Event-based Controlfor Multi-Agent Systems

Sammanfattning: In this thesis, a novel approach to the average consensus problem for multi-agent systems is followed. A new event-based control strategy is proposed, which incorporates event-based scheduling of state measurement broadcasts over the network. The control-laws are based on the resulting piecewise constant functions of these measurement values. This facilitates implementation on digital platforms such as microprocessors and reduces the number of interagent communications over the network. Starting from a basic problem setup with singleintegrator agents, fixed undirected connected communication topologies, and no time-delays, the novel strategy is developed. Different triggering conditions guaranteeing convergence to an adjustable region around the average consensus point or asymptotic convergence to this point, respectively, are discussed. Numerical simulations show the effectiveness of this approach, outperforming classical time-scheduled implementations of the consensus protocol in terms of load on the communication medium. Furthermore the problem class is extended to networks with directed communication links, switching topologies, and time-delays in the communication as well as to agents with double-integrator dynamics. As an illustrative example, the novel strategy is applied to a formation control problem of non-holonomic mobile robots in the plane.