Multi Agent Based Dynamic Task Allocation

Multi-Agent system, which is relatively a recent term, can be viewed as a sub-field of Distributed AI. In order to construct and solve complex real world problems like task allocation in an organization, a number of agents can perform work cooperatively and collaboratively. To achieve the maximum system utility through the proper task allocation among agents, there is a notion of dynamic team formation. In this paper we focus on task allocation mechanism to agents with dependencies among various tasks to agents either individually or by forming a team whenever needed according to their capability to do a particular task. We propose few algorithms which dynamically filter the capable agents for performing task and hence allocating them with the help of synergy value and their individual utility. We also simulate the task allocation mechanism done by our proposed algorithms and show the performance of the system by defining some metrics like, agent utility, team utility, system utility.