PhD Studentship in Aeronautics: Safe Data-Driven Control - Bridging the Gap Between Nonlinear Control and Machine Learning (AE0056)

Start Date: Between 1 August 2025 and 1 July 2026

Introduction: Obtaining highly accurate mathematical models of a system is often not feasible (or it is undesirable due to complexity). Nevertheless, from a control engineering perspective, we still wish to design control inputs to achieve rigorous performance guarantees, so that one can mathematically prove that desired behaviours will occur or that undesired behaviours (e.g. instability) will not. Machine Learning (ML) has emerged as a tool for ‘learning patters’ from data sets. Yet, there is a gap between the core aim of control engineering, obtaining rigorous performance guarantees, and ML, which lacks such guarantees.

Objectives: In this project we will seek to bridge the gap between traditional control engineering and ML by combining nonlinear control theory, system identification and ML to develop novel data-driven methods to design control strategies for partially/fully unknown dynamical systems. We aim to develop strategies to ensure the training phase is safe. Moreover, we will explore methods of experiment design to shorten its duration and integrate performance objectives directly into the training process. Second, we will exploit nonlinear control tools to obtain performance guarantees or to systematically construct warning signals that will indicate unexpected/unsafe behaviours before they occur. Such signals may not only serve as an indicator of potential issues but can also be used to signal when additional training is necessary. The latter aspect will serve as a stepping stone for real-time learning and control. We will explore applications of the developed theory, e.g. in the context of robotic systems.

Supervisors: Dr Thulasi Mylvaganam, Nonlinear control, dynamic optimization and data-driven control.

Learning opportunities: You will develop a strong expertise in control engineering, with a particular focus on fundamental aspects that can have far-reaching impacts in diverse fields.

Professional Development: You will have access to engaging professional development workshops in areas such as research communication, computing and data science, and professional progression through our Early Career Researcher Institute.

Duration: 3.5 years.

Funding: Full coverage of tuition fees and an annual tax-free stipend of £21,237 for Home, EU and International students. Information on fee status can be found on our fees and funding webpages.

Eligibility: You must possess (or expect to gain) a First class honours MEng/MSci or higher degree or equivalent in Engineering or Mathematics. A strong background in control engineering and mathematics is crucial, and familiarity with topics such as nonlinear control, optimal/robust/adaptive control or system identification is desired. The candidate must be motivated to undertake fundamental research that is highly mathematical in nature.

How to apply: Contact Dr Thulasi Mylvaganam with your CV and any additional supporting documents before submitting an application: t.mylvaganam@imperial.ac.uk

Submit your application via our Apply webpages. When submitting your application, you will need to use the following details:

• Search course/Programme: Aeronautics Research (PhD)
• Research Topic: Please use reference number AE0056
• Research Supervisor: Dr Thulasi Mylvaganam
• Research Group: Aero

For further information: For questions about the project, please email Dr Thulasi Mylvaganam, Senior Lecturer: t.mylvaganam@imperial.ac.uk

For queries regarding the application process, email the PhD Administrator.

Application deadline: 9 January 2025

Equality, Diversity and Inclusion: Imperial is committed to equality and valuing diversity. We are an Athena SWAN Silver Award winner, a Stonewall Diversity Champion, a Disability Confident Employer and are working in partnership with GIRES to promote respect for trans people.