UR5 Post Processor Joint values

General Questions
#1

Hi,

I’m trying to post process a simple UR5 robot program from the Visual Components to a real robot. I’m using the standard post processor which is available here in the forum (dated 29.01.2021).

Currently the problem is that the joint values of the robot are changing during the post processing. So the robot in Visual Components is moving as it should be but when post processing the program to a script and when running this script with the real UR5 the robot is reaching towards zero values. How can I fix this? What I’m doing wrong?

Thanks

#2

Can you post the model or at least the script from PP?

-k

#3

Hi,

Here is the VC model and the script from the post processor.

UR5test.vcmx (1.2 MB) UR5test.script (646 Bytes)

The wait and Set output commands are working as should be. Only problem is the robot movements.

Thanks!

#4

I don’t know what the issue could be. I tested the script on Polyscope and the motions seem to match VC simulation. Check the attached video.

I don’t know if there’s an option in UR to give joint positions as degrees instead of radians. In that case positions in script would be really small degree values so close to zero position. But at least in the documents I’ve seen there’s no option like this and joint positions should be given as radians. And that’s how it’s done in our PP.

>

-k

#5

Hi,

I created the same program manually with the robot. I noticed that in the script from the robot there is used command “get_inverse_kin” when using the movej command. In the script which is created with the Visual Components there isn’t that command (get_inverse_kin) when using the movej command. Should the post processor of the Visual Components create this get_inverse_kin command when transferring script to the UR5?

See the attachments (“UR5robotilta” is the script from the real robot and “UR5test” is the post processed script from the Visual Components).

UR5robotilta.script (54.4 KB)
UR5test.script (646 Bytes)

#6

get_inverse_kin shouldn’t be needed. I mean it just converts cartesian position to joint position but script should accept joint position directly. The polyscope test I made worked fine and also I know that PP has been used on many real UR robots. Here’s the UR script manual and for movej (p.24) the first argument a is a joint position and in its example there no get_inverse_kin but just joint position as radians: https://s3-eu-west-1.amazonaws.com/ur-support-site/62056/scriptManual.pdf
I don’t know if this could be controller software version issue or something.

-k