Modélisation et Commande des Manipulateurs Mobiles Continuums.

Abstract

This thesis focuses on the modelling and control of mobile continuum manipulators for the assistance of people with reduced mobility, using the Robotino XT robotic platform as a case study. The latter is composed of a continuum manipulator mounted on an omnidirectional mobile platform. In this thesis, we propose an artificial potential field (APF)-interval neuro ou controller type-2 (IT2FNN). This hybridisation aims to exploit the advantages o ered by each of these approaches to design a robust, intelligent, and computationally e cient controller. We also make a comparative study of four learning models, namely the multilayer perceptron (MLP), radial basis functions (RBF), support vector regression (SVR), and co-adaptive neuro-fuzzy inference system (CANFIS) for the geometric modelling of the manipulator part. Finally, the two previous controllers are coordinated by a neural system for the control of the whole Robotino XT. The performance of the difierent controllers was evaluated through a series of real-time simulations and experiments.