Optimal multi-agent planning solution for a sample gathering problem

Abstract

This research targets the problem of automatically planning a team of mobile agents such that they collect the samples scattered throughout an environment in minimum time. Each mobile agent can carry at most one sample at a time and it can travel a maximum total distance, given by agent's available energy. The environment is assumed already abstracted to a finite graph, where a node plays the role of the deposit where the samples should be gathered. Our solution consists in several steps that lead to a formulation of the initial problem as a Mixed Integer Linear Programming one. The solution yields a plan that imposes for each agent the samples and the order for collecting them. This result is optimal from the point of view of collecting all samples in minimum time.