs need to be in a know position when switching from forward to
> > > backward? Or walking to turning? If so, we need sensors for that...
> >
> > Yes. Changing the forward/backward/turning muxes needs to be done when the feet
> > are in the air. Observing any one foot is probably sufficient.
> >
> > The above costs us four motors, and one sensor.
>
> Ok, we [plan to] have a sensor on the front foot, and can detect when that is
> down. So, no more sensors are required.
>
> So you said, the forward/backward must be changed while the feet are in the air.
> We've been thinking of having a touch sensor on the front feet, so it can detect
> when it goes off an edge (like a table). However, it would not know the surface
> is not there, until it trys to put the foot down. So, all the feet would be
> down, and it needs to backup. Can it do that? Or will it need to keep walking
> a bit?
Hey that is interesting. How would the robot actually behave in this
circumstance? We'll have to build it and teach it and build as necc.
e
>
> Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
Ahh now we are getting somewhere.
> This sounds like we can NOT do the table edge detection stuff, using it's feet.
>
> Imagine you're crawling around on your hands and knees, with your eyes closed.
> When you get to the top of a stairway, you put your hand down, and it doesn't
> touch the ground.
>
> If I understand you correctly (which doesn't always happen via e-mail) you must
> continue to crawl forward, until your other hand and knee are off the ground.
>
> That will be a problem.
Hmmm, indeed, or perhaps if we monitor the time it takes for a Leg to hit the
ground and then install a logic command that will take action if the leg does
not hit the ground in a predetermined amount of time. If the leg's foot sensor
does not detect the presence of solid ground within the time it normally takes
to hit the ground, it can be presumes there is no ground and this could be
managed and corrected. So when a foot does not "feel" the ground ahead of it in
a predetermined amount of time; the time it normally takes to hit the ground, it
can begin the self preservation/course reversal actions.
Could we still do this then?
> It's fine if it can't turn, or switch to moving sideways while all feet are on
> the ground, but it really needs to be able to reverse direction.
>
> I suspect the front legs can be a small amount longer, so it will detect the
> lack of ground, before the weight is really shifted to the new legs.
>
> Will that work?
Or install a command that will trigger the reverse seqence, should the foot not
send a tounch down signal within the expected arrival time.
cool right. I got my hand in the cookie jar.
e
> Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
Ahem,
> So, we'll need to make sure the center of gravity is closer to the third set of
> legs, and it CAN'T be in front of the second set.
>
> I hope Eric's listening. :)
>
> Steve
So we can incorporate the tail's funtionality and weight to help aid the robot
by swinging or uncurling the tail so it faces away from the body and is extended
back. This shifts the center of gravity away from the front of the body. Hey a
Scorpion's tail is pretty handy huh?
e
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
>
> <snip>
> >
> > I hope Eric's listening. :)
>
> Is he back yet?
>
> >
> > Steve
>
> Kevin
I'm back I'm back. What can I do?
I want to build something.
I'm gonna make a concept moc moc up. Pun intended.
e
|
|
| |
In lugnet.technic, Eric Sophie wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> >
> > 1 Spybot, for IR direction control. This is mostly used as a sensor.
>
> Question please.
>
> The Spybot. That's for Directional Control.
> Can you eloborate? I find this interesting.
> I have conceived of possible contrivences, I'm interested in dicussing a method.
> What would the Spybot's two motors trigger exactly?...to aid in directional
> control?
>
> It also needs to be in a line of sight with the RCX right?
> What possible arrangement set ups could there be? /are best?
> Where in/on the body is best?
>
> What main factors need to be in play in order to utilize the Spybot?
> ie, what main valve switches, sensors, motor/polarity/pnuematic reversors need
> to be in alignment so the functions can be triggered or guided by the Spybot.
> Does it come down to having the said controls ending up at the outputs of the
> Sypbot? So ultimately, what should those master controls be?
>
> Thoughts?
Ok, so first of all, the Spybots, although limited, have one VERY cool feature.
It's their IR capability.
They have three IR sensors, so they can detect the direction of an IR signal
from a controller, and/or another Spybot. So, if you set a controller to beacon
mode, the Spybot can be programmed to follow it.
Basicly, I was thinking of using a Spybot as an external sensor for the main RCX
that's controlling SSClagorpion. When the Spybot receives signals from the
controller, it can process them, and send commands (via IR) to the RCX.
As for using the motors (so they don't go to waste) they could be connected
directly to pneumatic valves that open and close the claws.
The RCXs should be facing the Spybot. The Spybot can face any direction, but
should be in the same plane as the RCXs.
> Just a thought. Having them in the Claws would be cool. It does alter the scale.
> Or more so, it sets a certain scale range minimum.
I think it would be fine having a Spybot as the "head", connect two valves to
the motors, and only put the weight/size of a pneumatic cylinders in the claws.
Then the RCXs would be side-by-side in the neck/body of the beast.
> > We can use the Spybot remote
> > to send it commands, like "follow me" or "attack anything in front of you"
>
> Refer to above, like I was asking, if you'll indulge me, what would you say in
> your opinion is/are these funtions. What do they exactly refer to in the actual
> design set up?
Well, pretty much whatever commands we want to send to it. What do you want to
tell it to do? :)
As I said, we can have it follow the remote (turn towards the remote and walk
forward) or tell it to "STOP", or go forward...
|
|
| |
In lugnet.technic, Eric Sophie wrote:
> In lugnet.technic, Steve Hassenplug wrote:
>
> > This sounds like we can NOT do the table edge detection stuff, using it's feet.
> >
> > That will be a problem.
>
> Hmmm, indeed, or perhaps if we monitor the time it takes for a Leg to hit the
> ground and then install a logic command that will take action if the leg does
> not hit the ground in a predetermined amount of time.
Detecting the table edge is not the problem. We can do that with a sensor in
the hip/leg in addition to one in the foot. Then, when the leg is extended, and
the foot is not touching, there is an error.
The problem (as I see it) is that it must can not backup right away. It must
lift the other leg group off the surface, before it can backup.
I'm not sure the exact sequence (I still haven't looked at Kevin's 242 paper
long enough to understand it) but we may be able to start backing up as soon as
the valve is switched, causing the legs to start UP. So, the feet would still
(pretty much) be on the ground, and it will start backing up.
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > Well, it will walk forward enough to lift the other four feet off the ground,
> > > before it starts backward, again.
> >
> > Nope. The sequence is:
> >
> > 1. All feet down.
> > 2. Lift one foot group
> > 3. Sweep legs
> > 4. put feet down
> > 5. Lift other foot group
> > 6. Sweep legs
> >
> > So it does not need to walk forward to lift the feet.
>
> I keep thinking about a rotary type walker, where the feet more or less go in a
> circle and the body is always moving forward. Please forgive me. :)
>
> So, when it lifts one leg group, and someone switches it into reverse, will it
> put those legs back down, right away, and lift the other leg group?
Nope. It will force the legs to the correct position for walking backwards, and
then drop the feet.
Kevin
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Eric Sophie wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > >
> > > 1 Spybot, for IR direction control.
> Basicly, I was thinking of using a Spybot as an external sensor for the main RCX
> that's controlling SSClagorpion. When the Spybot receives signals from the
> controller, it can process them, and send commands (via IR) to the RCX.
Oh! That's cool, I get it.
> As for using the motors (so they don't go to waste) they could be connected
> directly to pneumatic valves that open and close the claws.
Right! Right! I get it, that answers the questions. I see. So the Spybots (oh
course doi!) wouldn't be in the claws. The motors of the Spybot (located in the
head or body or where ever) could trigger the Claw's valves. Because they are
hoses the "buisness end of the circut could be anywhere.
yep. yep.
> The RCXs should be facing the Spybot. The Spybot can face any direction, but
> should be in the same plane as the RCXs.
k.
> As I said, we can have it follow the remote (turn towards the remote and walk
> forward) or tell it to "STOP", or go forward...
nice. it will almost be alive.
Thought does come to mind.
Kevin, if you feel you need the Motor outputs from the Spybot to aid in the
walking systems et als... that option is also open. We don't have to use the
Spybot for the claws, but it is an example as I'm sure we all know. Though we
could use it as such.
e
|
|
| |
> I'm gonna make a concept moc moc up. Pun intended.
MOC-up! I love it!
|
|
| |
In lugnet.technic, Kevin L. Clague wrote:
> In lugnet.technic, Steve Hassenplug wrote:
> > So, when it lifts one leg group, and someone switches it into reverse, will it
> > put those legs back down, right away, and lift the other leg group?
>
> Nope. It will force the legs to the correct position for walking backwards, and
> then drop the feet.
I could be wrong (missing something) but I think if the switch is flipped
immediatly after they raise up, then the legs are in the correct position.
If you're walking forward, and switch to reverse right as your foot comes off
the ground, you just want to put it back on the ground, right where it was.
I don't think the legs need to be moved.
Of course I still haven't really looked at your drawing...
Am I missing something?
Steve
|
|
| |
In lugnet.technic, Steve Hassenplug wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > In lugnet.technic, Steve Hassenplug wrote:
> > > So, when it lifts one leg group, and someone switches it into reverse, will it
> > > put those legs back down, right away, and lift the other leg group?
> >
> > Nope. It will force the legs to the correct position for walking backwards, and
> > then drop the feet.
>
>
> I could be wrong (missing something) but I think if the switch is flipped
> immediatly after they raise up, then the legs are in the correct position.
>
> If you're walking forward, and switch to reverse right as your foot comes off
> the ground, you just want to put it back on the ground, right where it was.
>
> I don't think the legs need to be moved.
>
> Of course I still haven't really looked at your drawing...
>
> Am I missing something?
Probably not.
Kevin
>
> Steve
|
|
|
