Bioloid cartwheel?

[cdraptor]

I had played a bit with the Bioloid trying to get this to work I noticed a few things when comparing it to the RN-1. First major thing I noticed is check out the servo configuration of the RN legs versus the Bioloid is where the legs swing out, the RN-1 joint is much higher up into the body - the Bioloid joint is much lower where the leg swings out, the joint above it is the leg rotation. So by the time you would tip to the side and trying to roll-over on the arms, the weight of the Bioloid is on it's side and it's got more weight on the legs to try and roll over. Every attempt I did all I got was a Bioloid crashing over on his side, the body doesn't create as well of a circular structure like the RN.

[tempusmaster]

I had played a bit with the Bioloid trying to get this to work I noticed a few things when comparing it to the RN-1. First major thing I noticed is check out the servo configuration of the RN legs versus the Bioloid is where the legs swing out, the RN-1 joint is much higher up into the body - the Bioloid joint is much lower where the leg swings out, the joint above it is the leg rotation. So by the time you would tip to the side and trying to roll-over on the arms, the weight of the Bioloid is on it's side and it's got more weight on the legs to try and roll over. Every attempt I did all I got was a Bioloid crashing over on his side, the body doesn't create as well of a circular structure like the RN.

How about trying it with the Speecys?

[NovaOne]

Would longer arms help?

[cdraptor]

I think you could possibly do it if you extended the arms on the Bioloid. As for the SPC-101C the arm configuration would probably be perfect, but the hip servos do not provide very much freedom in extending the leg outward. So moves like a high side kick like the RBT-1 cannot be pulled off with the SPC-101C.

[srobot]

After about 3 hours of programming I got the first part to work, but as soon as William (my Bioloid) gets on his hands he flops over.

I'll keep working on it.

--Scotty

[srobot]

Here is an improved version of the first part of the cartwheel:

http://cid-d3d82c952d9be25d.skydrive.li ... rtwheel%20[1st%20Part]

Enjoy,
--srobot

[srobot]

It looks like I got it!

I am using an IMU from HUV Robotics so the motion is more reliable, although the cartwheel still works without it (a bit more sloppy)!

The battery is charging now, so I will do some more testing and improvements after it is charged, then I will post a video and code!

http://www.huvrobotics.com

Cheers,
--Scotty

[srobot]

Within 12 hours a video and working code should be posted!

With the HUV Robotics IMU about 3 in 5 cartwheels work, without the IMU 1 in 5 work.

Let the countdown begin!

--Scotty

[Bullit]

I'm currious how you are using the huv imu in your code?
I don't belive the stocks motion firmware supports it.

[srobot]

I'm currious how you are using the huv imu in your code?
I don't belive the stocks motion firmware supports it.

I use it in the BCP not the Motion Editor, but yes the stock firmware does support it. I just put the IMU in line with the Dynamixels and the CM-5 found it. I also tried this with an extra Robotis servo, but it got an err.

So to use it all you do is:

Open the BCP Editor.

Choose a spot where you want to add the IMU.

Choose "Dynamixel" out of "CM-5" and "Typing".

Type "120" into ID, and choose an Address out of the Ram table on the HUV website.

So to see if the Humanoid is laying down you can do:

If 120:28 [Comment: ID 120, and Address 28] > 395 Then

Let me know if this is hard to follow.

--Scotty

[Bullit]

So how do you use this to keep your robot balanced?

[srobot]

phpBB [media]

Although it looks like I am "helping" it in the second half of the cartwheels, I am not. My hand is 1 - 2 inches away. It was a reflex because of the few times William (the Bioloid in the video) slammed against the wood board in the back.

--srobot

Edit - Due to some comments this video has received, I will do a new video in a few days so my hand WILL NOT be near the robot, also so the camera DOES NOT have to move, AND the robot will stay in the ending spot longer.

It's late, so give me 'till tomorrow, or the next...

[jaybirdtke]

Good job. Quite creative the way you got him to roll. That's the most progress I've seen in making a bioloid do a cartwheel so far....
I'm curious why you aren't using the tradional poses that I've seen the khr and robonova use to perform a cartwheel.

[Rob]

Nice work, regarding the IMU I have read the documentation on the Twiki. I was wondering if you could either point me in the direction of more documentation or tell me what the ranges of possible values are for the table for the IMU measurements below.

26 (0x1A) Sideways Accel(L)
27 (0x1B) Sideways Accel(H)
28 (0x1C) Forwards Accel(L)
29 (0x1D) Forwards Accel(H)
30 (0x1E) Vertical Accel(L)
31 (0x1F) Vertical Accel(H)
32 (0x20) Pitch Rate(L)
33 (0x21) Pitch Rate(H)
34 (0x22) Roll Rate(L)
35 (0x23) Roll Rate(H)
36 (0x24) Yaw Rate(L)
37 (0x25) Yaw Rate(H)

Also you said previously 395 for Forwards Acceleration tells us the robot is lying down, couldn't the robot just be standing up and not moving forward and the IMU control table would have the same value?

[JonHylands]

Nice work, regarding the IMU I have read the documentation on the Twiki. I was wondering if you could either point me in the direction of more documentation or tell me what the ranges of possible values are for the table for the IMU measurements below.

26 (0x1A) Sideways Accel(L)
27 (0x1B) Sideways Accel(H)
28 (0x1C) Forwards Accel(L)
29 (0x1D) Forwards Accel(H)
30 (0x1E) Vertical Accel(L)
31 (0x1F) Vertical Accel(H)
32 (0x20) Pitch Rate(L)
33 (0x21) Pitch Rate(H)
34 (0x22) Roll Rate(L)
35 (0x23) Roll Rate(H)
36 (0x24) Yaw Rate(L)
37 (0x25) Yaw Rate(H)

Also you said previously 395 for Forwards Acceleration tells us the robot is lying down, couldn't the robot just be standing up and not moving forward and the IMU control table would have the same value?

You should read this thread:

http://robosavvy.com/forum/viewtopic.ph ... highlight=

where I document all of the ranges. I should add that information to the wiki.

Gravity is an acceleration, so the "at rest" value you get from the accelerometer is different depending on the orientation.

- Jon