ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Using IterativeSplineParameterization with computeCartesianPath

asked 2019-06-21 00:25:22 -0500

mjs gravatar image

updated 2019-06-22 14:27:28 -0500

gvdhoorn gravatar image

Hello, I've been trying to use the recently merged in IterativeSplineParameterization (https://github.com/ros-planning/movei...) trajectory processer without having to install moveit from source but I haven't been able to find a config file that allows me to do so. Has anyone used this algorithm without installing from source ?

edit retag flag offensive close merge delete

1 Answer

Sort by » oldest newest most voted
3

answered 2019-06-21 05:51:01 -0500

gvdhoorn gravatar image

updated 2019-06-22 01:45:13 -0500

Edit:

Your comment completely changes your question (please include all relevant information in your first post in the future), so here is an answer to the new question.

I’m specifically looking to use the iterative spline parametrization for Cartesian planning. Would updating this config file change the algorithm used by both freespace and cartesian plans ?

No, it doesn't, as the default implementation of computeCartesianPath(..) hard-codes the use of the IterativeParabolicTimeParameterization right now (here).

I would change the object instantiated here https://github.com/ros-planning/movei...

yes, you would have to do that.

You could potentially avoid building all of MoveIt from source by extracting the move_group package from the source tree, adapting the adapter and then building that.

Even better: create a copy of the adapter, place it in its own package and just build that.

Best: make this configurable (using a parameter), test it and submit a PR against ros-planning/moveit and get it merged.


There should be no need to build MoveIt from source (I'm also not entirely sure what you'd be changing specifically in that case).

The time parameterisation is done in a MoveIt Planning Request Adapter.

There are three of those, corresponding to the three algorithms currently available:

  • AddTimeParameterization (which applies the Iterative Parabolic Time Parameterization)
  • AddIterativeSplineParameterization (the spline based approach: Iterative Spline Parameterization)
  • AddTimeOptimalParameterization (using the Time-optimal Trajectory Generation library by Tobias Kunz)

So changing the algorithm used is a matter of changing the configuration of the planning request adapters.

For MoveIt configurations generated by the MoveIt Setup Assistant, that can be done by editing the ompl_planning_pipeline.launch.xml file.

Specifically, change this line and replace AddTimeParameterization with AddIterativeSplineParameterization.

Refer to the Time Parameterization page in the MoveIt documentation for some more info on this.

And refer to the planning_request_adapters_plugin_description.xml file for a list of available time parameterisation planning request adapters.

edit flag offensive delete link more

Comments

Thank you! I’m specifically looking to use the iterative spline parametrization for Cartesian planning. Would updating this config file change the algorithm used by both freespace and cartesian plans ?

As for building from source, I would change the object instantiated here https://github.com/ros-planning/movei...

mjs gravatar image mjs  ( 2019-06-21 10:56:56 -0500 )edit
1

I’m specifically looking to use the iterative spline parametrization for Cartesian planning.

Your original question did not mention this. Could you please add these sort of details to your question in the future?

gvdhoorn gravatar image gvdhoorn  ( 2019-06-22 01:39:03 -0500 )edit

Yes sorry about that, I will make sure to do so in the future

mjs gravatar image mjs  ( 2019-06-22 20:31:25 -0500 )edit

I think this answer should be included in the tutorial

Xinjue gravatar image Xinjue  ( 2023-02-21 04:07:38 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2019-06-21 00:24:52 -0500

Seen: 896 times

Last updated: Jun 22 '19