As an example, here is a paper I would like to implement: [1].
It shows me control design and goes through the calculation of the
newton-euler dynamics, but if I end up with a 125Hz hight level
"control-loop" e.g. from an UR and can only control with speed-velocity,
am I even able to solve problems like these?

Maybe other people here can help me out by telling me their strategies or even tell me that implementing these things is impossible on a high-level interface.

Hi there,

I have a more general question about robot software. My example(s) might be solved, but my question is how to implement these things myself (to understand it better).

When I find great looking robotics papers that solve a problem I often struggle to put the theory into practice. The literature often solves the problems e.g. by calculating necessary motor torques or by showing the optimal control designs.

My problem when trying to use a real robot to implement these is: The industrial robots (I used) from ABB, UR, etc. don't allow me to do control with joint torques or to mess with (their) low-level control or to even run in a fast-enough control loop.

So my very general question is: for real-world robots, how to I implement solutions from research? Is it even possible if you did not build the robot yourself (and thus don't have low-level access)?

As an example, here is a paper I would like to implement: [1]. It shows me control design and goes through the calculation of the newton-euler dynamics, but if I end up with a 125Hz hight level "control-loop" e.g. from an UR and can only control with speed-velocity, am I even able to solve problems like these?

Maybe other people here can help me out by telling me their strategies or even tell me that implementing these things is impossible on a high-level interface.