Impact of model accuracy on the performance of an optimization-based path planning algorithm for an articulated vehicle
Abstract: This thesis presents a study on how the accuracy of vehicle kinematic models can affect the performance of an optimization-based path planner. A bicycle model and an articulated machine model are chosen to be different accurate levels of modelling. The path planner generates a final path in two steps. First a coarse path is obtained by an adapted A' algorithm as an initial guess, then the final smooth collision-free path is generated by a nonlinear optimization algorithm according to the initial guess and the selected vehicle model. The differences regarding path length, success rate and computation time between these two models are compared in this thesis. The generated paths are further used in real tests as reference paths of an autonomous articulated hauler. The performance of different models is evaluated by the lateral error and the heading error. The findings suggest that for the articulated hauler, the articulated machine model provides a more accurate and more realistic path, yet in some cases, it can be replaced by the bicycle model that has better time efficiency in computation while maintaining the performance of the generated path.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)