This project has industrial collaboration, and is co-supervised by a member of staff at the UK Atomic Energy Authority (UKAEA).
Operating robots in nuclear environments presents unique challenges due to safety risks, complex manipulation tasks, and the need for precision. Effective operator training and autonomous robotic assistance are crucial to ensuring safe and efficient teleoperation in these extreme conditions. This PhD project addresses two intertwined goals:
- Improving Human Training: Developing adaptive haptic training strategies that help operators refine their skills through real-time skill estimation, multimodal feedback, and tailored trajectory guidance.
- Enhancing Robot Autonomy: Enabling robots to improve their own performance by learning from operator data, ultimately enhancing their ability to assist in complex tasks.
Key Scientific Questions:
- How can haptic training improve operator skill acquisition in nuclear teleoperation?
- Which trajectory features best capture expert skill, and how can these insights improve robot control strategies?
- How can bidirectional learning — from humans to robots and vice versa — be effectively integrated to improve both operator performance and robotic autonomy?
Our Approach:
The project will explore how learning from demonstration can be combined with data-driven analysis of trajectory data to capture the spectrum between expert and novice behaviours. Traditional approaches often rely on predefined summary statistics (e.g., average speed, smoothness), which may fail to capture the subtleties of operator skill in high-stakes nuclear teleoperation, and how skill is developed over time. Instead, this project will investigate richer representations of trajectories that account for constraints such as precise path-following, controlled force application, and timing requirements. The student will develop methods to track learning curves and quantify operator skill progression over repeated tasks, identifying points where adaptive haptic guidance can be most effective.
At the same time, insights from expert trajectories will inform robot learning from demonstration, allowing robots to emulate successful strategies and adapt their assistance to operator needs.
Outcomes:
By generating data tailored for nuclear teleoperation, the student will:
- Design adaptive haptic guidance methods that respond to operator skill levels.
- Identify trajectory features that characterise expert performance for training robots.
- Develop algorithms that allow robots to refine their control strategies based on observed human behaviour.
Collaboration:
The project will benefit from extending existing collaboration with RACE at UKAEA, providing insights from real nuclear operators and access to MASCOT, Dexter, Telbot platforms, as well as advanced haptic interfaces for observation, data collection and testing. Secondments and mentorship from industry supervisors will ensure strong practical grounding.
Prior Work:
This research builds on the supervisory team’s previous work in teleoperation and shared control:
- Odoh et al. (2024) – Workload assessment in robotic teleoperation (Scientific Reports) https://doi.org/10.1038/s41598-024-82112-4
- Ly et al. (2021) – Predictive haptic guidance in shared control (IEEE RO-MAN) https://ieeexplore.ieee.org/abstract/document/9515326
- Singh et al. (2020) – Haptic-guided teleoperation of collaborative robots (IEEE Transactions on Haptics) https://ieeexplore.ieee.org/abstract/document/8979376
- Kim et al. (2023 – Immersive Haptic Interfaces for Bilateral Teleoperation (IEEE SMC) https://doi.org/10.1109/SMC53992.2023.10393925.
- Naceri et al. (2021) – VR Interfaces for Robotic Teleoperation (Journal of Intelligent Robotic Systems) https://doi.org/10.1007/s10846-021-01311-7
Funding notes
This project is part-funded by a Community Studentship provided by the Fusion Engineering CDT, and hence the student will be based at the University of Nottingham, but should expect to engage fully with the 3-month full-time training programme in the Fusion Engineering CDT at the start of the course (October to December inclusive). CDT training will be delivered across the CDT partner universities at Sheffield, Manchester, Birmingham and Liverpool. The training course requires weekly travel to attend in-person training at these universities.
For further information about the CDT programme, please visit the CDT website at www.fusion-engineering-cdt.ac.uk or send an email to hello@fusion-engineering-cdt.ac.uk.
Candidate requirements:
- Due to the nature of the funding, the position is only available for candidates qualifying for UK Home fee status. Applications from non-UK candidates will not be considered.
- 1st or 2:1 academic qualification in Engineering or Physical Sciences or a related discipline, with expertise in fluid mechanics and heat transfer.
How to apply
- Please send an email with subject “CDT studentship: Adaptive Haptic Skill Transfer for Human-Robot Collaboration in Nuclear Teleoperation” to Ayse Kucukyilmaz ayse.kucukyilmaz@nottingham.ac.uk, attaching a cover letter, CV and academic transcripts.
- Incomplete applications will not be considered.
- Suitable applicants will be interviewed, and if successful, invited to make a formal application.
- Please note only shortlisted candidates will be contacted and notified.
Apply for this project here.
