ThRowbonova

DirtyRoboto · Posted: Sat Mar 03, 2007 8:16 pm Post subject: ThRowbonova

What you are about to see was done in the space of three hours. I had had a hard day at work (yes on a Saturday) and reached home about 17:00 after a wind down I decided to play with Takeshi (my RN).

After a bit of messing about I decided to use the new grippers and an idea about simulating the transfer of energy through an autonamous system.

I wanted to produce a mechanical simulation of how humans input power (energy) into objects to be thrown. There is a snap or whiplash action to transfer more power to the projectile.
I soon learned that the RN just using the shoulder rotation for power was not nearly enough. I then looked at producing a way of using balance to transfer the power, I ended up using an extended rocking motion to impart the extra power needed, by starting an arc using the rocking just before the arm is bought into motion, thus giving the arm a greater forward speed in relation to the ground.

Anyway 3 hours later I have recorded, edited and uploaded for your delight a compilation of my 3 hours of fun............

Marcus.
_________________
In servo's we trust!

NewBreedWarrior · Posted: Sat Mar 03, 2007 10:20 pm Post subject: hmmm

lol well that amused me! nice thought put in action, I understand how that would be very important. Maybe one day we can learn our RN-1's to use fluid motion combined with kicks and all fighting concepts
_________________
In the fell clutch of circumstance
I have not winced nor cried aloud.
Under the bludgeonings of chance
My head is bloody, but unbowed.

DirtyRoboto · Posted: Sun Mar 04, 2007 11:36 pm Post subject:

If there is no RN throwing record then I would like to set the challenge of throwing a match box or item of similar weight.
The match box Takeshi threw was 8g and his best throw was 28cm.
The RN hand and also the chest plate came in at 26g and both averaged at 22 cm.
The English 1p (penny) best throw was 32cm on a full charge. For the first 6 throws the distance was good then it started creeping down as the power was used up.

Marcus
_________________
In servo's we trust!

Humanoido · Posted: Mon Mar 05, 2007 7:59 am Post subject:

Marcus, I think you could get Takeshe to throw even farther. The goal is to release the projectile at exactly 45 degrees at the maximum velocity. Using the legs will help to increase the velocity, just like some forms of the shotput competition in track and field. If RN can balance, start from a crouched position and accelerate. A slight backwards bend to issue a little snap motion is helpful. You probably already experimented by increasing the speed of the servos. Maybe a ramping vector would help stabilize upon returning to the stance and maintaining balance. Great video and congratulations on this wonderful novel idea, and showing us your skill in successfully achieving it. I have to agree, it was very entertaining!

humanoido

I shall be telling this with a sigh
Somewhere ages and ages hence:
Two roads diverged in a wood, and I--
I took the one less traveled by,
And that has made all the difference.

mthomson · Posted: Mon Mar 05, 2007 2:23 pm Post subject:

Nice work Dirtyroboto.

After watching the movements of Tekeshi throwing overhand I'm left wondering if throwing underhand would yeild better results. Perhaps come up out of a crouched position and release the object right when RN1 gets to the standing position, thus utilizing the motion of standing and of the swinging arm together......just a thought.

As soon as I get my grippers and my rc tx/rx installed I'll give it a try.

Keep up the great work, it definitely inspires me to do more with my RN1.

mark

gdubb2 · Posted: Tue Mar 06, 2007 4:49 am Post subject:

Hey Marcus,
Did I hear a gauntlet hit the floor???? Why YES I DID.......I heard it clear over on this side of the pond.. Very Happy

I've been working on tossing a ping pong ball for the toss event at Robogames. I'm not doing very well, due to the distance required. But just for fun I thought I would take on the challange.

A ping pong ball only weighs about 3 grams. Well thats not exactly fair, so I shoved enough small steel wire inside the ball I used to make it weigh 8 grams.

Now depending on how you measure the distance, The ball went between 24 inches (61 cm), and 30 inches (76 cm). The large square floor tiles are 12 inch X 12 inch. A unweighted ping pong ball (3g) goes 1 meter.

There are 3 arm servos involved and one foot servo. The foot servo in this case did not add any distance, just made it look a little cooler. But I still need to get a ball to go 1 meter but also have an altitude of 60 cm. when it hits.

Gary

DirtyRoboto · Posted: Wed Mar 07, 2007 7:04 pm Post subject:

That was really cool, and very far!

I must point out the difference in what is being done by you and I.
I wished to explore a fixed hand set that would cope with a range of objects and have a good power grip. I also wished to throw forward for asthetics and a little experiment. My main focus was to throw an object and then aquire a distance reading for it. After 5 throws I wanted the RN to beep back the best throw.

Also, I had to set up a motion and trigger a release within that motion, this is what I was displaying. The fact that a release can be a seperate entity with in an already executed move.
_________________
In servo's we trust!

gdubb2 · Posted: Wed Mar 07, 2007 7:23 pm Post subject:

Hi Marcus..

I know there is a world of difference in the way we are throwing things. I haven't even tried to use the same method you are using. Although I did try to throw foreward before I switched to the side movement. I didn't get much distance.

I'm simply after height and distace and hopefully someday accuracy. After I pick up the ball, I hold it upright so it can't get away. But then the fun starts. 1 line of code before the throw sequence, I start closing the gripper. This starts the ball moving foreward in the wires. Then in the line of code that does the actual toss, I finish the closing the grip and do the foreward movement. The grip closing causes the ball to be propelled foreward. Kind of like being squirted from between your fingers. You don't see that part in the very poor video. It's hard to detect when looking right at it.

Good luck with your experiments, you are doing great.

Gary

DirtyRoboto · Posted: Wed Mar 07, 2007 7:52 pm Post subject:

I watched your clip quite a few times. At first I thought you were just using gravity to stick the ball to the bot. then I noticed that it was an actual grip of the ball and a release (of sorts).

My routine starts the arm off moving (the move is in two parts) first forward and then back. The back move is for balance only.
During this sequence I perform the release. I do this by setting a delay without WAIT so that the release is performed xxxx amount of time after the initial throw move is performed.

Marcus.
_________________
In servo's we trust!

Humanoido · Posted: Tue Mar 13, 2007 12:39 pm Post subject:

gdubb2, can you have the humanoid throw
the ball unassisted?

Perhaps the same type of springing motion
can be created by making one arm
stationary and holding the ball, while the
other acts as the loading element and
trigger.

humanoido

mthomson · Posted: Tue Mar 13, 2007 2:39 pm Post subject:

Humanoido,

Like an underhand volleyball serve?

Humanoido · Posted: Wed Mar 14, 2007 1:04 pm Post subject:

Yes, like an underhand volley ball serve, only, to cock the hand holding the ball you would program RN to pull the arm back a little to assist and then let it spring forth, just as you did to help RN in the video.

As I envision the process, both arms to the front near waist, right arm holding the ball (need to add some kind of sling for the ball), then RN left hand on top of right hand to apply some pressure (the cocking to trigger the release), and then the full release with the right arm.

humanoido