Abstract
Task allocation is still a fundamental problem in Multi-Agents System (MAS). It allows the formation a coalition of agents to cooperate together in order to carry out a complex task. Generally, the process of task allocation requires calculating the value of all the possible allocations, then determining which optimal. In the context of Massive Multi-Agent Systems (MMAS), one agent generates all the possible coalitions and then calculates the value of each, is inefficient. Moreover, the traditional approaches based on the negotiation between agents, are impractical because of the complexity of communications between agents. In this paper we propose a decentralized method, based on grouping agents using the Formal Concepts Analysis (FCA) approach for the task allocation in MMAS. In our model, agents are fully cooperative and each task is composed of several subtasks. The proposed solution is based on two steps: i) computing groups of agents having similar characteristics in the objective to distribute the task allocation process among agents and minimize the communication between agents. ii) Finding the optimal allocation by sharing the task allocation process among groups of agents.