Desenvolvimento de estratégias de controle para rastreamento e estabilização de sistemas não-holonômicos

Abstract

The main practical problem that motivates the development of this thesis was the necessity to control autonomous vehicles with nonholonomic constraints. Nonholonomic systems constitute a system class with special characteristics: despite their movements are constraint, they can reach any conguration in the space where they are dened (when controllable and reachable); unfortunately control laws to stabilize nonholonomic systems are not easy to be generated, and constitute an open research area. Specically, due to diferential equations structure governing the nonholonomic system, the regulation problem cannot be solved via a smooth, time-invariant pure state feedback law due to the implications of Brockett's condition. To control autonomous vehicles it is necessary a system that identies its position, a path planner system that generates path compatibles with the system constraints, and a control system to follow this trajectory and stabilize the vehicle at equilibrium point. The main subject of this work is the stabilization and trajectory tracking of systems with nonholonomic constraints. A discontinuous control law (state scaling) is used to exponentially stabilize nonholonomic chained form systems. The previous results are extended to trajectory tracking regulation of n-dimensional chained form systems using internal model. To the trajectory tracking regulation problem yet is proposed to use generalized predictive control (GPC). Employing the GPC is possible to use actual information from the process model dynamics. Control system robustness is improved by adapting the model to the new conditions. At the beginning a GPC controller is proposed to the diferential driven wheeled mobile robot in Cartesian discrete form and further the GPC is extended to the diferential driven wheeled mobile robot in discrete chained form. The results obtained at the last step are valid to nonlinear system with nonholonomic constraints in discrete chained form or that ones which can be transformed to this form with n states and two inputs or controls.