Validate Force distribution
Hi all, today I've tried to validate the force distribution. The test was supposed to be identical to tests which are already working, except that in addition the force distribution was computed, without even sending the result as feedback. I have no problems with that in simulation.
However, we did not manage to properly validate because the robot was making a terrible noise when executing the movement. We could not trace the origin of this noise, but it seems it was only doing it with the new test, not with the old one. This should be investigated further.
Also, when sending the robot in half sitting, more often than not the second joint of the left arm blocked itself, becoming red in the WebCommander. We could not understand why. We think it might come from the initial "clack" the robot does when activating the position controllers. But it was not doing it before!
@fcaminad and @frisbourg know more about it.
It might just be that the robot was not in good shape and on Monday everything will work.
If you want to test while I'm away:
in python/sot_talos_balance/test
, run
python test_dcmZmpControl_distribute.py
At one point it will ask:
Calibrate force sensors?
Say yes. Then, when it's done, put the robot on the ground. Then it will ask:
Use output of force distribution?
say no at first. Then
Execute a sinusoid?
Say yes. Finally
Raise the foot?
SAY NO! It's not ready yet.
To examine the output of the wrench distribution, in plot
:
rosrun plotjuggler Plotjuggler -l plot_wrenchDistrib.xml
If everything works fine (as it seemed to be), you can repeat the test by saying yes to the question
Use output of force distribution?