Authors: Ines Mahmoud; Ayachi Errachdi; Mohamed Benrejeb
Addresses: Laboratoire de Recherche Automatique, Université Tunis El Manar, B.P. 37 Le Belvédère, Tunis 1002, Tunisia ' Laboratoire de Recherche Automatique, Université Tunis El Manar, B.P. 37 Le Belvédère, Tunis 1002, Tunisia ' Laboratoire de Recherche Automatique, Université Tunis El Manar, B.P. 37 Le Belvédère, Tunis 1002, Tunisia
Abstract: This paper focuses on the neural modelling of a system in the presence of disturbances. It proposes an offline and online identification of the inputs of a disturbed mobile robot using artificial neural networks. When meeting strong disturbances, the deviations between the predicted signal-based model and the real signal of the system may become large since usually the disturbances are not measurable and not included in the predictive model. The used disturbance, in this paper, is a random signal added to the output signal. The results of simulation show that the use of the neural networks in the identification of the inputs of the disturbed robot model is very interesting since it enables to guarantee the time competition and the quality of the modelling. The effectiveness of the proposed algorithm applied to the modelling of behaviour of the CHAR robot is verified by simulation experiments with two strategies, online and offline modelling of robot CHAR inputs. Simulation results demonstrate the effectiveness of the proposed control method.
Keywords: inverse models; offline modelling; online modelling; artificial neural networks; ANNs; robot modelling; learning rate; disturbances; mobile robots; simulation; robot control.
International Journal of Computer Applications in Technology, 2017 Vol.55 No.1, pp.39 - 45
Received: 21 May 2015
Accepted: 11 Oct 2015
Published online: 14 Feb 2017 *