Multi-fidelity algorithms for the horizontal alignment problem in road design

Multi-fidelity refers to methods that employ low fidelity and high fidelity models in their design so as to lower the computation time and obtain high accuracy simultaneously. In particular, the algorithms that use the multi-fidelity technique in their design adaptively controls the error of the objective function. As they reach better positions in the search space, they update the objective function and use a new version with better precision. In this thesis, we study the multi-fidelity algorithms for solving the horizontal alignment problem in road design. The algorithms that we study here are a generalized pattern search with adaptive precision control and a trust region algorithm for unconstrained problems with controlled error. At first, in order to make a fair comparison, we tune the parameters of each algorithm on 5 small roads. We then test the algorithms on 35 roads, ranging from small to very large roads. The results of the comparison demonstrate that the use of the multi-fidelity framework helps the horizontal alignment algorithms reduce their computation time by an average factor of 2.52 while preserving the quality of solutions.Arts and Sciences, Irving K. Barber School of (Okanagan)Computer Science, Mathematics, Physics and Statistics, Department of (Okanagan)Graduat