Using the algebraic structure of articulated robot dynamics in control design

We discuss the algebraic structure of the dynamics of articulated multibody systems and how it may be effectively used in control design. The dynamics has a decomposition consisting of matrix operators satisfying many identities and giving insight into the dynamical system. To show the utility of this formalism, we investigate a disturbance attenuation control problem. We introduce a meaningful performance index which has an interesting interpretation in terms of the `normalized´ robot dynamics. Finally, we derive a new adaptive controller using the local structure found in the dynamics and test it on a 3-link planar robot arm