Static GUI for Simulink and add gearboxes inertia

[gabrielenava]

Following the issues:

• Simulink takes a lot of time in some iterations wb-toolbox#160

• Check if the Simulink required integration step is respected  #38

• Update all models to launch Simulink from command line #39

I created a static gui that allows to handle a Simulink model (run and stop) without the need of opening the Simulink model interface. This is sufficient to avoid the occasional blocking of Simulink while running, that seems to be related to the real-time update of the Simulink GUI. This new GUI is instead static, i.e. it is never updated while the Simulink model is running.

Also, following #49 I added the gearboxes inertia to the motors reflected inertia calculation.

TO DO:

• add info to understand if the simulink model is still compiling;
• close the model when the user closes the GUI;
• add a warning message if the required integration step is not respected.

[gabrielenava]

The two types of harmonic drives listed in the wiki page are the CSD-14 and CSD-17. Their inertia is indeed different:

I think it is better to ask if there is a table mapping the type of harmonic drive with the associated joint on the robot. @fiorisi do you know where we can retrieve this information?

[gabrielenava]

I asked the mechanics about a way of knowing which gearbox is associated with which joint. They told me one way is to check the CAD model (long) or to look is someone already documented this info somewhere. However, they also told me what follows:

• there is a third type of harmonic drive that is not included in the wiki. It is used in the forearm/neck joints, so it is not of our interest at the moment.

• there is a 1:1 correspondence between motor and harmonic drive, i.e. the CSD-14 type is used only with one type of motor and the CSD-17 is used only with another type of motor.

• there are only 2 type of motors and 2 types of harmonic drives used for the robot shoulder, elbow, legs and torso joints.

Based on these information, I concluded that it was sufficient to check the elbow and hip joints in the CAD, that have two different type of motors (the MOOG equivalent of the KOLLMORGEN RBE-01210 and the MOOG equivalent of the KOLLMORGEN RBE-01211, respectively), and check which harmonic drive is associated with which motor. Note that differently from what is written in the wiki we currently mount MOOG motors on the iCub, not KOLLMORGEN.

In particular, according to my reasoning it seems that the map is the following:

• torso yaw, shoulder and elbow ---> CSD-14

• legs, torso pitch and roll ---> CSD-17

I will apply this logic for the moment. @fiorisi can you later confirm that this mapping is correct?

cc @FabioBergonti

[gabrielenava]

Both the GUI and the gearbox inertia have been tested on iCubGenova04 for performing the Yoga ++ demo. The demo was performed as usual, and there were no qualitative differences. Before merging, I would like to fix a couple of issue concerning the GUI.

[gabrielenava]

All possible tests in simulation have been performed. We need to test this PR on the real iCub before merging.

[gabrielenava]

The PR passed the tests on iCubGenova04. I will anyways wait for a testing period before merging.

[gabrielenava]

I created a release of the current master branch. The controllers have been successfully tested more than one time on iCubGenova04, so I think it is time to merge this PR.