Closed proyan closed 7 years ago
This should close #12
I moved the IMU filtering as well because the torsoAcceleration and the torsoAngularVelocity signals were also moved. The main idea was to keep the f/t and current estimates in the previous entity, and move the others which are independent of f/t and are useful in Pyrene. Do we not need IMU filtering in Pyrene?
The filtered IMU values are only used to estimate the joint torques, so we won't need that in Pyrene. I would leave everything in the F/T estimator, except for the numerical derivatives of the encoders. The additional advantage of doing that is that the new entity is general purpose and it can be used to compute the derivatives of any arbitrary signal. I suggest calling it something like "NumericalDifference", which should take as input a signal x, and compute as output its first and second derivatives dx, and ddx, together with a smoothed version of the input x_smooth. Later we could implement different methods for numerical derivatives in this entity.
The last commit should take care of the issue
Which python scripts do you mean?
I mean create_entity_utils mainly. There may be others that do not come to my mind in this moment.
@andreadelprete I am ready to make a new pull request for #13 after you merge this
Everything looks good to me. Did you test it on the robot? If not and you don't know how to do it you can find the instructions I wrote here: https://wiki.laas.fr/robots/SoT-14-04
For the missing information (such as passwords) ask @thomasfla.
It works on the robot. I ran the run_sot_fast_da to check and the robot is still alive :laughing: I just mailed you the std output of the command
Takes out kinematic signals from the force-torque-estimator and puts them in a new entity vel-acc-estimator
force-torque-estimator is dependent on the angular velocity and acceleration signals from vel-acc-estimator