Identification of Differentially Driven Wheeled Mobile Robot using Neural Networks

In the field of control theory system identification is an important way of investigating and understanding the system. Several distinct approaches of system identification produce the different form of knowledge about the system. In system identification, neural network plays an important role. A novel technique has been considered to identify the nonlinear multi input multi output (MIMO) crossed coupled system using neural network. A very interesting mechatronics platform, namely, wheeled mobile robot, is a highly nonlinear cross coupled dynamic system and pose challenges for many classical control techniques, has been considered as a candidate system to illustrate the proposed neural networks based system identification. Static and dynamic back propagation methods are used for system identification. This paper verifies the ability of the neural networks and its applicability for the system identification of practical nonlinear systems.