A real-time algorithm for task allocation

One of the main difficulties in designing multi-agent systems for environments with hard time constraints is the problem of task allocation, assigning an agent or group of agents to accomplish a specific system subgoal. In this paper we introduce a viable real-time task assignment algorithm for multi-agent systems. The algorithm, referred to as MATS (multi-agent task selection), is applied to two simulation environments and compared with optimal assignment algorithms, both in terms of speed and efficacy. Results are summarized and future areas of research discussed.