ur10 not performing well with current physic settings
In this commit:
https://bitbucket.org/osrf/ariac/commits/26186a4ff11723ec8c651f720bff97fa35e3d5e7
We set the max step size to 0.005 and that seems to cause shaking and incorrect joint displacement in the ur10 joints. I only figured this out after discussing it with @scpeters. Setting them back to the defaults resolved the issue for me.
I don't know why we decided to do this, but I would guess it was to improve the simulation performance. So we'll either need to change this back to the default (and live with slower sim time values) or we'll have to figure out what other settings to tweak in order to get reasonable behavior out of the arm.
I can describe the incorrect behavior caused by this in more detail if necessary in person, but I think it is responsible for the shaking of the ur10 described by some users.
Comments (4)
-
-
I've reverted the change to the physics settings in https://bitbucket.org/osrf/ariac/commits/27746d3acc7c02b0fd17fabed23037bb303ccf24 - we'll likely have to revisit this issue when we start using the belt again. I opened https://bitbucket.org/osrf/ariac/issues/29 to track this
-
- changed status to resolved
-
in Gilbreth (ARIAC spin-off) they found some values that are working for them: https://github.com/swri-robotics/gilbreth/pull/45#issuecomment-370064158 (settled on max step size: 0.00625 - real time update rate: 160)
- Log in to comment
yep, it was changed for performance reasons (https://bitbucket.org/osrf/ariac/pull-requests/11/performance-improvements/diff#comment-23550676)
There could be a value between 0.001 and 0.005 that works well enough for both