Robots have an obvious role to play in radioactive environments on nuclear decommissioning sites. They can handle radioactive materials or perform other activities in places that are hazardous for humans to enter, or where humans can only do so with significant protection. However, robots are not always the completely ‘hands off’ solution one might think. Sometimes their human programmers need to enter the active environment, under strict safety protocols, to make upgrades to processes related to their software, or to set new paths. This requires those programmers to don an air-fed suit, with two sets of gloves and a respirator, so they can enter the cell and work on the robot. It’s uncomfortable, costly, and not without risk.
Solutions to this problem are arising in the form of simulators – highly accurate virtual representations of the robot’s environment. These provide a safe ‘sandbox’ environment on which new setups and paths can be trialled without risk, with optimal settings then transferred to the robot’s controller. That could save days of time in robot upgrades, reduce discomfort and radiation exposure risk, support robot operator training, and dramatically cut the cost associated with humans entering a contaminated cell.
A Pioneering robot simulation at Sellafield
The first such simulator being deployed at Sellafield is the Pile Fuel Cladding Silo Operational Simulator (PFCS OpSim), developed with Sellafield Ltd by a team at the Robotics and Artificial Intelligence Collaboration (RAICo) – a collaboration between the UK Atomic Energy Authority (UKAEA), Nuclear Decommissioning Authority (NDA), Sellafield Ltd, the University of Manchester and more recently AWE Nuclear Security Technologies.
The simulator is of a Waste Container Handling Facility (WCHF), a concrete room where robots process waste containers. This is housed within Sellafield’s Pile Fuel Cladding Silo, a legacy waste storage facility and priority for decommissioning on the site, built in the 1950s to store radioactive cladding from the nuclear fuel used by the very first reactors on the Sellafield site. Today, it is one of the most hazardous facilities on the site and needs to be emptied so the waste can be repackaged into safe, modern storage.

Within the WCHF is a large 6-axis industrial robot arm, which unbolts empty containers, sends them to be filled with waste, then securely bolts the returning containers for onward transfer. The arm also swabs the outside of the container to check for contamination and puts the swab in a deposit box for analysis.
When upgrades are needed, such as software modifications or replacing tools like the swab gripper or nut runner, engineers or maintenance personnel have to take the robot out of service, and robot programmers need to suit up in protective gear to make changes. Positional adjustments require the robot to be positioned using a KUKA Smartpad and then each position has to be ‘touched up’, where the robot software records the co-ordinates.

Changes need multiple test runs, often with several human re-entries to refine the setup. To keep them safe, each entry needs planning, clearance and monitoring, and support teams on hand.
A typical upgrade can take 10 days of testing, which is slow work since the discomfort of thick protective gear limits time humans can spend on the upgrades in any one session. Such changes can happen due to breakdowns or off-normal conditions, and can be more frequent during modification periods. Each entry increases the risk of radiation exposure or contamination, and the risk of damaging the robot or in-cell equipment from human error, whilst trial runs for new setups create downtime.
The PFCS OpSim could solve many of these issues. It provides a simulation that allows operators to visualise the robot within a realistic but virtual WCHF. That allows new programmes and other changes to be tested and validated in the simulation before deployment into the facility. Once validated, software updates could be transferred digitally to the robot.
Creating a waste handling facility simulation
The project began with a visit by Sellafield Ltd to the UKAEA’s Remote Applications in Challenging Environments (RACE) facility in Culham, where the team saw a virtual simulator of the two-armed ‘MASCOT’ manipulator used in the JET fusion machine.
Recognising its potential, conversations began around creating something similar for Sellafield, which ultimately led to the project with RAICo. This is exactly why RAICo was set up, to allow members to share challenges and work together on solutions.
To map the waste handling facility, a LiDAR scanner was used to collect precise positional data of everything in the space. Those data points were sent to RAICo alongside Sellafield Ltd’s CAD models from the original facility design.
RAICo used these assets to build a precise digital replica of Sellafield’s PFCS that mirrored its exact layout. This was done using RAICo’s in-house-developed 3D visualisation software platform, RHOVR (Remote Handling Operations Virtual Reality), which also harnesses the Unreal Engine – better known for its use in video games – to create hyper-realistic 3D environments.
Next, an off-the-shelf simulator of the robot’s hardware and software was acquired and integrated. This was the most technically challenging element of the project – effectively adding a virtual replica of the robot into the virtual PFCS environment.
The completed product meant users could programme the robot simulator (as they would the real robot), and see the virtual instance of the robot run that programme in a photorealistic virtual version of the PFCS. Programmers can then visually assess outcomes – such as whether the robot takes the most efficient route, whether it still performs actions such as swabbing the waste container correctly after a new tool has been added, and if it knocks into anything on the way – before deploying the changes onto the real robot control software.
The simulator was validated over six months, and demonstrated at the RAICo1 facility in Whitehaven, Cumbria, in March 2025.
From demonstration to deployment
The demonstration was well-received, and a working version was subsequently transferred to Sellafield’s Engineering Centre of Excellence. The tool is still new and its primary tangible value to date has been training, allowing operators to now get up close and personal with the robot virtually, instead of watching it through CCTV screens, which lack reliable depth perception. It also provides a useful tool for demonstrating Sellafield’s robotics capabilities to a wide range of internal and external stakeholders.
But the real value is in its potential to programme the waste handling robot virtually. As with any safety critical applications, there are processes to follow. Sequences need to be rigorously tested to confirm they are suitable. Sellafield Ltd also needs to establish the rigorous processes and cybersecurity measures needed to enable the programme to be transferred from the simulator – which operates in a sandbox environment, air gapped from Sellafield’s IT systems – onto the robot. But the team is optimistic this can be navigated, and the first simulator-tested programme will be running by next year.
The benefits of robot simulation
Virtual simulations are not perfect, and some in-cell fine tuning will still be needed. But, all being well, the benefits of such a simulator should be significant.
The potential time saving is the most obvious advantage. Sellafield Ltd estimates that a best-case scenario would see the number of days human programmers need to be sent into cells cut from 10 to two. That frees up engineers, but also reduces the cost of sending people into a cell in full air-fed PPE.
Risk reduction is just as important. Less time in the cell means less exposure to hazards. This supports ALARP (As Low as Reasonably Practicable) – a UK safety principle requiring decommissioning sites to keep risks to people as low as possible while at a reasonable cost/effort. The simulator also enables more experimentation. New sequences can be trialled to explore the robot’s capabilities, for example, creating a sequence to pick up a tool should it become dislodged.

All of this excites the robot programmers at Sellafield, who look forward to spending more time at a computer, and less time in uncomfortable suits going through laborious programming motions.
Finally, the simulator is a powerful tool for training for the robot’s operators. Because it mirrors the real cell visually and operationally, the simulator is ideal for SQEP (Suitably Qualified and Experienced Person) training and familiarisation, maintenance rehearsals, and pre-job briefs. Instead of looking through CCTV to make assessments, operators can step inside an interactive, risk-free version of the environment, making learning faster, safer, and more accurate.
The wider potential of Sellefield simulation
The PFCS Operational Simulator is a first of its kind at Sellafield but hopes not to be the last. Early candidates for expansion include the Box Encapsulation Plant (BEP) and the Encapsulated Product Store-Waste Transfer Route (EPS-WTR), where robots carry out waste manipulation, size reduction, bolting, swabbing and other precision handling tasks in areas with little or no human access. In BEP, for example, the level of hazard in some cells will make human entry impossible, and operators will have to rely on camera feeds with limited depth perception to carry out operations. A simulator here would allow safe and accurate programming and troubleshooting, as well as greater availability of training.
With two similar robots already in operation, and at least another 10 coming once BEP is online, there is clear potential for scaling robot simulators across Sellafield. What’s more, other NDA estate sites face similar challenges, and the technology could serve as a blueprint for robot simulations across many sites.
The value of collaboration
The ability to create this simulator rests on the combination of expertise brought together through the RAICo partnership. Sellafield Ltd had the knowledge of its own site and has long recognised the value of advanced digital simulation, but did not have the specialist robotics and 3D visualisation capabilities in-house.
RAICo, with its blend of nuclear engineering, robotics, and software development capabilities had developed tools and expertise to solve this problem, though working collaboratively with Sellafield to understand the specific on-site setups and challenges was just as important. The result is a robust tool, delivered faster and more cost-effectively – a win for Sellafield Ltd and the taxpayer.
Commenting, Rav Chunilal, Head of Robotics and Artificial Intelligence at Sellafield Ltd, said “this collaboration with RAICo is accelerating our mission and setting a new benchmark for innovation in nuclear decommissioning which can be repeated across Sellafield and other NDA group operating companies.”
Simulations are not entirely new. They are proven, well-understood, and widely used in industries like aerospace. However, they are still novel in nuclear decommissioning which has been slower to embrace them due to strict safety protocols around deploying new technologies. This project is changing that and, once proven, Sellafield Ltd hopes to accelerate the use of simulators quickly.
This project is, of course, part of a wider drive across the Sellafield site and the NDA estate to embrace digital tools to improve safety and efficiency, examples of which range from UAVs and robot quadrupeds, to virtual training and glovebox automation.
Nuclear decommissioning is enormously complex and must be done to extremely high safety standards. Today’s decommissioning sites involve a wide range of remediation tasks in a wide variety of challenging spaces – none of which were designed with modern technologies in mind. Robotics and digital technologies hold obvious potential to accelerate this mission. There is no single solution that will solve everything – the challenges and risks are too diverse. Sites like Sellafield must deploy robots and AI with precision, targeted challenge-by-challenge, with each deployment making nuclear decommissioning tasks safer and more efficient – collectively accelerating the process and reducing the risk and cost of nuclear decommissioning.