search

robotology / whole-body-controllers

Simulink-based whole body controllers for humanoid robots.
123 stars 42 forks source link

Investigate if it makes sense to port Friction Exploitation control in wbc #73

Closed gabrielenava closed 5 years ago

gabrielenava commented 5 years ago

I would like to port the friction exploitation controller in wbc, but before it is necessary to understand the ratio between effort and benefits.

gabrielenava commented 5 years ago

I think the answer is to the issue title is NO for the following reasons:

1) it is a huge work of porting and update, for a controller that is not effectively used on the real iCub; 2) the stability of the zero dynamics of the proposed controller has never been verified. I remember the momentum control was defining a stable closed loop momentum dynamics, but then additional terms appear in the postural task and the effect has not been investigated that much. 3) the controller can be tested on the robot, but requires to modify the low-level controller parameters. This means every time is used we have to remove friction compensation from the low-lever controller configuration files. Otherwise, we may break the robot. 4) current control has recently been implemented. At this point, it makes sense to wait until whole-body-current-control is available, and add a new controller to the repo that also implements friction exploitation in a cleaner framework.