Variable Utility in Multi-robot Task Allocation Systems

Abstract

Multi-Robot Task Allocation (MRTA) systems have been thoroughly studied by the robotic community in the last decade, with the purpose of creating systems that can solve real problems with good results. Most MRTA architectures consider non-variable utility measures to value task execution payoff. However, in realistic settings there are problems such as tasks having priority and time lifespan, that can be better modeled with variable utilities. This paper introduces variable utility applied to MRTA systems, and exemplifies it by presenting a model for a box-pushing problem with this new concept.