Added finger coupling management for iCub new Mk3 hand.

[ale-git]

I've added the coupling class of the new iCub hand under developmente at present.
Since the coupling between the second and first phalanges is not linear, and thus not easily invertible, it makes use of an interpolated look up table that will be also useful in firmware implementation in control boards.

[CLAassistant]

All committers have signed the CLA.

[ale-git]

Ciao @traversaro, can this be merged now?

[traversaro]

Ciao @traversaro, can this be merged now?

Sorry, I had a leftover review comment that was not sent, yes I think we can merge it now.

[traversaro]

Ah, given the recent work on coupling controlboard perhaps @xEnVrE may be also interested in reviewing this.

cc @randaz81

[randaz81]

LGTM 👍
Just for additional clarity, can you add to the description of this PR the following infos (if available):

1. a link to the relevant hand documentation (e.g. its geometry...)
2. a link to the sdf model of the robot which uses the new hand mk3.

[xEnVrE]

I agree with @randaz81. Also, some of the implemented equations are complex and if, in the future, someone wants to make a comparison between the implemented code and the documentation, it would be great if the equations are available somewhere.

[xEnVrE]

I also want to stress the fact that with the actual code of the controlboard, while the decoupling between coupled and controlled joints will be fine, the joints limits used to instantiate the trajectory generator in the coupled-joints space will be wrong as longs as #472 this is not handled. As a result, the overall range of motion of the fingers will be different from the expected.

[ale-git]

Link to geometry documentation:

https://github.com/icub-tech-iit/dev/blob/master/projects/icub-hand-Mk3/calculations/FingerkinV2.docx

Link to the URDF model:

https://github.com/icub-tech-iit/dev/tree/master/projects/icub-hand-Mk3/simulation_model/gazebo/iCubHandMK3_model

[traversaro]

I also want to stress the fact that with the actual code of the controlboard, while the decoupling between coupled and controlled joints will be fine, the joints limits used to instantiate the trajectory generator in the coupled-joints space will be wrong as longs as #472 this is not handled. As a result, the overall range of motion of the fingers will be different from the expected.

I guess a possible workaround for now is to pass really loose limits, to ensure that the trajectory generation is not affected?

[ale-git]

Although scary, the relation is pretty almost linear:

[traversaro]

Link to geometry documentation:

https://github.com/icub-tech-iit/dev/blob/master/projects/icub-hand-Mk3/calculations/FingerkinV2.docx

Link to the URDF model:

https://github.com/icub-tech-iit/dev/tree/master/projects/icub-hand-Mk3/simulation_model/gazebo/iCubHandMK3_model

I am not sure if everyone has access to this private repos. In any case, adding these models to public repos is tracked in robotology/icub-models-generator#130, so eventually we could add the links suggested by @randaz81 .

[traversaro]

Just to avoid confusion, the only comment left is #469 (comment) .

[traversaro]

Thanks @ale-git ! @xEnVrE it is ok if I now merge this PR and then you rebase #499 on the top of the merged devel ?

[xEnVrE]

@xEnVrE it is ok if I now merge this PR and then you rebase #499 on the top of the merged devel ?

Yep!