Safe adaptive control. Data-driven stability analysis and robust synthesis. (Q1958455)

The book is devoted to the analysis and design problems for adaptive control of highly uncertain systems. The basis for the results in the book is formed by the idea of data-driven unfalsified control theory which, roughly speaking, means that the acquisition of the process knowledge is performed by learning about the process through behavior of its signals and not through parameter estimation of its mathematical model. On the other hand the model is used to design a set of controllers switches which define the structure of the closed loop system. The concept of unfalsified cost levels enables a unified approach for all adaptive control techniques by recognizing the causes and repair mechanisms for the model mismatch instabilities. It leads to design procedures for the safe adaptive controllers which ensure robustness of stability and performance of the systems operating under uncertain conditions. The monograph consists of two parts. In the first part the authors discuss theoretical results dealing with robust stability and performance, and convergence of the general switching control framework including the case of an infinite set of candidate controllers. It enables to treat in a unified way both continuously parametrized adaptive controllers and multiple model/controller adaptive systems. The second part is devoted to practical examples which demonstrate pitfalls of the standard adaptive control design and potential advantages of the safe adaptive control theory. The examples include unfalsified adaptive spacecraft attitude control, automatic PID tuning, switching control of missile autopilot, switching congestion control for satellite networks, and direct adaptive control of a two-link robot arm. The authors also present some open questions and future research directions.