search

open-dynamic-robot-initiative / robot_properties_solo

BSD 3-Clause "New" or "Revised" License
40 stars 20 forks source link

Undue collision due to URDF meshes placement #19

Closed thibnoel closed 4 years ago

thibnoel commented 4 years ago

Hello, I recently started working on the SOLO-12 model, especially on the characterization of the collisions. For now I only looked at the collision pair between the front left leg and the body (base link). I'm working with the FCL library to check the distance between the meshes in order to build a collision map in the shoulder's rotations space. My problem is that there is a subspace of the configuration space where the collision looks wrong (see attached pictures) : FCL detects a collision when there should not be one on the real robot in the same configuration.

problematic_collision_view0 problematic_collision_view1

Here as you can see, the collision that is detected is between the edge of the motor and the edge of the base link. This problematic configuration is not unique and the subspace where it happens can be (I hope) better seen here, along with a more complete view of the collision map for the whole configuration space collision_config_space zoom_problematic_config

PS : please tell me if this issue doesnt fit the standards for this repo

fgrimminger commented 4 years ago

Hi - thanks for pointing this out. There was a mistake in one of the dimensions of the simplified robot base model. This should be fixed now. Best. Felix

https://github.com/open-dynamic-robot-initiative/open_robot_actuator_hardware/blob/master/mechanics/quadruped_robot_12dof_v1/stl_files_for_visualization/solo_12_base.STL

jviereck commented 4 years ago

Hi folks, thanks @thibnoel for reporting and @fgrimminger for providing an updated STL!

@thibnoel: Can you try to import the fixed solo_12_base.STL file from the open_robot_actuator_hardware repo into the robot_properties_solo repo and check if the collisions disappear?

Best, Julian

thibnoel commented 4 years ago

Hello, thanks for the quick answer! I tried the updated stl model for the base and it looks like the fix is working perfectly, the faulty collision is gone (figure attached) fixed_problematic_collision_view0

Thanks a lot again Best, Thibault

thomasfla commented 4 years ago

Thanks @fgrimminger ! @thibnoel I don't think this issue should be closed yet since we have to update the STL here first.. Could you do a PR if you validated the new stl?

thibnoel commented 4 years ago

Yes you're right, I submitted the PR

jviereck commented 4 years ago

Thanks @thibnoel .