Feedback insensitivity to small measurement errors

This paper provides a precise result which shows that, if a (continuous-time, finite-dimensional) system is stabilizable in any way whatsoever (even by means of a dynamic, time varying, discontinuous feedback) then it can also be stabilized in a way which is insensitive to small measurement errors. This robustness property cannot be achieved, in general, by memoryless state-feedback controllers; the solution given involves a hybrid strategy based on the idea of sampling at a "slow enough" rate.