Add support for external axes on the ABB driver #150
Conversation
* Motion server now takes into account all 4 additional joints present in the message and assigned them to `eax_a`, `eax_b`, `eax_c` and `eax_d`. * Status server sends all connected external axes. Non-connected are zero filled.
|
Thanks for the PR, much appreciated 👍 🎉 Quick question: is there a conversion from |
|
@gavanderhoorn you're totally right, one sec... |
abb_driver/rapid/ROS_motion.mod
Outdated
| @@ -54,6 +54,7 @@ PROC main() | |||
| IF (trajectory_size > 0) THEN | |||
| FOR current_index FROM 1 TO trajectory_size DO | |||
| target.robax := trajectory{current_index}.joint_pos; | |||
| target.extax := trajectory{current_index}.extax_pos; | |||
|
|
|||
| skip_move := (current_index = 1) AND is_near(target.robax, 0.1); | |||
There was a problem hiding this comment.
Would this also not need to check the pose(s) of the external axes? Otherwise we could run the risk of skipping points that satisfy is_near(..) for the robot itself, but not for the ext axes.
There was a problem hiding this comment.
Fixed it on 3b1d413
|
Fixed unit conversion on 289fc4b |
|
@gavanderhoorn thanks for checking this! hope it's better now 😉 |
|
@gonzalocasas: thanks for iterating. You wrote you've tested this on your hw. Did you only look at the |
|
Both. I was sending trajectories using this little roslibpy library that we released recently so I was testing the full loop. |
|
Using metres now, I hope? :) |
|
Yes, indeed. 😉 |
|
@gonzalocasas: my apologies for letting this slip. I'll confirm this implementation tomorrow in RobotStudio and merge afterwards. |
|
Thanks again for contributing this @gonzalocasas 👍 much appreciated. Could I ask you to add a Features section (after the Overview section) to wiki/abb_driver describing this new support? I'll add other things we might want to list (such as basic 6 axis robot support, agnostic of actual robot model configured, etc). |
|
One thing I forgot to ask/mention: this PR assumes that external axes are all linear units. Afaik, that is not necessarily true. I'm not sure how to determine at runtime whether an axis is a linear unit or a rotational one, but that would seem to be required here. |
@gavanderhoorn thx for adding that. I thought I edited already, but I think I left the tab open in preview, silly me. Anyway, I added one bit to that. And maybe a triviality, but instead of saying |
|
@gonzalocasas wrote:
Yes, that may be too limited a description. I just wanted to add a statement about what is at least definitely supported. If you can verify support for 4-axis robots (and I actually expect that to work right away), please update that bullet point.
Thanks. I've changed the text around a bit more again. Mostly order of things. |
Squashed commits: * Add support for external axes on the ABB driver * Motion server now takes into account all 4 additional joints present in the message and assigned them to `eax_a`, `eax_b`, `eax_c` and `eax_d`. * Status server sends all connected external axes. Non-connected are zero filled. * Convert to and from ROS and ABB units (meters and millimeters respectively) * Include eax in `is_near` check
As discussed on https://github.com/ros-industrial/abb/issues/135, here's a proposal for integrated external axes support on the ABB driver. I tested on IRB4600 real hardware on a linear track.
eax_a,eax_b,eax_candeax_d.I explicitly did not change tabs to spaces in places where it would have been logical to do so (e.g.
ROS_common.sys) to avoid merge conflicts with #145