Identification of friction and rigid-body dynamics of parallel kinematic structures for model-based control

Until now, the experimental identification of the dynamics of parallel robots is restricted to simple models in combination with adaptive control algorithms. This gap is closed by a new approach presented in this paper, which is suited for even complex parallel kinematic structures. The approach consists of two steps and utilizes simple point-to-point (PTP) motions that lead to a separation of friction and rigid-body dynamics. In the first step, local models are determined for a lot of different configurations, i.e. end-effector positions. In the second step, the overall friction model and the overall rigid-body model are calculated from the local models by linear Least-Squares estimators. The use of linear estimators is based on a formulation of the dynamic equations, which is linear with respect to a dynamic parameter vector of minimal dimension. This formulation is automatically obtained by an algorithm, which utilizes Jourdain's principle of virtual power. The experimental application of the identified model to model-based feedforward control of the innovative hexapod PaLiDA, which has been developed by the Institute of Production Engineering and Machine Tools of the University of Hannover, proves the capability and efficiency of the presented algorithms.