Semi-Decentralized Optimal Control of a Cooperative Team of Agents

In this project a decentralized optimal control law is proposed for a team of dynamic agents which is also called a multi agent system. The purpose of this controller for the group of agents is to accomplish consensus in a leaderless structure. In this order a semi-decentralized optimal control strategy is designed. The optimal control law is based on the local cost function of each of the agents and this cost function is minimized based on the Hamilton Jacobi Bellman (HJB) set of equations. The dynamic of the agents is governed based on the interaction between agents which is basically the flow of information about states of agents toward each other. Beside this flow of information there is also a local control input to each agent which also plays a roll in controlling dynamics of each agent. The optimal control law essentially effects the system through these two control inputs.It is worth noting that the consensus algorithm is derived in a formal way that is based on conventional multi agents control systems methodologies. Finally, the simulation results are presented in order to show the effectiveness of the proposed optimal control law.