Fix 6DOF arm example #804
Fix 6DOF arm example #804
Conversation
github-actions
bot
requested review from
anfemosa,
bmagyar,
christophfroehlich,
DasRoteSkelett,
homalozoa,
mamueluth and
mcbed
|
@mikelikesrobots did you run into this in your recent tutorial BTW? This is how we found this issue... |
I didn't run into this at all! The issue I had was that it didn't always receive the trajectory, or it ignored it. When it did receive the trajectory, it did a perfect circle. Your code changes do point out an issue, though. |
... and I didn't see that myself. I am using Cyclone DDS instead of the default. Maybe give this a quick try on your end to see if that could be the difference? |
it does ;) |
There was a problem hiding this comment.
The changes LGTM
Thank you for the changes @sea-bass
Thanks for your review and follow-up @mikelikesrobots
There was a problem hiding this comment.
The changes LGTM. I had problems in my local dev environment (see #805), but @saikishor confirmed that it works so let's go ahead
christophfroehlich
merged commit 1dc20c1
into
ros-controls:master
(cherry picked from commit 1dc20c1) # Conflicts: # example_7/controller/r6bot_controller.cpp # example_7/reference_generator/send_trajectory.cpp
* Backport changes from #623 * Backport changes from #753 * Fix 6DOF arm example (#804) * Revert changes to handles --------- Co-authored-by: Christoph Froehlich <[email protected]> Co-authored-by: Sebastian Castro <[email protected]>
Thanks to debugging by @catarina-p and myself, we found issues with the 6DOF arm example.
Turns out the controller kept extrapolating forever past the final point, meaning that instead of slowing down from positive to zero velocity, the velocities kept going negative further and further.
Before this fix:
https://github.com/user-attachments/assets/af815279-818a-4922-837d-94b0208035cc
After this fix:
https://github.com/user-attachments/assets/70f4796e-38e1-4ba5-b2f2-5e402c324f2c
I also did some cleanup of the trajectory time calculations since the final point's time was not exactly 3.0 as expected.
Edit: As a side note, I am somewhat surprised that this example doesn't just use the regular joint trajectory controller that is already available in ros2_control. Although it's a nice example for people writing their own controllers.