The Surrey Technology for Autonomous Systems and Robotics (STAR) Lab is currently conducting a four-year research project into the potential uses of robotics within the nuclear sector, funded by the UK Nuclear Industry and Engineering & Physical Sciences Research Council (EPSRC)s Impact Acceleration Account. The team is working in collaboration with Sellafield and the National Nuclear Laboratory (NNL) to examine how robotics could be used within nuclear decommissioning specifically.

The project, known as Automated Robot Waste Processing, began last year and is making fast progress. As the worlds nuclear fleet ages an increasing number of reactors will need to be decommissioned, a traditionally expensive and hazardous process. The STAR Lab is hoping to help by using robust robotics running autonomously to dismantle and clear nuclear sites.

Their robots combine a number of technological advancements to suit nuclear decommissioning, taking inspiration from space robotics. They include high-performing visual imagining systems, along with robotic arms, combinations of which are entering the testing stage of the project.

STAR Labs main focus since its inception in 2007 has been space robotics, and the new research reflects this. But according to Professor Yang Gao, who leads the Surrey team, nuclear sites and space actually have a lot in common. Space is remote and hostile, and overcoming these problems has provided many clues for working within nuclear sites.

One area [of comparison] is autonomous software, as you can imagine space missions tend to deal with locations that are very, very far from earth so remote operation in many scenarios is not feasible, says Gao. So were developing software that can potentially do the models which involve the mission goals and then make decisions. Essentially autonomous software.

"Space is remote and hostile, and overcoming these problems has provided many clues for working within nuclear sites."

Because of the very high requirements of space missions, software has to be very reliable and robust, so we have techniques while designing autonomous software to take into consideration robustness and reliability.

The nuclear decommissioning process, which can cost anything from $13m to $19m, currently takes years and necessitates humans working to dismantle the site and decontaminate the waste. This is often made more challenging by the remote location that nuclear plants are often purposefully situated in, and the hazardous residual radiation.

Another aspect of comparison is probably more in line with what is happening with the hardware of the robotics systems, says Gao. It needs to be radiation proof, so the mechanisms we develop, the different actuators, the different censors, they all need to be able to cope with that sort of environment. These challenges are very much shared with the space environment, there's a great synergy in that respect.

The obvious and most important benefit of robotics in nuclear decommissioning is removing humans from harms way. The environment is classified as an extreme environment as there are many hazards, so using humans to perform those tasks is risky, says Gao. Using machines, using autonomous systems to replace humans from that sort of environment is definitely more desirable.

This is particularly relevant for nuclear sites which have deferred dismantling, and have instead enclosed the site for anything from 40 to 60 years. When such a site is eventually dismantled, surveys must be undertaken to ensure that the radiation level has dropped as much as expected and is below 25 millirem per year, the level required for the site to be redeveloped for other uses. When entering a site, however, it is difficult determine the amount of residual radiation, and whether it will affect the humans undertaking the survey. Using robotics would remove this danger.

There are other benefits to increased use of robotics, too, such as accuracy and productivity. Sometimes there are some anomalies in sites that, although they are visible the human eye, could [be] quite hard to see, because either they are quite small or because some of the facilities are actually very homogeneous, says Gao. For the human eye to react to that sort of homogeneity is not easy. The machine would do a better job.

The robots are equipped especially for this, with 3D light scans and cameras. We can also introduce measuring techniques such as deep-learning to allow us to very reliably map the site and detect what has changed, says Gao. Even though this change could be very, very small and probably quite difficult for human eyes to detect, with our learned-visual model we can achieve this anomaly detection reliably.

Decommissioning nuclear sites requires a lot of monotonous dismantling, tasks well suited to robots whose productivity is not deterred by fatigue or boredom. For the decommissioning task we envision using robotic arms, which could definitely improve productivity because it's actually just repetitive work, says Gao. A machine cannot be annoyed by that because there is no emotion, they will actually be able to keep to the same rate of productively.

The project has already overcome challenges, but Gao is expecting more as integrated trials begin at the NNLs testing facility. We'll never stop facing challenges I'm afraid that's just reality. In the research phase, and currently in the development phase, there have been a lot of challenges in terms of getting the actual hardware working at the actual conditions that we anticipated.

These challenges were created not just by the environment but also by the integration of technologies. Once you start putting different hard components together, integrating different software packages, you start to find some real problems, says Gao. Some of them are due to communication issues, and some of them due to signal issues within some of the components.

"We envisage that challenge might come from the culture of such a traditional sector."

Whilst this research collaboration only began last year, it relies on theoretical elements Sellafield and STAR have been working on for years; it is often in fitting these aspects together that things do not run smoothly.

A further challenge may be the nature of the nuclear industry, as change can take a long time to be accepted. I think we envisage that challenge might come from the culture of such a traditional sector. Of course people need to get used to things, even if they know, they agree and they appreciate that this is coming in to help, says Gao. But you can imagine when you introduce new machines into a sector people are not used to, there will always be some cultural impact.

Professor Gao and her team are working on nuclear decommissioning but there are several projects running parallel to each other tackling different elements of the nuclear industry. The current project is to try and solve decommissioning problems, so waste management, but I also want to mention that at the same time we're running projects looking into other areas, for example atomic fusion, says Gao.

As such she believes that robotics is set to play a much bigger role in the nuclear industry in the future. Gaos team is currently running subsection trials, with full integrated system trials coming up in the next couple of months, and hopefully a full test within an actual nuclear site by the end of the year. If successful this technology could soon become commonplace in the nuclear industry.

Looking ahead, the success of robotics also relies on training a new generation of engineers. At the moment we work closely with engineers and scientists at NNL, but what we want to do next, probably towards the end of this year or early next year, is to help also train the site engineers and operators, says Gao. This will eventually allow STAR to step back, and allow others to use the robots for decommissioning, site maintenance and other areas.

The STAR Labs research is progressing at an impressive rate thanks to collaboration. At this rate, it seems that robots are likely to become a key tool in nuclear decommissioning.

Original post:

Nuclear decommissioning: sending the robots in - Power Technology