Inconsistency in using derivatives of ZYX Euler angle vs. angular velocity in rnea() and dccrba() of Pinocchio

[YangLiu-tum]

Hi ocs2 team,

Really appreciate for the efforts and work in creating and sharing this valuable repo with the robotics community! It has been a great resource for many developers and researchers!

Recently I have been reading and learning the code of ocs2. And I've just come across an tiny inconsistency in the way the angular velocities are represented in the code. Particularly when calculating the inverse dynamics per rnea() and the derivative of the centroidal momentum matrix Adot using Pinocchio library functions.

ocs2/ocs2_pinocchio/ocs2_centroidal_model/src/CentroidalModelRbdConversions.cpp

Line 225 in ebde452

return pinocchio::rnea(model, data, qDesired, vDesired, aAugmented, fextDesired);

ocs2/ocs2_pinocchio/ocs2_centroidal_model/src/CentroidalModelRbdConversions.cpp

Line 84 in ebde452

const auto Adot = pinocchio::dccrba(model, data, qPinocchio, vPinocchio);

In computing the Adot, the vPinocchio is used as function input. The vPinocchio is defined in as follows, where base_orientation_zyx_derivatives is used to represent the change of base orientation.

ocs2/ocs2_pinocchio/ocs2_centroidal_model/include/ocs2_centroidal_model/CentroidalModelPinocchioMapping.h

Line 58 in ebde452

* Pinocchio Joint Velocities: vPinocchio = [ base_linear_velocity, base_orientation_zyx_derivatives, joint_velocities ]'

However, when computing the rnea(), the vDesired contains the base angular velocities expressed in the world frame, since it consists of the output from computeBaseKinematicsFromCentroidalModel().

ocs2/ocs2_pinocchio/ocs2_centroidal_model/src/CentroidalModelRbdConversions.cpp

Line 186 in ebde452

vDesired << desiredBaseVelocity, centroidal_model::getJointVelocities(desiredInput, info);

ocs2/ocs2_pinocchio/ocs2_centroidal_model/src/CentroidalModelRbdConversions.cpp

Line 82 in ebde452

baseVelocity.tail<3>() = getGlobalAngularVelocityFromEulerAnglesZyxDerivatives<scalar_t>(baseOrientation, derivativeEulerAnglesZyx);

I'm curious as to the convention used here. Is there something crucial I'm missing or some specific requirements related to the Pinocchio library?