Abstract
It is of great interest in an increasing number of fields and applications to use autonomous vehicles in varying and challenging terrain conditions. In these conditions, the vehicle's movement is strongly affected by the type of terrain being traversed and unpredictable disturbances from the environment. Varying ground friction, uneven surface and slopes are examples of these effects. Consequently, producing path following controllers becomes challenging, as ordinary approaches are based on ideal models of the vehicle, where the surface is flat and the wheels have perfect traction with the ground. Thus, using ordinary approaches in off-road terrain may lead to severely decreased performance, and in the worst case, failure to follow predetermined paths. Therefore, this thesis investigates how ordinary path following controllers are affected by varying terrain conditions, and how extending the vehicle model can improve performance. Three methods from the research literature were chosen, each extending the ordinary vehicle model with relevant parameters, to obtain a more accurate vehicle model. Computer simulations were carried out, and the results show two of the proposed approaches outperforming an ordinary controller in some of the simulated terrain conditions. Thus, indicating the possibility of improving path following performance by extending ordinary vehicle models with relevant parameters. Possible implications are broadening the applications of autonomous vehicles and improving robustness in their performance. More research is needed to investigate whether similar results can be expected in real-world operation.