Employing a Recurrent Linear Model to Guide Mobile Robots Adaptively

Controller adaptation is always a major concern. A controller that meets certain performance design objectives can not be satisfactory unless it can preserve them in the presence of system parameters fluctuation over a wide range of operation conditions. For autonomous robots, a controller that preserves stability and robustness dynamically is especially desirable. This paper presents, therefore, a design of an adaptive controller using recurrent linear model (RLM) to guide a robot driven by multi-references. Moreover, it shows how the artificial neural networks (ANN) span from their classical pattern recognition to the adaptive behaviour learning. The main objective of that is to maintain the stability and to reinforce the robustness of the overall control system by which the transition among different set points will be smoothed. The RLM is systematic, methodical and easy to design.