Time-effective direct dynamics identification of parallel manipulators for model-based feedforward control

摘要

Model-based control of robotic manipulators necessitates accurately parameterized models. This issue is especially important for complex and coupled mechanisms like parallel manipulators. This paper provides a methodology that permits the estimation of model parameters in a time-inexpensive manner. Due to the limited workspace, a bounded and periodic excitation trajectory is advantageous. Only one measurement procedure is necessary for accurate identification, which is carried out by means of the Markov estimator. Rigid-body and friction models are identified and integrated in a feedforward control. The success, accuracy and robustness of the proposed methodology are substantiated with a multitude of experimental results