ROS Resources: Documentation | Support | Discussion Forum | Service Status | Q&A answers.ros.org

MoveIt Maintainer Meeting - Feb 27th

The next Maintainer Meeting will be February 27th at 8am PST

Your time zone: 2020-02-27T16:00:00Z

Agenda:

Please request here any agenda topics you would like included.

Meeting ID
meet.google.com/fpo-srqg-feg

Phone Numbers
(‪US‬) ‪+1 347-486-5750‬
PIN: ‪644 991 049#‬

Country-specific phone numbers are available.

6 Likes

Thanks for the invitation. I think we will be done with the technical parts of merging our pilz planner into moveit (https://github.com/ros-planning/moveit/pull/1893) and can then discuss further action points.

Decide on reasonable default scaling factors for the new feature: https://github.com/ros-planning/moveit/pull/1890/files#r380817565

  • Execution speed should be low enough to give beginners time to hit the emergency stop if, e.g., IK yielded an inadequate joint solution
  • The feature will affect all currently running robot demos that do not explicitly set the scaling factors to 1.0 beforehand. This is likely quite a lot.
    Are we ok with setting the default to very low values and require people to explicitly specify 1.0 in their joint_limits.yaml if they want acceptable behavior again?
  • 0.1, 0.2, and 0.5 have been proposed in the thread
1 Like

Regarding the default scaling factors, I’ll quote myself from the Github discussion:

I tested a few parameter values with our UR5 to check which speeds would still be surprising a user in this type of situation.

  • 0.5 is barely different from full speed for small motions
  • 0.2 seemed to be still pretty fast
  • 0.1 seemed safe

0.1 for a UR5 is about 10 cm/s in most configurations, which I think is a good compromise.

I also explained my thoughts over there. I probably won’t make it to the meeting, so I’d be happy if someone else could champion safe defaults.

1 Like

Also, I agree with Robert’s proposal about the code coverage checks. The current setting leads to a lot of “not mergeable” flags on benign PRs.

Examples of affected PRs:

  • Here, the tests are part of a separate PR.
  • Here, convenience changes are made to the setup assistant, and no runtime code is affected.
  • Here, the default scaling factors discussed in the post above are introduced. How do you test for this? Either way, the change does not seem dangerous.

From a developer perspective, it is confusing not to see at first glance if there is a functional issue with the code or if there is a housekeeping issue. It seems clear that the severity is not the same. It would be great if that could be represented somehow, and the code coverage feedback given differently.

As pointed out in https://github.com/ros-planning/moveit/issues/130#issuecomment-589544315 and https://github.com/ros-planning/moveit/pull/1909, there is an inconsistency in how MoveIt and rviz handle empty collision geometries. While MoveIt copies visual geometries over to collision geometries, rviz doesn’t display collision geometries in this case. We should discuss the different options to resolve this inconsistency offered in https://github.com/ros-planning/moveit/issues/130.

1 Like

The minutes of the meeting:

Participants: Audrow Nash, Christian Hankel, collaed 11, Dave Coleman, Henning Kayser, Jafar Abdi, Jere Liukkonen, Joachim (Pilz), Mark Moll, Michael Gorner, Mike Lautman, Omid Heidari, Robert Haschke, Tyler Weaver

  • Becoming official ROS 2 working group (Dave Coleman)
  • Code Coverage checking (Robert Haschke, Tyler Weaver)
  • MoveIt and Rviz handles empty collision geometries differently (Michael Gorner, Robert Haschke)
  • MoveIt 2 release recap (Henning Kayser)
    • MoveIt 2 Beta Released! summarize it
    • Should we have eloquent-devel branch .? (Dave Coleman)
      • Prefer creating the branch once we have a full release (Henning Kayser)
      • As soon as the binary release is done we should have it (Robert Haschke)
  • Discussion about a bug in collision checking
  • Pilz planner integration (Christian Hankel, Joachim (Pilz))
    • Almost done in terms of technical integration
    • Might be one thing to explicitly state somewhere is this’s not an actual planner but trajectory generator (Robert Haschke)
    • Technically It’s an incomplete planner because it doesn’t return a plan with collision (it do a collision checking) (Michael Gorner)
  • Default scaling factor (Michael Gorner)
2 Likes

Would Acutronic moveit2 master branch supported on eloquent? I had encounter lot of dependency errors while building it.