Revision history [back]

I am pretty sure this is precisly the problem @RobertWilbrandt explained:

This is caused by "jumps" in the IK, where each sampled point is individually fine, but the interpolation between them leads to these "flips".

Based on my limited experience with orientation constraint planning I think using enforce_joint_model_state_space and increasing the planning time (which you both already mentioned) is your best option at the moment in MoveIt.

There is also the planning with a constraint database option if your setup stays the same and you want faster planning, but I've not tested this myself yet.

The position only IK solution @RobertWilbrandt proposed is interesting, I'm not even sure it was intended to solve this problem, but it certainly could help a lot depending on the implementation details inside MoveIt.

In a couple of months, you will hopefully have another option when this pull request and it's little brother for orientation constraints gets merged.

I am pretty sure this is precisly precisely the problem @RobertWilbrandt explained:

This is caused by "jumps" in the IK, where each sampled point is individually fine, but the interpolation between them leads to these "flips".

Based on my limited experience with orientation constraint planning I think using enforce_joint_model_state_space and increasing the planning time (which you both already mentioned) is your best option at the moment in MoveIt.

There is also the planning with a constraint database option if your setup stays the same and you want faster planning, but I've not tested this myself yet.

The position only IK solution @RobertWilbrandt proposed is interesting, I'm not even sure it was intended to solve this problem, but it certainly could help a lot depending on the implementation details inside MoveIt.

In a couple of months, you will hopefully have another option when this pull request and it's little brother for orientation constraints gets merged.