Multi-Agent Positional Consensus Under Various Information Paradigms

摘要: This thesis addresses the problem of positional consensus multi-agent systems. A is achieved when agents converge to a point. Some applications this class in mid-air refueling aircraft or UAVs, targeting geographical location, etc. In research work some algorithms have been developed. They can be categorized two part (i) Broadcast control based algorithm (ii) Distributed algorithm. case broadcast strategies for group developed achieve consensus. The constrained by requirement that every agent must given same input through communication mechanism. Although command computed using state information global framework, implemented local coordinate frame. mathematical formulation has done linear programming framework computationally less intensive than earlier proposed methods. Moreover, random perturbation command, helps reasonable proximity among even large number agents, which was not possible with existing strategy literature, introduced. method extended at pre-specified location. comparison between LP approach and SOCP also presented. demonstrated successfully on robotic platform made from LEGO Mindstorms NXT Robots. second algorithm, decentralized multiple autonomous concept. Even here, framework. Each sensing radius it capable position orientation other within their region. feasible easy implement. distributed algorithms, efficient rendezvous uses rectilinear decision domain (RDD), as against circular assumed available literature. reducing its computational complexity considerably. An extensive analysis carried out prove convergence