Contraction analysis of switched systems with application to control and observer design

In many control problems, such as tracking and regulation, observer design, coordination and synchronization, it is more natural to describe the stability problem in terms of the asymptotic convergence of trajectories with respect to one another, a property known as incremental stability. Contraction analysis exploits the stability properties of the linearized dynamics to infer incremental stability properties of nonlinear systems. However, results available in the literature do not fully encompass the case of switched dynamical systems. To overcome these limitations, in this thesis we present a novel extension of contraction analysis to such systems based on matrix measures and differential Lyapunov functions. The analysis is conducted first regularizing the system, i.e. approximating it with a smooth dynamical system, and then applying standard contraction results. Based on our new conditions, we present design procedures to synthesize switching control inputs to incrementally stabilize a class of smooth nonlinear systems, and to design state observers for a large class of nonlinear switched systems including those exhibiting sliding motion. In addition, as further work, we present new conditions for the onset of synchronization and consensus patterns in complex networks. Specifically, we show that if network nodes exhibit some symmetry and if the network topology is properly balanced by an appropriate designed communication protocol, then symmetry of the nodes can be exploited to achieve a synchronization/consensus pattern.