Repeat experiments after change in robot.timeStep?
A bug was discovered in sot-dynamic-pinocchio, which caused the timeStep
variable in the robot
object to be incorrectly set to 0.005 instead of 0.001.
Unfortunately, we were systematically making use of this value.
Normally, the problem in our code has been fixed by commits 1ad5670f and 1aeb5298.
However, gains and time durations now need to be re-tuned. The CoM oscillation trajectory, for instance, is now really slow. I think all it takes is just multiplying by 5 any integrator gains, and dividing by 5 any derivatives.
Once this is settled, should we repeat the experiments just to make sure everything still works?