PhD Applications

Pursuing a PhD in our lab opens doors to exceptional opportunities in both academia and industry. Graduates I've worked with go on to secure research scientist positions at leading tech companies, academic appointments with the potential to establish their own research groups, and launch innovative startups for those with entrepreneurial vision.

As a PhD student in SWIRL, you'll engage in cutting-edge research within a collaborative and intellectually stimulating environment. You'll have access to state-of-the-art computational resources, work closely with me and fellow researchers, and present your work at premier international conferences and workshops.

Important: Before submitting an official application, please complete the form at the bottom of this page. This helps me assess whether a full application would be suitable and allows me to provide initial guidance.

If invited to proceed, submit your official application through the Department of Computing's online application system:

• Academic Programme: Select "Computing Research PhD" (not "AI and Machine Learning PhD 4YFT")

• Proposed research supervisor: Enter your supervisor's name

• Proposed research group: Enter "Foundational Reinforcement Learning Lab"

• Proposed research topic: Provide a title based on your research proposal

The minimum requirement is a Master's degree with grades equivalent to a UK distinction. However, given the competitive nature of applications to SWIRL, successful candidates typically demonstrate:

• Outstanding academic performance across their degree

• Deep knowledge and genuine interest in reinforcement learning, robotics, machine learning, and related areas beyond standard coursework

• Research experience or potential evidenced through publications, significant personal projects, or a compelling and well-reasoned research proposal

While most students begin in October, alternative start dates may be possible depending on circumstances.

Application deadlines:

• 15th October 2025

• 15th December 2025

• 15th February 2026

• 15th April 2026

Applications are reviewed after each deadline, with promising candidates invited for interviews. The review process may take several weeks to months, and applications from different deadlines may be considered together.

PhD offers are conditional on securing appropriate funding to cover tuition fees and living expenses. Funding options include:

• Self-funding or external scholarships: Students may fund their studies independently or through scholarships from their home countries. PhD programs at Imperial typically span 4 years, requiring tuition fees for 3 years and living costs for 4 years (no tuition fees in the final year).

• College scholarships: After receiving a conditional offer, your application proceeds to the department's funding committee. Imperial offers various scholarships covering full tuition and living costs. Funding decisions are based on merit among qualified candidates.

Your eligibility depends on your fee status:

• Home students: UK nationals or those with settled/pre-settled status

• Overseas students: All other applicants

Additional external scholarships may be available based on your nationality. If you already have funding secured, please indicate this in your application.

Your research proposal is crucial for assessment and will guide interview discussions. While your actual PhD may evolve based on supervision, interests, and field developments, the proposal demonstrates your research thinking and communication skills.

Key recommendations:

• Length: Maximum 4 pages, prioritizing conciseness

• Audience: Write for a specialist audience (primarily me)

• Focus: Get straight to the point—avoid extensive literature reviews

• Visuals: Include figures to illustrate your ideas clearly

Research Focus Areas

Research in FRL centers on foundational aspects of reinforcement learning. We're particularly interested in:

• Sample-efficient learning: Minimizing real-world trials while maximizing skill acquisition

• Safe exploration: Learning without catastrophic failures or unsafe behaviors

• Multi-paradigm integration: Combining online RL, offline RL, and imitation learning effectively

• Embodied learning: Leveraging complete physical morphology for sensing and control

• Soft robotics: Learning in high-dimensional, continuous deformation spaces

• Compliant interaction: Safe physical interaction with humans and environments

• Sample complexity: Understanding fundamental limits and achieving optimal rates

• Transfer and generalization: Learning across morphologies, tasks, and environments
  
  

We value research that combines theoretical rigor with practical impact, particularly work that addresses the fundamental challenges of making intelligent robots both capable and safe for real-world deployment.