Path Following Control of an Agricultural Skid-Steering Robot: An MPC Approach

The purpose of this thesis is to describe and implement a control algorithm for an agricultural rover moving inside an orchard. This work is done on a real autonomous agricultural rover crafted at CASY Laboratory and is taught to help farmers in doing the most demanding jobs on the orchard. In particular, the work is not focused on the simple navigation of the rover inside the rows, but on more difficult trajectories. The thesis focuses on the tracking of non-trivial trajectories required by the challenging scenario in which the rover is moving. Due to the orchards shape and to the ground geometry, often circular or squareform trajectories are required. In addition, the rover can have the necessity to change the row to reach every part of the orchard and there can be some area that can be restricted and so the rover has to perform non-trivial trajectory even within a single row. In order to make the rover realise its purpose of following a path, the control method chosen is the Model Predictive Control, a powerful approach that is well known and really effective, as shown throughout the development of the thesis. The realised controller has been tested and analyzed with different trajectories and, moreover, with different parameters and constraints.