Abstract-- In experimentally assessing the stability and performance level of a large collection of controllers in feedback with a single unknown actual plant, we develop a method of searching for a subset of controllers to manage the number of the experimental assessments. Avoiding testing every prospective controller, we check whether every controller in the large set of controllers will work satisfactorily when they are connected with an actual but unknown plant by experimenting only a small subset of candidate controllers. In doing this, the -gap metric will be a main tool.
Jisang Park, Robert R. Bitmead