UR10 wrist3 collision and visual mesh are 1mm off

[alek5k]

There is a 1mm error in both the UR10 collision and visual meshes in the Z direction relative to the origin. This could cause an issue where if you planned a movement in ROS requiring a tolerance of <1mm, a collision could result.

I could not seem to find any corrections for this 1mm in robot kinematics yaml definition.

Evidence (image dimension scaled by 10x):

[Ruoyu-Jia]

I met similar problem rencently. The precision of existing dae flies is not good due to reduce the size of files. Is there any way to get more accurate urdf?

[fmauch]

Thanks for reporting this. The UR10 meshes are rather old, so that probably exists in the ROS1 description, as well.