robotology
Make sure ARE is working in Gazebo simulation
Several users are experiencing issues that (I think) are related to not properly tuned joint-level PID position loops of the iCub model in Gazebo, see:
- https://github.com/robotology/QA/issues/155
- https://github.com/robotology/icub-gazebo/issues/37
- https://github.com/robotology/icub-gazebo/issues/35
- https://github.com/robotology/icub-gazebo/issues/31
We should tune the position gains with respect to some clearly defined scenario, such as the ARE one described by @frank-foerster in https://github.com/robotology/icub-gazebo/issues/37 . For reference we should document on which scenario we tuned the gains, such that in the future we can simplify the process of re-tuning the gains for different robots (such as R1).
There's a bunch of bugs that prevent ARE from working in gazebo. I'll remove this message when the fixes will be pushed.
Is this still the case @pattacini ? In any case, I think the proper place for this issue is icub-models.
Most of the basic components of ARE (Cartesian Controller, Gaze Controller, hand controller) are working nicely now as demonstrated in https://github.com/robotology/icub-gazebo-grasping-sandbox.
Never tested ARE per se though.
