The Nuclear Decommissioning Authority (NDA) is responsible for the safe and secure decommissioning of the UK’s 17 former nuclear fission sites. It has its work cut out. These are complex facilities, which date back to the 1950s in some instances, and were not designed with decommissioning in mind. Whilst there are many shared challenges across sites, they are far from ‘cookie-cutter’ designs.
Technology has always been a key element of addressing these diverse challenges, especially technologies that can be used in hazardous environments. But in recent years a number of ‘commercial-off-the-shelf’ technologies – particularly in general-purpose robotics and the digital sphere – have emerged.
These technologies provide ‘platforms’, highly functional and flexible tools, on top of which we can build solutions to decommissioning challenges, which can then be customised to the different needs of each site.
Robots to the rescue
Robots are one of the most visible examples of this ‘platform’ approach, notably the use of robotic quadrupeds, which first appeared in viable forms around 2020. These are highly versatile machines that can walk around complex environments that humans cannot safely enter, navigating debris and climbing stairs in a way that previous generations of robots could not. They can be fitted with all sorts of sensors, and their manipulator arms can grip tools and have the potential to perform human-like movements.
Through the Robotics and Artificial Intelligence Collaboration (RAICo) programme we have worked with decommissioning sites to adapt quadrupeds to tasks ranging from operating switches to loading scaffold pipes into a cutting station. We are currently working closely with the Remote Technologies Group at Sellafield Ltd to explore how a new 3D-printed swabbing tool and control system can be used on their site. RAICo is a collaboration formed to accelerate the deployment of robotics and AI to solve shared nuclear challenges in nuclear decommissioning and fusion engineering.
The NDA estate has used quadrupeds to survey the reactor bioshield at Trawsfynydd and create a 3D radioactivity map of a cell in a Fuel Cycle Area at Dounreay.
Similarly, nuclear sites are exploring ways to adapt autonomous and remotely operated drones and underwater vehicles to the challenges of a nuclear site.
None of these robots was purpose-built for nuclear, and they have many other applications, from oil rig inspections to disaster response. This is what makes them so valuable – their developers have done the hard work of creating a versatile robot, we adapt it for our needs. Many of those needs are replicated across the NDA estate, so can be addressed by a single robot type with relatively minor changes between sites.
Off-the-shelf does not just mean physical pieces of equipment like robots. Commercially available digital tools are just as important. Many generic platforms for building digital twins have become invaluable for understanding risk profiles of sites and planning safe and efficient decommissioning tasks. Mapping tools can take LiDAR and dosimetry data collected by robots and create radiation risk maps, which can then be integrated into these digital twins. Analytics tools can recommend optimal courses of action. Tools from the video game industry, such as Unreal Engine, are also proving powerful, allowing us to create hyper-realistic 3D environments, which provide safe places to train new operators or test new robot setups.
Adapting commercial technologies
The term commercial-off-the-shelf, or COTS, is often used to describe these physical and digital technologies, but a better term might be MOTS – modified-off-the-shelf. None of these technologies can go straight out of the box to deployment in a radioactive environment. They are platforms with useful functionality, which can be modified to suit specific decommissioning challenges.
Some modifications are about preparing the technology for the environment, for example through radiation protection and contamination prevention. Others are about creating new functionality for specific decommissioning tasks, such as upgrading quadruped control software so its manipulator arm can swab the curved surface of a waste drum.
Modifications can be significant. For example, RAICo has developed digital sandboxes for simulation, training and operations, which use Unreal Engine to create hyper-realistic 3D environments, combined with software developed in-house and built around the end users’ needs. Whilst not strictly off-the-shelf, this is a general purpose tool that mixes commercial and bespoke technologies, and can be adapted to a wide range of decommissioning use cases.
Ensuring commercial tech meet nuclear needs
Whilst commercial technologies are reducing an enormous amount of bespoke development work, they still need to work in the high stakes world of nuclear decommissioning. Ours is a highly regulated industry, and technologies must comply with site and regulator expectations on safety.
That means development of modifications must be done in close collaboration with end users on decommissioning sites like Sellafield, and working with them to perform rigorous testing of each development in sandboxed environments, before gradually progressing to non-active, then active demonstrations, and finally deployments. At each stage we need to show that risk is ‘as low as reasonably practicable’ (ALARP). That includes not only meeting site safety processes, but training, adoption, and workforce support.
That is only possible through close collaboration with technology deployment groups on the sites – the people who understand the site-specific challenges and have the knowledge and expertise to lead the on-site deployments. And testing is not just physical. Integration with IT and cyber requirements is essential. Ensuring data interoperability between vendor platforms and NDA estate systems is a persistent deployment challenge. And once it’s proven at one site, learnings need to be shared, and technology adapted for different environments. This is far more efficient than developing technology from scratch, but it is still a job to ensure each technology goes through the site’s proper safety processes. Nonetheless, learning from one site vastly speed up deployments on subsequent sites.
‘Off-the-shelf’ is therefore something of a simplification. Adaptation is required in various degrees to make these commercial technologies viable for nuclear. But the point is that general-purpose innovations have arisen in recent years that hold enormous potential – well beyond what any one nuclear site could develop in house.
A moment for platform technologies
What is driving this ‘platform’ approach to solving shared challenges? A big driver is of course technological possibilities. Recent advances in robotics and digital technologies make them more practical. The current generation of commercial robots have more fine-tuned controls, often with haptic feedback, and high degrees of freedom, enabling dexterous tasks with high reliability.
Advances in reinforcement learning enable robots to learn from tasks and improve, without laborious human programming for every scenario. Advances in digital twins and VR (much driven by the video game industry) have hugely increased what is possible through simulation and data driven analysis.
More mundanely, this is all much easier in a world with the IT backbone to support remote working, online communication, and secure data sharing between sites.
But NDA strategy has also been critical. As technology makes more possible, the NDA has emphasised ever greater collaboration, open innovation, and use of common digital tools. The funding of the RAICo programme since 2022 by the NDA and others is of course a prime example of this ambition.
Can we continue to benefit?
This article has described a few examples of how commercial robots and digital technologies are helping solve decommissioning challenges. Impressive though progress has been, it is just the start. In the commercial world, such technologies are advancing at a mind-boggling pace.
Embodied intelligence – the integration of AI models with physical robotic systems – will allow robots to follow voice commands and learn about their environment. As an example, the Figure 03 humanoid robot, launched in October 2025, claims to be capable of tasks comparable to those performed with human hands, and can learn from watching and listening to people. We are keeping a close eye on such innovations. Meanwhile, AI is advancing in all sorts of areas, from pattern recognition to data analysis, to document management, all of which hold promise in decommissioning. Will decommissioning continue to benefit from all this innovation? Will this collaborative approach to investigating and deploying new technologies still hold as tech enters new territories?
AI powered robots that make their own decisions in a nuclear environment will of course ring alarm bells. That doesn’t mean there aren’t safe ways to deploy them, though they may be different to how things are done now. Just as we must keep an eye on commercial innovations, so must we review our own processes to check they still serve us as technology changes.
We will probably always be a little behind fast-paced industries like self-driving vehicles, which is no bad thing. Those industries will provide lessons on how to assess new generations of technology.
When confronted with a potentially game-changing technology, we will need to ask ‘how can we benefit from this safely’ rather than ‘how does this fit into our current processes’. That needs vision and flexibility at all levels, and clear communication to everyone involved. Sellafield Ltd’s decommissioning programme expects to run for 100 years – in that time commercial technologies will advance in ways we can’t even imagine. If technological change is adopted adequately, there is a high potential of reducing the forecasted 100 years.
The promise of commercial innovation
Commercial technology holds enormous promise for nuclear decommissioning, as well as for the emerging fusion industry. Progress in recent years has shown we can be ambitious in embracing the possibilities of new off-the-shelf technologies, without compromising safety. If we continue to do so, we can accelerate nuclear decommissioning, enhance safety, and make huge savings to the cost of UK nuclear decommissioning, and so to the taxpayer.
