Controle preditivo aplicado ao seguimento de trajetória de robô móvel com rodas

Abstract

This work proposes a study and application of advanced controller to trajectory tracking of wheeled mobile robots. This kind of problem is a challenger for controllers once its models has two inputs and three outputs and is a non-linear model. In the literature there are various solutions to wheeled mobile robots trajectory tracking, among them the Model Predictive Control (MPC) with linearization model and a non linear control which in this work will be nominated as Klancar Controller. The Predictive Controllers can be applied efficiently in plants which has multiple inputs an multiple outputs, in situation that a future reference trajectory is known and systems with input and output constraints . However, the main disadvantage of MPC is the high computational effort which limits its practical application. Thus, this specific controller uses the plants linearization model. On the other hand, the Klancar Controller may be more efficient than the ones based on linear models, once the model is non linear. However, its solution, by definition, does not match the optimized criteria which can be a disadvantage mainly in systems that has constrains and a known future reference. Furthermore, this work proposes the application of the Predictive Control Extended Prediction Self Adaptive Control (EPSAC) to wheeled mobile robot trajectory tracking. This control strategy uses explicitly the non linear robot model, future reference, constraints on the variables and has a optimized solution. And, to the matter of this work, it has not been found reports of the EPSAC applied in mobile robotics, and is thus an unprecedented application. Simulation results are presented comparing the controllers studied using performance indices. Else, the controllers were applied in a mobile robot.