iCubGazeboV3 model feet contacts are unstable in Gazebo v11.13 #209
Comments
|
Actually, I did more tests, and I found that probably updating Gazebo version is not the cause of this issue. When I did my test with the latest Gazebo version, I used a clean Conda environment that also installed the latest of The reason for this issue seems to be updating Which most likely is due to robotology/icub-models-generator#243 @gabrielenava you can try to downgrade |
Fix robotology/icub-models#209 . In #243 the location of the frames of the links of the soles ( with names like `(r|l)_foot_(front|back)` ) have been moved to the origin of the FT sensors that connect them with the `(r|l)_ankle_2` link. All the quantities expressed in this frame have been automatically modified to account for its new location by the simmechanics-to-urdf script, except for the location of the assigned collision, that are hardcoded in the .yaml file . This created a regression, has it effectly moved the sole assigned collisions up, so the robot was actually touching the ground with the (r|l)_ankle_2 links, that do not have any contact parameter assigned. This commit fixes the situation by taking the existing z value of the assigned collision origin, and adding the offset with which the other elements (visual, inertial) have been modified (see robotology/icub-models@799ae07#diff-9d811d6ae4a80f936a2466cdc3b852ac4fb6d6e3a4cd40d2c42e794f42d6415aL260), to obtain: newCollisionZ = oldCollisionZ + (newVisualZ - oldVisualZ) = 0.003 + (0.9241470000000105 - 0.9464470000000105 ) = -0.0193
|
Fix proposed in robotology/icub-models-generator#251 . |
|
@gabrielenava @HosameldinMohamed can you try with the latest devel, i.e. 78b150c ? If it works, I can do a release. |
|
For me it works from the
|
|
looks fine also for me! contacts are stable and the robot does not drift anymore!
|
|
Fixed released in https://github.com/robotology/icub-models/releases/tag/v2.4.0 . |
I have Gazebo v11.13. after dropping the
iCubGazeboV3model in the simulator, the robot starts to drift on the floor and the feet contacts are very unstable:vokoscreen-2023-09-12_17-44-52.mp4
this is not occurring in @HosameldinMohamed (previous) configuration, with Gazebo 11.12 (see also robotology/whole-body-estimators#169 (comment)):
Screencast_20230912_235556.mp4
after he also updated to Gazebo 11.13, the same issue appeared on his laptop.
cc @traversaro
The text was updated successfully, but these errors were encountered: