Multi-Agent Path Finding for Non-Conflicting Paths: An Infinite Speed Conflict-Based Search Approach

Yaodong Hou, Gang Tao, Zheng Zang, Guojie Kong, Jianwei Gong*

*此作品的通讯作者

科研成果: 书/报告/会议事项章节会议稿件同行评审

摘要

Multi-Agent Path Finding (MAPF) is the problem of finding a set of non-conflicting paths for multiple agents on a graph, which is NP-hard to solve optimally. This paper proposes an infinite speed Conflict-Based Search (IS-CBS) algorithm to solve the MAPF problem when the agents move at different speeds and the edges correspond to roads of varying lengths. In addition, three distinct approaches are introduced to calculate admissible heuristics for IS-CBS: the sum approach which uses the sum of all heuristics as the final heuristic, the Pareto approach which builds a Pareto set and selects nodes from it, and the alternation approach which alternates between heuristics in search iterations. In our experiment, IS-CBS with heuristics outperforms IS-CBS in terms of success rate, runtime, and number of expanded nodes which is reduced by up to a factor of 21.

源语言英语
主期刊名2023 42nd Chinese Control Conference, CCC 2023
出版商IEEE Computer Society
5500-5505
页数6
ISBN(电子版)9789887581543
DOI
出版状态已出版 - 2023
活动42nd Chinese Control Conference, CCC 2023 - Tianjin, 中国
期限: 24 7月 202326 7月 2023

出版系列

姓名Chinese Control Conference, CCC
2023-July
ISSN(印刷版)1934-1768
ISSN(电子版)2161-2927

会议

会议42nd Chinese Control Conference, CCC 2023
国家/地区中国
Tianjin
时期24/07/2326/07/23

指纹

探究 'Multi-Agent Path Finding for Non-Conflicting Paths: An Infinite Speed Conflict-Based Search Approach' 的科研主题。它们共同构成独一无二的指纹。

引用此