Estimation of impedance control parameters with artificial neural networks for variable robotic resistive therapy

The aim of this study is to improve the modeling of physiotherapist behaviors on therapy. In order to contribute to a more consistent therapy of the rehabilitation robots used for lower limb, it was aimed that the rehabilitation applications would be made by considering also patient physical information. At this point, the control algorithm of the therapy by means of impedance control has been extended by evaluation of patient physical information can be grouped as weight and length of patient body in addition to force and position (angle) knowledge. The control algorithm using patient physical information as an input was developed by the method of Artificial Neural Networks (ANN) and the architecture of ANN written as multi-layer perceptron (MLP). Also, back propagation learning method is used to train the ANN. The control algorithm computes the impedance parameters by estimating. The proposed method generated successful results in terms of parameter estimation. The obtained results are sufficient for modeling the movements of physiotherapist.