Design Framework for Achieving Guarantees with Learning-Based Observers

The paper proposes a novel framework for state observer design, in which learning-based observers are incorporated. aim of the method is to provide framework, able guarantee limitation observation error, even if error under all scenarios cannot be verified. based on robust H∞ design method, guarantees resulted observer. Moreover, process extended with controller leads joint controller-observer design. In this proposed applied vehicle control problem, such as lateral path following. problem goal an accurate velocity signal vehicle, used controlled system generation front wheel steering angle. effectiveness illustrated through simulation examples high-fidelity dynamic simulator CarMaker.