I have a similar problem. I have built an Isaac Sim simulation with a Universal robot and the RMPflow motion policy. This works quite well so far. Now I am not sure what approaches there are to transfer the RMPflow motion policy to a real universal robot, for example in Python or as ROS block…
We don’t have anything out-of-the-box like that for the UR line at the moment like we do for Franka. What you’d need to do is implement a controller which accepts the RMPflow command stream and translates that to desired velocities for the UR velocity control interface. See this discussion for a discussion of how to use the velocity controller:
This would likely involve translating the desired positions to velocity commands using a law similar to
alpha and beta are both between 0 and 1. The first line nudges the velocity command in the direction of the current position error, while the second blends the desired command toward the desired velocity as a regularizer. You might also need to blend toward zero (regularize back toward zero).
In the end, you’ll likely have to play with the parameters, or the form of the controller. These are well studied problems, so there are some good referenced out there on how to do this. It’d help to find a controls expert to consult with.