Dynamics-based linearization and control of an autonomous ground vehicle.

Abstract

Several aspects of dynamics based motion control of autonomous ground vehicles have been studied. For example kinematic and dynamic modelling of the vehicle, path and motion planning, path following and trajectory tracking, stabilization to a manifold and a to a desired position with specified orientation have been considered. Assuming the (ideal) rolling without slipping conditions for a wheel imposes the nonholonomic (non-integrable) constraints on the motion of the wheel and the vehicle. Nonholonomic systems are found not to be asymptotically stabilizable by a smooth state feedback laws. In this research, first, kinematics and Newtonian dynamics of a tricycle with front wheel steering and driving are developed and then using the concepts of Lagrange equations of the first kind, the dynamic models of the tricycle in the form of x = f(x,u) with torque inputs are derived. Nonlinear control theory and input-output feedback linearization as a systematic approach have been applied on these dynamic models of the vehicle. Cartesian space feedback linearization is utilized for trajectory tracking and asymptotic stabilization to a one dimensional manifold and simulation results are provided. Curvilinear space feedback linearization approach for asymptotic stabilization of two outputs with open-loop control of the third planar variable using a planned path between two end positions and orientations has been verified by simulations and implemented on an experimental setup. Internal dynamics of the system has been investigated, and its dimension, state variable and stability have been analysed.