EPSRC DLA: Foundation Models for Context-Aware Perception and User Trust in Autonomous Vehicles on UK Roads

PGR-P-2446

Key facts

Type of research degree: PhD
Application deadline: Tuesday 31 March 2026
Project start date: Thursday 1 October 2026
Country eligibility: UK only
Funding: Funded
Source of funding: Research council
Supervisors: Dr Mahdi Rezaei
Additional supervisors: Milad Mehdizadeh
Schools: Institute for Transport Studies

<h2 class="heading hide-accessible">Summary</h2>

One full scholarship is available in the Institute for Transport Studies in 2026/27. This scholarship is open to UK applicants and covers fees plus maintenance. The Institute for Transport Studies invites applications from prospective postgraduate researchers who wish to commence study for a PhD in the academic year 2026/27 Institute for Transport Studies EPSRC DLA Scholarship.

<h2 class="heading hide-accessible">Full description</h2>

Background: Recent academic and policy studies have raised concerns about the readiness of the UK road network for large-scale deployment of automated vehicles (AVs) (Tengilimoglu et al., 2024). Unlike controlled pilot environments, UK roads are characterised by winding geometries, heterogeneous layouts, and frequent marking and surface degradation from persistent adverse weather conditions. UK motoring surveys report that nearly 72% of drivers say road markings are significantly faded (RAC, 2026). Another study reports up to 1.9 million potholes annually on UK roads (AIA, 2025). These conditions challenge AV perception systems, which rely on consistent visual and geometric cues, and also affect public trust and AV adoption, as users view the roads-AVs as interconnected elements of future mobility. However, despite millions of miles driven, most AV trials occur in well-maintained, highly structured settings that do not reflect the UK network’s operational reality or reflect user acceptance. Even with improved maintenance, infrastructure degradation and weather-induced variability remain persistent, rather than exceptions, especially in adverse climates like the UK. Research Gap: While recent research has explored multimodal perception and end-to-end learning for autonomous driving (Chen et al., 2024), context-aware world modelling approaches that integrate road condition, weather effects, and geometric uncertainty remain underexplored. Methods are typically optimised for standard conditions and structured infrastructure, limiting robustness and generalisation on degraded networks common in the UK; consequently, AV perception struggles to sustain reliable situational awareness in ambiguous or weather-affected scenarios. Furthermore, few studies connect these technical limitations to user perceptions, behavioural intentions, and public acceptance of AVs, which therefore constrains policymakers’ and developers’ ability to assess readiness. Bridging this gap requires evaluating both technical performance and perceived readiness. Objectives: This EPSRC DLA project addresses the gap by defining an AI-driven, user-centric research direction for autonomous driving via multimodal foundation models for context-aware perception. Objectives include: • Designing a unified multimodal world model fusing camera and LiDAR into a self-updating representation encoding scene geometry, road condition, and weather-induced uncertainty. • Evaluating robustness, interpretability, and generalisability under adverse weather and degraded roads versus state-of-the-art modular pipelines. • Examining how infrastructure quality and AV perception performance are perceived by UK road users, and how these perceptions relate to trust, perceived safety, and behavioural intentions toward AV adoption. Methodology: Recent advances in multimodal transformers, world models (Ha & Schmidhuber, 2018), and large-scale representation learning (Sathyam & Li, 2025) provide an opportunity to rethink the AV perception pipeline. Building on these, this project will develop a customised foundation model to jointly encode static and dynamic driving context (e.g. road geometry, surface quality, visibility, and environmental conditions) using standard onboard sensors (cameras and LiDAR). While the student retains methodological flexibility, the approach will comprise three main phases: 1. Developing a computer vision multimodal foundation model to learn unified scene representations from raw video and LiDAR. 2. Extending these representations to forecast short-term environmental states and potential hazards, producing calibrated uncertainty for safety-critical reasoning. 3. Designing a UK survey to assess perceptions of AV safety, trust, and readiness under varying infrastructure and performance. The survey will capture perceived road quality, AV’s capability in risk perception (from Phases 1–2), and users’ willingness to use or interact with AVs. Datasets: The project has access to various multimodal datasets, including synchronised camera and LiDAR across diverse conditions, provided by Stellantis, our industrial partner, and also public datasets (e.g. nuScenes, Waymo Open) for large-scale training, robustness analysis, and domain generalisation User perception will be collected through bespoke UK-based surveys, ensuring alignment with national infrastructure characteristics, regulatory context, and public attitudes towards automated vehicles. Impact: This research will deliver a timely UK impact by addressing an important barrier to safe, scalable AV deployment: alignment between AV capabilities, infrastructure conditions, and public trust. By jointly evaluating infrastructure and system performance alongside user perceptions, the project will provide evidence to inform decisions on prioritising technological innovation, infrastructure investment, and public engagement. Collaborating with industrial partners, the research will develop deployable methods for perception robustness and uncertainty awareness and inform approaches to human-centred validation and safety assurance. Outcomes will also support evidence-based policy to enable safer AV trials, strengthen public trust, and contribute to long-term economic benefits from AI-driven mobility. References: 1] Tengilimoglu, O., Carsten, O., & Wadud, Z. (2024). Are current roads ready for highly ‎automated driving? A conceptual model for road readiness for AVs applied to the UK city ‎of Leeds. Transportation Research Part A: Policy and Practice, 186, 104148.‎ ‎[2] RAC Survey (2026). Vanishing act: vital road markings are rapidly disappearing from ‎Britain’s roads.‎ ‎[3] ALARM Survey (2025). “30th Survey: Key facts and findings on the condition of local ‎roads in England and Wales. ‎ ‎[4] Sathyam, R. & Li, Y. (2025). "Foundation Models for Autonomous Driving Perception: ‎A Survey Through Core Capabilities," in IEEE Journal of Vehicular Technology, vol. 6, pp. ‎‎2554-2582.‎ ‎[5] Rasouli, A., Kotseruba, I., & Tsotsos, J. K. (2018). "Understanding Pedestrian ‎Behavior in Complex Traffic Scenes," in IEEE Transactions on Intelligent Vehicles, vol. 3.‎ ‎[6] Chen, L. et al. (2024). “End-to-End Autonomous Driving: Challenges and Frontiers”, ‎IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 46. ‎ ‎[7] Ha, D., & Schmidhuber, J. (2018). Recurrent World Models Facilitate Policy Evolution. ‎Advances in Neural Information Processing Systems, Vol. 31.‎

<h2 class="heading">How to apply</h2>

Formal applications for research degree study should be made online through the <a href="https://www.leeds.ac.uk/research-applying/doc/applying-research-degrees">University's website</a>. Please state clearly in the research information section that the research degree you wish to be considered for is Foundation Models for Context-Aware Perception and User Trust in Autonomous Vehicles on UK Roads as well as <a href="https://environment.leeds.ac.uk/transport/staff/9408/dr-mahdi-rezaei">Dr. Mahdi Rezaei </a>as your proposed supervisor. If English is not your first language, you must provide evidence that you meet the University's minimum English language requirements (below). As an international research-intensive university, we welcome students from all walks of life and from across the world. We foster an inclusive environment where all can flourish and prosper, and we are proud of our strong commitment to student education. Across all Faculties we are dedicated to diversifying our community and we welcome the unique contributions that individuals can bring, and particularly encourage applications from, but not limited to Black, Asian, people who belong to a minority ethnic community, people who identify as LGBT+ and people with disabilities. Applicants will always be selected based on merit and ability.

<h2 class="heading heading--sm">Entry requirements</h2>

Applicants to research degree programmes should normally have at least a first class or an upper second class British Bachelors Honours degree (or equivalent) in an appropriate discipline. A Master's degree in engineering, computer science, AI, data analytics, transport studies, or other relevant fields is desired.

<h2 class="heading heading--sm">English language requirements</h2>

The minimum English language entry requirement for research postgraduate research study is an IELTS of 6.0 overall with at least 5.5 in each component (reading, writing, listening and speaking) or equivalent. The test must be dated within two years of the start date of the course in order to be valid. Some schools and faculties have a higher requirement.

<h2 class="heading">Funding on offer</h2>

We are offering a fully funded scholarship to study the project Foundation Models for Context-Aware Perception and User Trust in Autonomous Vehicles for UK Road Environments, at the Institute for Transport Studies, Univeristy of Leeds for one UK status candidate. The funding covers UK tuition fees as well as a UKRI matched stipend (currently £20,780 in 2025/26) per year, subject to satisfactory progress. Eligibility Criteria Applicants must be eligible for UK (Home) fees/funding. If you are unsure whether you are eligible for UK fees/funding, please see our <a href="https://www.leeds.ac.uk/undergraduate-fees/doc/fee-assessment">fee assessment page.</a>

<h2 class="heading">Contact details</h2>

For further project information please contact the Mahdi Rezaei: <a href="mailto:M.Rezaei@leeds.ac.uk">M.Rezaei@leeds.ac.uk</a> For application advice or queries, please contact Environment PGR Admissions: ENV-PGR@leeds.ac.uk