CONTROL THEORY AND OPTIMISATION

My team’s research is centred around developing methods in two main areas:

1. Designing controllers using optimisation: My group develops numerical methods and optimisation techniques specifically aimed at designing controllers, with particular emphasis on managing constraints, nonlinear behaviour, and uncertainty in real-time applications.
2. Optimisation of closed-loop systems: We also examine how advanced optimisation methods can enhance performance across entire closed-loop systems, adopting a holistic rather than isolated subsystem perspective.

MODEL PREDICTIVE CONTROL (MPC)

My team extensively researches Model Predictive Control, an optimisation-based approach valued for systematically managing constraints and nonlinear dynamics. A key challenge is MPC’s heavy computational burden. To address this, our group develops novel numerical methods, including integrated residual techniques for efficiently solving nonlinear optimisation problems. We also investigate local reduction methods that enable effective handling of uncertainties, significantly outperforming traditional Monte Carlo methods. Additionally, we are creating entirely new classes of both derivative-based and derivative-free optimisation solvers, enhancing computational efficiency in MPC.

OPTIMAL CYBER-PHYSICAL CO-DESIGN

My group also specialises in the joint optimisation of control algorithms alongside physical systems and computing resources. Instead of independently designing these components, we use mathematical optimisation to jointly determine control algorithms, computing hardware, and physical parameters. Our methods explicitly address trade-offs between performance, cost, reliability, and resource usage, providing systematic solutions to co-design challenges.

Our ongoing research addresses three key areas:

1. Hybrid and real-time aspects: Cyber-physical systems combine discrete computational processes with continuous physical dynamics. We develop optimisation methods explicitly tailored to these hybrid characteristics.
2. Nonlinear optimisation under uncertainty,: We formulate and solve optimisation problems involving uncertain parameters, nonlinearities, and constraints, frequently encountered in real-world applications. Here, our local reduction methods are particularly effective in managing uncertainty efficiently.
3. Real-time computation in embedded and distributed systems: Our group designs numerical algorithms capable of running reliably and efficiently on embedded systems, where computational resources are limited.

APPLICATIONS

My team validates and inspires our theoretical work through real-world applications in multiple domains:

1. Information systems and robotics: We solve complex control and scheduling problems in multi-agent robotic systems and communication networks, integrating computation, communication, and actuation decisions.
2. Aerospace systems: Our aerospace work includes autonomous energy-efficient vehicles, load reduction strategies for wind turbines, and aerodynamic shape optimisation. We combine physics-based and data-driven optimisation to co-optimise control strategies with physical design decisions, such as aerodynamic profiles and sensor placement.
3. Building energy systems: We also actively develop control and optimisation techniques for energy-efficient buildings. By leveraging advanced optimisation methods, our team addresses challenges related to energy consumption, occupant comfort, and dynamic environmental conditions, achieving performance improvements over conventional methods.