Research On Flocking Control Of Nonholonomic Multi-agent Systems

Flocking behavior is a ubiquitous phenomenon in the biological community.The systematical and in-depth research on flocking control problem can not only help to reveal the inherent mechanism of the flocking behavior of the biological community,but also provide the theoretical basis and technical guidance for the coordinated operation of man-made industrial systems,such as multi-robot formation,unmanned aerial vehicle network and wireless sensor network.Therefore,the research on flocking control problem has important theoretical significance and practical application value.Based on the above observations,for the two-dimensional and three-dimensional nonholonomic multi-agent systems,using the tools from graph theory,stability theory and coordinated control theory of multi-agent system,we analyze the nonlinear property of different types of nonholonomic system models,demonstrate and implement the different flocking control based on the design of the control protocol,including flocking control,attitude-synchronization flocking and multi-flocking control.The main contents of this thesis are as follows.Considering the virtual leader whose dynamics is unknown,state information is time-varying and not available to all agents,we investigate the distributed flocking control problem of two-dimensional nonholonomic multi-agent systems under both fixed and switching topologies.Based on the relative velocity and heading angle information of neighboring agents,two distributed discontinuous control protocols are designed for fixed and switching topologies,respectively,which don't invoke the potential energy function using position information.Based on theoretical analysis,the proposed distributed discontinuous control protocols guarantee that the velocities and heading angles of the agents exponentially converge to the velocity and heading angle of the virtual leader,respectively,while ensure collision avoidance and cohesiveness of the whole group.For the three-dimensional nonholonomic system model with fixed linear velocity,we study the flocking control problem of three-dimensional nonholonomic multi-agent systems with proximity graphs.The theoretical analysis reveals that the three-dimensional system model with constant linear velocity possesses the characteristic of Lipshcitz type,and the corresponding Lipschitz constant is obtained.A distributed linear control protocol is designed only using relative attitude information,and guarantees that the attitudes of all agents exponentially converge to synchronization,and simultaneously ensures connectivity preservation and collision avoidance of the whole group,which don't invoke the potential energy function using position information and saves the communication energy.While discussing collision avoidance among agents,we fully consider the volume factor of each agent,and design an adjustable minimum relative distance value for collision avoidance.Furthermore,we also discuss the communication delay problem in the distributed control protocol through simulations,which shows that the proposed control protocol can also drive the three-dimensional nonholonomic multi-agent systems into a stable flocking behavior if the communication delay is a small constant.For the three-dimensional nonholonomic system model with time-varying linear velocity,we study the attitude-synchronization flocking control problem of the general three-dimensional nonholonomic multi-agent systems.Considering that the general three-dimensional system model does not satisfy the characteristic of Lipshcitz type,we design a corresponding distributed linear control protocol only using relative velocity and attitude information,and make the general three-dimensional system model satisfy the characteristic of Lipshcitz type under the proposed control protocol.While discussing collision avoidance among agents,we fully consider the volume factor of each agent,and design an adjustable minimum relative distance value for collision avoidance.Furthermore,the proposed distributed control protocol can ensure the three flocking rules and attitude synchronization meanwhile,and get the sufficient condition of attitude-synchronization flocking control problem for the three-dimensional case.For task requirements of multi-object and multi-task in the practical application,we develop the collective motion of multi-flocking of two-dimensional nonholonomic multi-agent systems.Considering the effect of the actuation,the actuation point has a skewing to the center of mass in our presented nonholonomic system model.In the paper,we relax the condition on the persistent excitation,and the collective motion of networked nonholonomic mobile agents will be developed no matter the velocities are persistently exciting or not.Thus,the assumption on velocity in this paper is less conservative.We design a fully distributed control protocol for multi-flocking problem of networked nonholonomic mobile agents based on our proposed novel potential function.By combing algebraic graph theory with Barbalat's lemma,the proposed distributed control protocol guarantees that the agents in the same subgroup asymptotically move with the same velocity and heading angle,while ensuring connectivity preservation of each subgraph and collision avoidance of the whole systems.