Contraction Hierarchies (CHs) have been successfully used as a preprocessing technique in single-objective graph search for finding shortest paths. However, only a few existing works on utilizing CHs for bi-objective search exist, and none of them uses CHs to compute Pareto frontiers. This paper proposes an CH-based approach capable of efficiently computing Pareto frontiers for bi-objective search along with several speedup techniques. Specifically, we propose a new preprocessing approach that computes CHs with fewer edges than the existing preprocessing approach, which reduces both the preprocessing times (up to 3× in our experiments) and the query times. Furthermore, we propose a partial-expansion technique, which dramatically speeds up the query times. We demonstrate the advantages of our approach on road networks with 1 to 14 million states. The longest preprocessing time is less than 6 hours, and the average speedup in query times is roughly two orders of magnitude compared to BOA*, a state-of-the-art single-query bi-objective search algorithm.
|Number of pages||10|
|Journal||Proceedings International Conference on Automated Planning and Scheduling, ICAPS|
|State||Published - 2023|
|Event||33rd International Conference on Automated Planning and Scheduling, ICAPS 2023 - Prague, Czech Republic|
Duration: 8 Jul 2023 → 13 Jul 2023
Bibliographical notePublisher Copyright:
Copyright © 2023, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.