Replicating TUG in gazebo

[vvasco]

This PR replicates the TUG in gazebo.
We can handle:

• the verbal interaction between the robot and the human model:
  • when pressing the wifi button, the human model / the robot (depending which one is moving) stops and a question can be asked;
  • a question is simulated through the question-simulator.sh script. We can choose 4 keywords, specifically speed,repetition,aid,feedback, each associated with two or three questions (as defined here). The script randomly selects a question associated to the chosen keyword and sends it to googleSpeechProcess, which analyses the sentence;
  • when googleSpeechProcess provides the output to managerTUG, the human model / the robot starts again.
• following the human model through navigation: when the TUG starts, the robot starts tracking the skeleton by moving its base;
• synchronization between robot commands and human animations: the human model seats when the robot asks to seat and the human model stands up, reaches the target, goes back and seats again when the robot gives the start;
• TUG failures:
  • we can provide a target to reach different than the finish line: the robot says that the test was not passed;
  • the human model can be paused during the TUG: the robot encourages the human model to finish the test, reminding to push the button to ask questions. If the human model does not continue the test, the test is not passed.

The following videos show the described functionalities (click on the images to open the videos):

• Successful TUG and questions:

• Target not reached:

• Human model stopping:

[pattacini]

OMG, that's simply awesome! ❤️

To be definitely merged after CI completes (keep an eye at Codacy and BCH failures if they're related) 🚀

For crafting the corresponding documentation, I would consider remaking the videos with less background noise and more machines in order to speed up the avatar's movements. This way, we would be able to bundle that material as a showcase to FDG. Also, R1's home position could be pushed closer to the chair, what do you think?

Again, superb ✨

[vvasco]

Having more machines would definitely help. For the documentation I will remake the videos as you suggested.
Regarding R1's position, I put it a bit far from the chair to be sure that the skeleton remains in the FOV when moving closer to the robot (I have to solve the issue with the skeletonLocker, but once done I'll move it closer).

I'll have a look to the CI failures.