An efficient data acquisition methodology for inverse dynamics model learning of manipulator based on analytical method. I

In the last few years, learning inverse dynamic models of manipulators from data has shown considerable successes and become a progressively developing topic in dynamic modeling of manipulators. In this paper, we presented an efficient data acquisition methodology for inverse dynamics model learning. Our method is based around the parametric physical model of a manipulator that obtained from the rigid body dynamics using the analytical method. Our framework consists of Denavit — Hartenberg method for the generation of the manipulator workspace. The received datasets are validated by the results of simulation of kinematic and dynamic modeling of the tested manipulator.