Some of our realtime types use float values/vectors/matrices and some others are based on double.
Is there any reason for that? Otherwise we should unify the types...
The following types are based on double:
dynamics/Torques
dynamics/Forces
dynamics/Wrench
geometry/Rotation
geometry/Translation
The following types are based on float:
robot/Weights.hpp
robot/JointState.hpp
kinematics/JointAccelerations.hpp
kinematics/JointVelocities.hpp
kinematics/JointJerks.hpp
kinematics/JointAngles.hpp
dynamics/JointTorques.hpp
dynamics/JointImpedance.hpp
The KDL-library uses double vectors. At the moment we have to convert our float-based jointangles to double before employing KDL and afterwards we have to convert it back.
This wastes lots of time...
However, it seems that we can not log double-based types via the OCL::FileReporting component (I am not completely sure about that, but I had an issue with logging the dynamics/Wrench type).
I have the same issue with casting. From float->double->float seems pointless to me. At least there should be templating option for tst-rt types for those who do not care about OCL::FileReporting (e.g., me!)
Some of our realtime types use float values/vectors/matrices and some others are based on double. Is there any reason for that? Otherwise we should unify the types...
The following types are based on double:
The following types are based on float:
The KDL-library uses double vectors. At the moment we have to convert our float-based jointangles to double before employing KDL and afterwards we have to convert it back. This wastes lots of time... However, it seems that we can not log double-based types via the OCL::FileReporting component (I am not completely sure about that, but I had an issue with logging the dynamics/Wrench type).