Disable specific collisions

👍 Gathering votes

Is it possible to disable collision checks between desired links (robot links and or environment obstacles)?
I saw the soft collision check option, but it can sometimes be useful to completely disable collisions between specific pairs of links/obstacles.

Thanks

Alban Laflaquiere

11 months ago

Activity
Yahav

Hi Alban,
Would it be sufficient to ignore collision checking with the end effector? Alternatively, could you describe a use case where you would want to disable collision checking with specific robot joints?

0    8 months ago    Reply

Comment must be at least 20 characters.
Cancel

We appreciate your feedback.

Alban Laflaquiere

Hi Yahav,
I think disabling collision with the end-effector would cover most use cases I can think about for now.
Thanks

0    7 months ago    Reply

Comment must be at least 20 characters.
Cancel

We appreciate your feedback.

Yahav

@Alban Laflaquiere: could you share real wolrd use cases to motivate this feature? it could help us figure out how to enable it on the platform

0    6 months ago    Reply

Comment must be at least 20 characters.
Cancel

We appreciate your feedback.

Alban Laflaquiere

@Yahav: I guess the most intuitive one would be when the robot is moving very close to (or even in) an obstacle to do some precise manipulation. In that context, the collision shapes of the gripper and obstacles would be too coarse, or maybe it's not even possible to properly model the actual obstacle in the real world (for instance it could partly non-rigid), and disabling (temporarily?) the collisions between the gripper and obstacle would allow us to do the manipulation we want to plan.

0    6 months ago    Reply

Comment must be at least 20 characters.
Cancel

We appreciate your feedback.

Lars Berscheid changed status to 👍 Gathering votes

11 months ago

Lars Berscheid

This is currently not possible, but sounds like a useful feature!

0    11 months ago    Reply

Comment must be at least 20 characters.
Cancel

We appreciate your feedback.

2 votes