Robust Superstabilizing Controller Design from Open-Loop Experimental Input/Output Data

This paper considers the problem of directly synthesizing a stabilizing controller from experimental data, avoiding an explicit plant identification step. Its main result shows that, if a slightly stronger result is desired, namely super-stability rather than just stability, then the problem admits a solution if and only if a related Linear Program, obtained using robust optimization techniques, is feasible. In this case, solving this LP leads to a robust controller guaranteed to super-stabilize the entire consistency set while optimizing worse-case performance.