University of Oxford
| Qualification Type: | PhD |
| Location: | Oxford |
| Funding for: | UK Students |
| Funding amount: | Funding for this project is provided by the EPSRC and Sellafield |
| Hours: | Full Time |
| Placed On: | 3rd February 2026 |
| Closes: | 4th May 2026 |
PhD Studentship available on the RAINZ CDT programme at the University of Oxford.
Project Overview
- Project title: Long-Term 3D Change Detection of Industrial Facilities
- Supervisors: Prof Maurice Fallon
- Cohort research challenge: Long-term autonomous monitoring and maintenance of assets
- Year 1 MSc Course: MSc Robotics
- Year 2 – 4 PhD Location:University of Oxford
Abstract: Nuclear facilities require regular inspection to monitor specific locations (for example valves, levers and read-offs) but also generally around the facility to identify the presence or disappearance of plant or equipment as well as more subtle changes such as the growth of a crack or the spread of corrosion.
Thus, long-term 3D mapping and change detection is of great interest for remote maintenance. Previous research at the Oxford Robotics Institute has developed a real-time 3D LIDAR mapping system (called VILENS) which is capable of surveying large multi-floor buildings in the time taken to walk around the site. This system has been demonstrated in various nuclear facilities in the UK.
However, the identity of the objects which caused the change in these clouds is not captured by this approach. Furthermore, the minimum identifiable change discernible from LIDAR is rather large – only down to about the size of a drinks can.
In this project we will use emerging 3D visual sensing technologies such as implicit neural rendering (NeRF, Gaussian Splatting) and Geometric Foundation Models (MapAnything, VGGT) to infer 3D change from a much simpler sensing source – the humble camera. The project will pair this direction with fundamental research into multi-session visual SLAM and semantic segmentation to develop a scalable, 4D (3D plus time) representation of an industrial facility suitable for verification of condition or detection of change over time.
About us
The EPSRC Centre for Doctoral Training in Robotics and Artificial Intelligence for Net Zero is a partnership between three of the UK’s leading universities (The University of Manchester, University of Glasgow and University of Oxford).
Robotics and Autonomous Systems (RAS) is an essential enabling technology for the Net Zero transition in the UK’s energy sector. However, significant technological and cultural barriers are limiting its effectiveness. Overcoming these barriers is a key target of this CDT. The CDT’s research projects focus on how RAS can be used for the inspection, maintenance and repair of new infrastructure in renewables (wind, solar, geothermal, tidal, hydrogen) and nuclear (fission and fusion), and to support the decarbonization of existing maintenance and decommissioning of assets.
Funding:
This 4-year studentship covers tuition fees at Home student rate, a tax-free stipend, and a Research Training and Support Grant. Funding for this project is provided by the EPSRC and Sellafield.
Eligibility
Applicants should hold a First or strong Upper Second-class honours degree (2:1 with 65% average), or international equivalent, in Engineering, Computer Science, Physics, Mathematics, or a related discipline, and demonstrate evidence of programming experience.
How to Apply
Applications should be submitted through the RAINZ CDT website via the above 'Apply' button by 13 February 2026, where further information about the CDT is also available.
Email: rainz@manchester.ac.uk