Dominik Kirchler, Leo Liberti, Thomas Pajor, and Roberto Wolfler Calvo
Shortest paths on road networks can be efficiently calculated using Dijkstra's algorithm (D). In addition to roads, multi-modal transportation networks include public transportation, bicycle lanes, etc. For paths on this type of network, further constraints, e.g., preferences in using certain modes of transportation, may arise. The regular language constrained shortest path problem deals with this kind of problem. It uses a regular language to model the constraints. The problem can be solved efficiently by using a generalization of Dijkstra's algorithm (DRegLC). In this paper we propose an adaption of the speed-up technique uniALT, in order to accelerate DRegLC. We call our algorithm SDALT. We provide experimental results on a realistic multi-modal public transportation network including time-dependent cost functions on arcs. The experiments show that our algorithm performs well, with speed-ups of a factor 2 to 20.
|Published in||Proceedings of the 11th Workshop on Algorithmic Approaches for Transportation Modeling, Optimization, and Systems (ATMOS'11)|
|Series||OpenAccess Series in Informatics (OASIcs)|