couplingXCubHandMk5: relation between coupled angles and finger motor

[PasMarra]

There are three angles ($q_0$, $q_1$ and $q_2$) in Hand Mk5 Coupling Laws that are interdependent, but there is no explicit relation between these angles and the motor controlling the finger.

For this reason, it is not clear to me which one of these angles and how it is affected by the motor rotation, even if here:

ergocub-software/src/modules/couplingXCubHandMk5/CouplingXCubHandMk5.cpp

Line 217 in 67534e3

physJointsPos[2] = evaluateCoupledJoint(physJointsPos[1], "thumb");

it is assumed that: q1 = thumb_prox = thumb_oc.

cc: @pattacini @marcellosavoldi

[pattacini]

cc @Nicogene @fiorisi

[pattacini]

Hi @PasMarra

I've just updated the documentation to indicate where the joint angle encoder is mounted:

• icub-tech-iit/documentation@b67e49f
• icub-tech-iit/documentation@02b004a

Given that, the control loop is closed around $q_1$ and as such you could continue with your control synthesis considering only the nonlinear mapping $q_2=f\left( q_1 \right)$.

That said, during our @icub-tech-iit/board meetup, we also said that it'd be beneficial for possibly improving the control performance to know the nonlinear actuation chain from the motor shaft to $q_0$ (the documentation already explains how to go from $q_0$ to $q_1$).

For this reason, it is not clear to me which one of these angles and how it is affected by the motor rotation, even if here:

The angle is $q_1$. The impact of motor rotation on $q_1$ is eventually nonlinear and not documented so far, but could be neglected to start with given that the control loop is closed around $q_1$.

[pattacini]

That said, during our @icub-tech-iit/board meetup, we also said that it'd be beneficial for possibly improving the control performance to know the nonlinear actuation chain from the motor shaft to $q_0$

cc @fiorisi