[Jorge]

I`ve downloaded the Franka Emika Robot from the Robot Library. When I try to plot the robot for a joint position, I don't see the correct pose representation in RoboDK. Maybe there is a wrong joint definition in the Franka Emika Robot that was uploaded in the Robot Library.

I have also tested in Matlab, using Peter Corke's toolbox. The representation in Matlab is correct, but the representation in RoboDK is incorrect. I share here the script in Matlab:

L1 = 0.333;
L2 = 0.316;
L3 = 0.0825;
L4 = 0.384;
L5 = 0.088;
L6 = 0.107;

L(1) = Link('revolute','d',L1,'a',0,'alpha',-pi/2);
L(2) = Link('revolute','d',0,'a',0,'alpha',pi/2);
L(3) = Link('revolute','d',L2,'a',L3,'alpha',pi/2);
L(4) = Link('revolute','d',0,'a',-L3,'alpha',-pi/2);
L(5) = Link('revolute','d',L4,'a',0,'alpha',pi/2);
L(6) = Link('revolute','d',0,'a',L5,'alpha',pi/2);
L(7) = Link('revolute','d',L6,'a',0,'alpha',0);

R = SerialLink(L);
R.name = 'Franka';
q = [0 -pi/4 0 -3*pi/4 0 pi/2 pi/4];
R.plot(q)

[Jeremy]

Thanks for noticing us.

We will try to fix that quickly.

Jeremy

[Mad]

Hi Jeremy,

I am having some troubles with that Model too:
1. The command MoveJ_Test (python) is very slow with this Model. Actually almost unusable.
2. It is not possible to display the workspace. There are only some dots somewhere along the x-plane. Maybe I don't understand it, but it doesn't look right to me.
3. Crossing the x-plane (spanned by the unit x and unit z vector) is causing problems. One joint is not wrapping around and stays locked at one end position, causing the arm to point into some random position. (Not sure if that is really the problem, but just for describing the problem.)

[Albert]

Collision detection time depends on many things. I recommend you to take a look at this page to speed it up:
//www.sinclairbody.com/doc/en/Collision-Avoi...isionCheck

We'll try to improve displaying the workspace for the Franka Emika Panda robot but it may take some time.

If you can share the RDK file it will help us understand the 3rd issue you mentioned.

[Mad]

Collision detection time depends on many things. I recommend you to take a look at this page to speed it up:
//www.sinclairbody.com/doc/en/Collision-Avoi...isionCheck

We'll try to improve displaying the workspace for the Franka Emika Panda robot but it may take some time.

If you can share the RDK file it will help us understand the 3rd issue you mentioned.

谢谢你的联系。我能理解你的观点at collision detection depends on other variables too.

I want to add one observation which makes me think it is model specific: When I run the MoveJ_Test command with other models (even other 7 DOF models) the command is much faster. I also noticed the size of the Franka Emika Model is quite big. Is there maybe some correlation?

I already tried to simplify the 3D model and fix one of the 7DOF (joint 3) as in//www.sinclairbody.com/forum/Thread-New-arm-error-Panda. (Unfortunately I wasn't successful yet, since the 6 DOF industrial robot arm template does not allow for a z shift of the flange joint.)

I will check if I can share RDK files with you. Maybe some minimal working (failing) example.

[Mad]

I added some files:

- RoboDK_test.py is telling the robot arm where to go. Create a new station inside RoboDK, load the franka emika robot arm and run the script.

- FrankaEmikaFail.mp4 shows what happens on my PC and xArm6Working.mp4 how it should look like - the blue tool vector should touch the red base vector tip

I also added two images of the problematic positions - First iteration works, second does not

[Mad]

Just noticed a bug in the python script, (I used two times the variable "i" for the iterator) I can't edit the old post so I just post the updated version here.

From the functionality point of view nothing changes.
Sorry

[Mad]

Is there any update on this?

[Mad]

Are there some news?