Graph Laplacian Based Distributed Formation Maneuvering Control Of Multi-agent Systems

A significant amount of research efforts has been diverted towards the study of multi-agent coordination due to its widely applications and many remaining theoretical challenges.Formation maneuvering control,as one of the fundamental topics in multi-agent coordination,aims to provide a control strategy to achieve a desired group formation shape and a synchronous group maneuver-ing velocity.However,in a complex and dynamic environment,as the multi-agent system evolves towards a desired formation maneuvering,individual agents may collide with each other,encounter obstacles and even lose the sensing or communication links due to the physical constraints.There-fore,it is important to improve the environmental adaptability of formation maneuvering control in practical applications.This thesis studies formation maneuvering control of multi-agent systems for leader-follower networks and leaderless networks.The main work and results of this thesis can be summarized as follows:For multi-agent systems modeled by leader-follower networks,formation maneuvering con-trol strategies are proposed based on graph Laplacian.For both single-integrator kinematics and double-integrator dynamics,the formation maneuvering control strategies are proposed which en-sure global asymptotic stability of the system.Moreover,motivated by the observation,the results are extended to the formation maneuvering control strategy with good adaptability in complex and dynamic environment.The proposed graph Laplacian based formation maneuvering control strate-gies steer a team of multiple mobile agents into a formation with variable size and maneuvering with a synchronized velocity for both single-integrator kinematics and double-integrator dynamics.The formation control with size scaling is effectively improve the environmental adaptability of moving formation.For multi-agent systems modeled by leaderless networks,local control laws are proposed for each agent in the network,with the parameters capable of being designed in a distributed man-ner,and is implementable locally via relative sensing and communication with neighbors.In other words,the idea of distributed control goes through the design and implementation of this strat-egy,so we call it a fully distributed control strategy.For both single-integrator kinematics and double-integrator dynamics,fully distributed control laws are proposed based on which necessary and sufficient conditions regarding a critical control parameter are obtained to guarantee globally asymptotical convergence of the overall system.Moreover,as collision may occur between agents and the sensing or communication links may be lost due to the physical constraints,a switching con-trol strategy is proposed based on a distance potential function for formation maneuvering control of multi-agent systems with guaranteed collision avoidance and connectivity maintenance properties.