DYNAMIC CONSENSUS CONTROL OF LINEAR DISTRIBUTED MULTI-AGENT SYSTEMS

Abstract

Due to the recent advancements in the modern control theory, the technology for controlling a single system has reached a considerably matured phase. A single complex system when equiv alently replaced by several simpler agents, greater benefits can be achieved. Multi-agent system (MAS) is defined as a loosely coupled structure composed of multiple agents, which interact each other to solve problems. With more research and observations done on the self organiza tion and local interactions among various biological populations in nature, like the division of labour among ant colonies, formation of bird groups MASs emerged. MASs are widely used in spacecraft, automated highway systems, unmanned aerial vehicles, attitude synchronization of satellites etc. Cooperative applications of MAS include flocking, formation and consensus. The first stage of the work analyses consensus problem which is an integral area of research under MAS. Static consensus control of MAS is studied using state feedback control. Consid ering the limitations of the state feedback controller in handling constraints, the control law is extended using an optimum control strategy utilizing a distributed model predictive controller (DMPC), which incorporates the constraints involved in the problem. Dynamic consensus con trol of linear distributed homogeneous agents with fixed communication topology is then stud ied utilizing the problem of vehicle platooning. The vehicle’s state space model is derived from its lateral dynamics. The DMPC control is examined with and without constraints, with differ ent sets of initial conditions. The results of MATLAB simulations demonstrate the effectiveness of the recommended control scheme to solve the benchmark problem of vehicle platooning.