wholeBodyDynamicsTree about codyco-modules HOT 4 CLOSED

To add on what Silvio wrote, the idea is also to provide a second implementation of the wholeBodyState interface for iCub that, rather than computing the estimations, read them from the Yarp ports of wholeBodyDynamicsTree. This allows the user to choose between two scenarios:

• the estimation are computed on the same machine on which the control code is running (using the first implementation of icubWholeBodyStates)
• the estimations are computed by wholeBodyDynamicsTree (which may run on a different machine) and the controller reads them through Yarp ports (using the second implementation of icubWholeBodyStates)

Since these two scenarios differ only for which implementation of the wholeBodyState is used, switching from one to the other is painless (i.e. just a matter of changing one line of code). The user choice will of course depend on the computational burden of the estimation process and the computational power of the computer. In general, if the computer is powerful enough, it is better to perform the estimation on the same machine so that we avoid delays introduced by the network communication.

from codyco-modules.

cc @serena-ivaldi

from codyco-modules.

Initial work is the wholeBodyDynamicsTree branch: https://github.com/robotology/codyco/tree/wholeBodyDynamicsTree

from codyco-modules.

wholeBodyDynamicsTree is correctly working at the moment, so I think this issue can be considered solved.

from codyco-modules.