Simulação de um controlador preditivo não linear para o controle de um veículo autônomo

Abstract

This paper presents the development and validation of a model-based predictive controller (MPC) for the control of autonomous ground vehicles. The study uses the kinematic model of a bicycle as a basis for representing the vehicle’s dynamics and proposes control strategies aimed at following predefined trajectories, while respecting physical limitations and comfort criteria. To validate the controller’s performance, two simulation environments are used: a simplified simulator, developed by the author, and the CARLA simulator, which offers a more realistic and complex scenario. The controller is tested on different trajectories (circular, oval and square), and aspects such as tracking error, acceleration and steering angle are analyzed. The simulated results show that the MPC is a promising approach for autonomous applications with high levels of automation, but with a very high computational cost that can hinder or even make its application unfeasible.