Abstract
This paper presents two innovative approaches to design symbolic controllers for dynamical systems. The first novelty involves a new trajectory-based strategy for defining the states of a symbolic model, which provides a more accurate representation of the system’s dynamics than the traditional grid-based technique. The second novelty concerns using a Bounded-parameter Markov Decision Process rather than a Finite Transition System to model the behavior of a symbolic model. This procedure allows for handling the system’s stochastic behavior and considers uncertainties. The effectiveness of the novel approaches presented is demonstrated through numerical results.
Type
Publication
2023 62nd IEEE Conference on Decision and Control (CDC)
