Spatially Distributed Multiagent Path Planning

Abstract

Multiagent path planning is important in a variety of fields, ranging from games to robotics and warehouse management. Although centralized control in the joint action space can provide optimal plans, this often is computationally infeasi- ble. Decoupled planning is much more scalable. Traditional decoupled approaches perform a unit-centric decomposition, replacing a multi-agent search with a series of single-agent searches, one for each mobile unit. We introduce an orthogonal, significantly different approach, following a spatial distribution that partitions a map into high- contention, bottleneck areas and low-contention areas. Lo- cal agents called controllers are in charge with one local area each, routing mobile units in their corresponding area. Dis- tributing the knowledge across the map, each controller can observe only the state of its own area. Adjacent controllers can communicate to negotiate the transfer of mobile units. We evaluate our implemented algorithm, SDP, on real game maps with a mixture of larger areas and narrow, bottleneck gateways. The results demonstrate that spatially distributed planning can have substantial benefits in terms of makespan quality and computation speed.

Distributed Planning with Asynchronous Execution with Local Navigation for Multi-agent Pickup and Delivery Problem

“…However, the methods from this approach are sometimes incomplete. For example, Wilt and Botea [26] proposed a spatially distributed planner in which each controller agent manages a subarea and communicates with the adjacent controller confirming the transfer of a mobile unit to another subarea. However, this method is partially centralized to make it complete.…”

A Hierarchical Approach to Multi-Agent Path Finding

“…There are only few existing works on solving MAPF using spatial hierarchies. The Spatially Distributed Multi-Agent Planner (SDP) (Wilt and Botea 2014) partitions a map into high-contention and low-contention regions and uses different MAPF solvers for regions of different types. Different from HMAPP, SDP does not partition the map into smaller regions if no high-contention regions are found.…”