Coordination in hierarchical pickup and delivery problems using delegate multi-agent systems

Pickup and delivery problems are a generalization of the planning problem faced by transport companies. Large logistics providers often employ a hierarchical ‘hub and spoke’ overlay network to connect pickup and drop off sites. Hierarchical pickup and delivery problems pose two major challenges: (1) determining suitable routes from the origin to the destination of packages through the logistics providers’ network, and (2) allocating the resources that are required for pickup, transporting and delivering the packages along such a route. By combining traditional resource allocation techniques with swarm algorithms, the approach in this paper offers a decentralized solution to pickup and delivery problems in hierarchical environments. Resources are scheduled locally at the node from which they operate, resulting in a distribution of many local resource schedules. A swarm approach called delegate multiagent systems is used to extract information from relevant localized schedules and combine them in consistent global paths. The ant-like agents in these delegate multi-agent systems also redistribute feedback from the path planning mechanism to the decentralized resource scheduling mechanism. Results obtained by this hybrid approach show that it outperforms greedy and static alternatives.