The challenge

Nuclear decommissioning and the safe disposal of nuclear waste is a global problem of enormous societal importance. The UK alone contains 4.9million tonnes of legacy nuclear waste, representing the largest, and most complex, environmental remediation project in the whole of Europe.

UK nuclear clean-up is expected to take over 100 years to complete, with estimated costs as high as £220billion. At least 20% of this expenditure must involve remote interventions using robotics, because the materials and environments are too hazardous for human workers, even wearing protective air-fed suits.

The legacy waste problem

Much of the legacy nuclear waste is very old. Nuclear operations in USA and UK began in the 1940s, and greatly accelerated in both countries following the first USSR atomic bomb test in 1949. The UK pioneered peaceful use of atomic energy, with the world's first industrial scale civil nuclear power station coming online at the Sellafield site in 1956. Many of the UK’s nuclear facilities, including test reactors, storage areas and fuel re-processing plants, are now in urgent need of decommissioning, with the UK Nuclear Decommissioning Authority (NDA) describing some facilities, particularly the ponds and silos at the Sellafield site in Cumbria, as intolerable risks.

In high gamma radiation environments, no human entry is possible at all. In contrast, a large part of nuclear decommissioning involves dismantling and waste handling in alpha radiation environments (e.g. facilities contaminated by plutonium dust). Currently such work is done manually by human workers wearing air-fed plastic suits.

Figure 1. Copyright Sellafield Ltd.

Fig. 1  shows a typical alpha-decommissioning operation at Sellafield, wearing an air-fed plastic suit underneath a thick leather overcoat to protect the plastic suit from punctures, and many layers of gloves. Workers must handle power tools to cut contaminated lab equipment and place it in storage containers. Working in such conditions is stressful, physically demanding and poses risks to human safety. Unless advanced robotics alternatives can be developed, it is expected that 1 million suited human entries will be required to complete decommissioning. Despite an extremely rigorous safety culture, statistically it is highly likely that such a huge number of human entries will result in a number of serious injuries or deaths.

Repackaging storage containers

Record keeping in the early days was not rigorous by modern standards, and there exist legacy waste storage containers with unknown contents or contents of mixed contamination levels. At the Sellafield site alone, 69,600 m3 of legacy waste must be cut open, examined, and re-packaged into 179,000 modern storage containers. To avoid wastefully filling expensive high-level containers with low-level waste, many old legacy containers must be cut open, and their contents “sorted and segregated”. This engenders an enormous requirement for complex remote manipulations which will have to be robotic.

Fig 2: Left, 1949 MSM device designed by R.C.Goertz. Middle, similar devices used today. Right, a human operator controls a Brokk robot

Robots in nuclear environments

Perhaps surprisingly, there has so far been remarkably little use of robots in the nuclear industry at all. The vast majority of remote manipulation is performed by an aging workforce (mean age 55 in UK) of highly skilled experts, using mechanical Master-Slave Manipulators (MSM), Fig. 2. Such devices date back to at least 1949  and have changed little in design since the 1960s. Where robots have been deployed, these have predominantly been directly teleoperated in rudimentary ways. Fig. 2 (above, right) shows a human operator tele-operating a Brokk robot arm. Such robots, widely trusted in the nuclear industry due to their ruggedness and reliability, do not have joint-encoders, and no inverse-kinematics solving is possible to enable Cartesian work-space control via a joystick. The operator is looking at the robot through a 1.6 m thick lead-glass window, with very limited situational awareness or depth perception. He controls each joint directly with individual switches, while guessing the inverse kinematics (how the end of the arm will move given the movements of each joint) from experience. Such control methods may seem very archaic to lab-based academic robotics researchers, but are standard and common in the nuclear industry.

Sellafield Ltd are now committed to building a new Box Encapsulation Plant, where large robot arms will be used to import and cut open decades-old waste containers (often with unknown contents), sort through the contents and repackage waste items into safe, modern storage containers. Currently a robotic test-bed has been created to explore how robots can be controlled to achieve this.

Using robotics to help solve the problems

To overcome these problems, the University of Birmingham Extreme Robotics Lab is leading the €6.8million H2020 project RoMaNS (Robotics Manipulation for Nuclear Sort and Segregation), in partnership with the nuclear agencies of UK (NNL Ltd) and France (CEA) and support from Sellafield Ltd. The projects section gives further details on how RoMaNS is developing state-of-the-art robotics and AI methods to assist the Sellafield Box Encapsulation Plant development.

RoMaNS technologies include advanced exoskeletons which allow the human to, not only control a remote slave robot, but also feel the forces encountered by the remote robot’s fingers when it grasps an object. We are also developing fully autonomous robotic grasping and manipulation methods, where the robot arm is controlled by an AI and guided by a robotic vision system. We are also exploring how the AI and human can collaborate to control a remote robot via a “shared control” system.