Identification of Friction and Rigid-Body Dynamics of Parallel Kinematic Structures for Model-Based Control

Abstract

Until now, the experimental identification of the dynamics of parallelrobots is restricted to simple models in combination with adaptivecontrol algorithms. This gap is closed by a new approach presented inthis paper, which is suited for even complex parallel kinematicstructures. The approach consists of two steps and utilizes simplepoint-to-point (PTP) motions that lead to a separation of friction andrigid-body dynamics. In the first step, local models are determined fora lot of different configurations, i.e. end-effector positions. In thesecond step, the overall friction model and the overall rigid-bodymodel are calculated from the local models by linear Least-Squaresestimators. The use of linear estimators is based on a formulation ofthe dynamic equations, which is linear with respect to a dynamicparameter vector of minimal dimension. This formulation is automaticallyobtained by an algorithm, which utilizes Jourdain's principle of virtualpower. The experimental application of the identified model tomodel-based feedforward control of the innovative hexapod PaLiDA, whichhas been developed by the Institute of Production Engineering andMachine Tools of the University of Hannover, proves the capability andefficiency of the presented algorithms.