System identification and model validation of nonholonomic wheeled mobile robots

This paper focuses on the developing a regression model which describes linear and nonlinear behaviors of the robots. Several techniques of system identification such as AutoRegression Moving Average Exogenous Input (ARMAX), Nonlinear ARMAX (NARMAX), and Least Square (LS) method are used to develop the mentioned model. Recursive LS (RLS) with forgetting factor is employed to illustrate the convergence of the parameters of the models. The advantage of this paper in comparison with other similar works is developing a timedependent linear in parameters model (ARMAX and NARMAX) for this nonlinear robot. Furthermore, these models are functional for practical applications and for designing adaptive controllers. To validate the proposed models for this robot, several statistical tests are implemented and results are compared. Also, R-squared test is used to verify capability of mentioned models. In addition to numerical simulations, we applied our theoretical outcomes to experimental data and results perfectly show the validity of the proposed method.