Control of robots with elastic joints based on automatic generation of inverse dynamics models

The paper shows how the tool vibrations of elastic joint robots can be reduced significantly by feedforward control based on inverse dynamics models. The key ideas are: (1) to use available algorithms for differential-algebraic equation systems to derive inverse plant models automatically and reliably, (2) to automatically construct the high-order differentiability of the inverse plant model inputs by prefilters, and (3) to use a controller with two structural degrees of freedom. The technique is demonstrated with a 6 axes elastic joint robot model and with the experimental results from our laboratory robot.