The Effects of Inter-Agent Variation on Developing Stable and Robust Teams

This paper provides a formal analysis of a multi-agent task allocation problem and how variation in agent behavior in the form of response probabilities can be used to build redundancy in the multi-agent system (MAS). In problems where experience is beneficial redundancy provides an MAS with a back-up pool of actors if the primary actors are unavailable. We examine how to ensure a complete team of agents needed for a particular task will be formed, as well as two different ways of determining how to ensure some level of redundancy.