sometimes we do have the problem that, when we place the cylinder in the target zone, it oscillates for ever after the gripper opened. After some time, this causes the cylinder to leave its target zone. We haven't found a way to prevent that from happening yet.
this late in the challenge we can hardly change anything on the simulation parameters.
i can only suggest to carefully place the cylinder, bringing it into contact with the table (do not let it drop from above), hold it still, (very) slowly open the gripper, and then move the gripper away without touching the object again.
there will be a new euroc-c2s1-scenes_1.0.20_i386.deb package with considerably higher rolling friction for all cylinders. this will not make the simulation more realistic (rather the opposite) but should stop the oscillating cylinder driftig on the table.
please wait for the announcemant via the "EuRoC C2S1 Software Versions" thread.
Is it possible that due to this changes, the friction increased in all tasks?
Especially concerning task 5: The forces which are necessary to push a part have increased dramatically. Sometimes it is more likely that the gripper of the LWR is forced to tilt instead of a pushed part starting to slide.
red wrote:Is it possible that due to this changes, the friction increased in all tasks?
the euroc-c2s1-scenes_1.0.25_i386.deb and euroc-c2s1-scenes_1.0.20_i386.deb only changed all the cylinders to use surface_material "aluminium2" instead of "aluminium". no other part/object is using the new "aluminium2". no other surface_material was changed.
i will check which foces are needed to push parts when they lie on the fixute before euroc-c2s1-scenes_1.0.20_i386.deb and after euroc-c2s1-scenes_1.0.25_i386.deb and will come back with results.
i've checked euroc-c2s1-scenes (1.0.13) from 20 Oct 2014 and euroc-c2s1-scenes (1.0.25) from 04 Nov 2014.
i've testes task5_v1 & task5_v2 multiple part on the puzzle fixture. (parts are surface material "aluminium" while the puzzle fixture is "wood" - this combination is defined to have a friction coefficient of 0.6 in the yaml fieles)
i tested which force (parallel to fixture plane) is needed to move the parts. depending on their mass i always observed a friction coefficient very close to 0.6 like expected.
the wood-aluminium combination has since the beginning this friction settings: friction: 0.6, rolling_friction: 0.0006
if you are experiencing a friction coefficient other that ~0.6 than you probably also introduce torques and forces which are not parallel to the surface...