Online convex optimization for control of dynamical systems
The project aims to advance the field of Online Convex Optimisation (OCO) in the context of control of dynamical systems. The goal is to develop and investigate OCO-based control schemes that can handle general cost functions and constraints without relying on restrictive assumptions. These algorithms will be applied to directly control dynamical systems and combined with established control strategies such as reference governors and model predictive control (MPC).
The project also aims to investigate the regret of general MPC schemes in the context of time-varying and a priori unknown cost functions. The expected outcome of the project is to achieve efficient and effective control of dynamical systems in real-world applications, especially in situations where the cost functions are a priori unknown and time-varying, which is a common challenge in many practical applications. The project will include theoretical analysis and algorithm development, as well as numerical simulations and experimental validation.