ICBS: The Improved Conflict-Based Search Algorithm for Multi-Agent Pathfinding

A Multi-Agent Path Finding (MAPF) problem is defined by a graph, G = (V,E) and a set of k agents labeled a1 . . . ak, where each agent ai has a start position si ∈ V and a goal position gi ∈ V . At each time step an agent can either move to an adjacent location or wait in its current location. The task is to plan a sequence of move/wait actions for each agent ai, moving it from si to gi such that agents do not conflict, i.e., occupy the same location at the same time. Conflict-Based Search (CBS) (Sharon et al. 2012), is an effective optimal MAPF solver. CBS has two levels. The low-level finds optimal paths for the individual agents. If the paths conflict, the high level, via a split action, imposes constraints on the conflicting agents to avoid these conflicts. Meta-agent CBS (MA-CBS) (Sharon et al. 2015) generalizes CBS by merging groups of agents into meta-agents when beneficial. In this paper we introduce Improved-CBS (ICBS) which further adds three new improvements to MA-CBS.