Comparison of command governor augmentation and modified linear extended state observers for uncertain dynamical systems

In this paper, two approaches for model following control in the presence of uncertainties are compared. Specifically, the so-called command governor framework, originally developed as an augmentation to adaptive control, and a modified linear extended state observer are evaluated. Both approaches estimate the disturbance signal causing the plant to deviate from the desired reference model. The modified linear extended state observer uses an observer with a linear Luenberger gain to approximate the disturbance signal while the command governor augments the reference signal applied to both, reference model and uncertain plant, by filtered differentiation of the control error. By qualitative and numerical comparison of these approaches, it is illustrated that the command governor is a specific implementation of the more general modified linear extended state observer. Note that the compared approaches do not rely on nonlinear or hybrid control laws, but may be realized as a continuous time, LTI system. Therefore, these methods may be advantageous to similar approaches such as adaptive control with respect to certification.