Pruning Techniques for the Increasing Cost Tree Search for Optimal Multi-agent Pathfinding

Abstract

We address the problem of optimal path finding for multiple agents where agents must not collide and their total travel cost should be minimized. Previous work used traditional single-agent search variants of the A* algorithm. In Sharon et. al. (2011), we introduced a novel two-level search algorithm framework for this problem. The high-level searches a novel search tree called increasing cost tree (ICT). The low-level performs a goal test on each ICT node. The new framework, called ICT search (ICTS), showed to run faster than the previous state-of-the-art A* approach by up to three orders of magnitude in many cases. In this paper we focus on the low-level of ICTS which performs the goal test. We introduce a number of optional pruning techniques that can significantly speed up the goal test. We discuss these pruning techniques and provide supporting experimental results.

Meta-Agent Conflict-Based Search For Optimal Multi-Agent Path Finding

“…Additionally, MA-CBS with intermediate values also outperforms the ICTS solver in the brc202d and ost003d maps, and performed very similar to ICTS for the den520d map. This supports the understanding from (Sharon et al 2011b) that ICTS is especially effective for maps with large open spaces such as den520d but not as good for domains with corridors and deadends such as brc202d.…”

“…Only results for EPEA* as the low-level is presented as A* results were similar. Additionally, for comparison we also report the success rate of the best ICTS variant (Sharon et al 2011b). Note that curves are ordered according to their performance in the given map.…”

“…al. (2011a;2011b) suggested the increasing cost search tree (ICTS) -a two-level framework where the high-level phase searches a tree with exact path costs for each of the agents and the low-level phase aims to verify whether there is a solution of this cost.…”

“…At each node in the constraint tree a low-level search is per-formed to find individual paths for all agents under the constraints given by the high-level node. Sharon et al (2011a;2011b;2012) showed that the behavior of optimal MAPF algorithms can be very sensitive to characteristics of the given problem instance such as the topology and size of the graph, the number of agents, the branching factor etc. There is no universally dominant algorithm; different algorithms work well in different circumstances.…”