[iCubGazeboV3] Ankles roll goes in HF for a given value of ankle pitch about icub-models

[GiulioRomualdi]

Hi @traversaro changing the max_step_size from 0,001 to 0.0001 fixes the problem

[Nicogene]

Probably we have to fake inertias also here

[traversaro]

The HF in the simulated robot is raised whenever the requested torque is higher then 2000 Nm, see https://github.com/robotology/gazebo-yarp-plugins/blob/52491d69365a32ad930f2b858b9f3a43e7596510/plugins/controlboard/src/ControlBoardDriver.cpp#L69 and https://github.com/robotology/gazebo-yarp-plugins/blob/52491d69365a32ad930f2b858b9f3a43e7596510/plugins/controlboard/src/ControlBoardDriver.cpp#L568 . What is happened there at the first glance seems related to some numerical error that leads the simulated system to diverge once controlled, i.e. a problem similar to #53 . Does the problem disappears if you increase some inertias of some links of the foot, or if similarly you decrease the simulation timestep?

[Nicogene]

@GiulioRomualdi it should be fine now, can you please double-check?