The work presented in this thesis aims at proposing methods for computing bicycle paths across a metropolitan. Several criteria such as distance, safety and effort must be considered in the path computation. The difficulty is to compute paths under a time constraint of a few seconds, in order to integrate the computation in the respond-time of a web page. Two approaches were discussed to solve this problem. The first one is an a posteriori approach where all compromise solutions are computed and the second approach is an a priori method that takes user preferences into account to guide the search by the selection of the most promising sub-paths first. Finally, we propose to model the road network as a line graph to take into account new criteria, requiring costs on arc sequences for example. All this work was necessary to develop the service Géovélo, which is a multiob- jective route planner adapted to bicycle. The service is available on a website and as mobile applications.
展开▼
