A Label Correcting Algorithm for the Shortest Path Problem on a Multi-modal Route Network

We consider shortest paths on a multi-modal transportation network where restrictions or preferences on the use of certain modes of transportation may arise. The regular language constraint shortest path problem deals with this kind of problem. It models constraints by using regular languages. The problem can be solved efficiently by using a generalization of Dijkstra’s algorithm (DRegLC). Recently a speed-up technique called SDALT has been proposed to speed up DRegLC. In this paper we present a new label correcting algorithm (lcSDALT) which overcomes some limitations of SDALT. We evaluated lcSDALT on a realistic multi-modal transportation network with various modes of transport and road types, including timedependent cost functions on arcs. We observed a speed-up of a factor 3 to 30, with respect to DRegLC and of up to 50% with respect to SDALT, while at the same time reducing memory requirements.