Is Controller Fragility a problem inherent to Robust Control?


Andrew Packard
Department of Mechanical Engineering
UC Berkeley

We will criticize most of the conclusions reached in a recent IEEE Transactions on Automatic Control publication, ``Robust, Fragile or Optimal,'' (August 1997, Keel and Bhattacharyya). While we do not dispute the numerical correctness of the paper's results, we believe that the interpretations reached by the authors are misleading. A thorough analysis of each example, modeling controllers as dynamic systems with imperfect components, bolsters our viewpoint. We will also comment on a published letter to TAC and the 1998 ACC session on the topic.

Our perspective is joint work with David Faris, Tom Pare, Jonathan How and Khalid Al-Ali.