Multi-agent Coordination Control With Nonlinear Dynamics,Quantized Communication And Structure-constraint

Coordination control of multi-agent systems has attracted more and more attention from control and robotics community due to its board application in areas such as coordinated robots,distributed sensor fusion,formation flight and distributed optimization,etc.In recent years,studies on distributed coordination control are still ongoing,both in depth and breadth.However,there are still many insufficiencies from theoretical research to practical application.Motivated by this observation,this dissertation mainly focuses on the core elements of multi-agent systems including agent dynamics,data transmission and interconnection structure and solves the critical challenges caused by nonlinear dynamics,quantized communication and structure constraint.The main contents and contributions can be summarized as follows:For directed topology structure,we explore the model reduction problem based on the edge agreement dynamics of multi-agent system.To begin with,the general concepts of weighted edge Laplacian of directed graph are originally proposed and its algebraic properties are further explored.Firstly,we point out that multi-agent system can be transformed into an output feedback interconnection of the spanning tree subsystem and the co-spanning tree subsystem.Then,based on the essential edge Laplacian,we propose a model reduction representation of the closed-loop multi-agent systems based on the spanning tree subgraph.Compared to traditional modeling approach,the edge agreement dynamics reveals the nature of multi-agent systems.Considering the inherently nonlinear dynamics of practical systems,we study the nonlinear consensus problem of multi-agent systems under directed topology by combining the graph-theoretic tools and ISS(input-to-state stability)cyclic-small-gain theorem.The second-order edge agreement problem for nonlinear multi-agent systems with unknown locally Lipschitz dynamics under directed topologies is discussed.By using the backstepping design,the original multi-agent system can be remodeled as several interacted subsystems with proven ISS properties.Additionally,the interactions of the interacted subsystems can be explicitly illustrated as a gain-interconnection digraph.With the aid of the ISS cyclic-small-gain theorem,the asymptotic stability of the whole system can be guaranteed.The quantized edge agreement problem of second-order nonlinear multi-agent systems is studied,in which both static and dynamic quantization are considered.For static quantization,we do not only guarantee the stability of the proposed quantized control law,but also reveal the explicit mathematical connection of the quantized interval and the convergence properties for both uniform and logarithmic quantizers.To further reduce the bit depth(number of bits available)and to obtain better precision,the dynamic quantized communication strategy referring to zooming in-zooming out scheme is also studied.Based on the reduced model associated with the essential edge Laplacian,the asymptotic stability of second-order multi-agent systems under dynamic quantized effects with only finite quantization level can be guaranteed.Inspired by the encirclement behavior of dolphins to entrap the fishes,we investigate the bionic encirclement control and circumnavigation control by employing the bearing rigidity theory.To begin with,we transform the encirclement problem into target formation with time-varying bearing constraint.Then,to achieve the encirclement behavior as well as the scale shrinking mechanism,we propose a distributed control law by combining orthogonal projection operator and the consensus protocol.In addition,we analyze the stability of the coordination tasks,including formation encirclement,formation tracking and formation scale-shrinking,by using Lyapunov method and small-gain theorem.Based on the encirclement control law,we further introduce the distance constraints to multi-agent systems to achieve circumnavigation.Numerical examples and experiment results are given to verify the efficiency and feasibility of the encirclement and circumnavigation control law.