Robotic solution to aid nuclear decommissioning

An EPSRC-funded project led by Sussex University aims to develop a robotic solution to address nuclear decommissioning challenges.

The robotic solution aims to help with clean-up at the Fukushima Daiichi Nuclear Power Plant in Japan. Image credit: Shutterstock

In particular, the project’s goal is to aid in both the clean-up of the primary containment vessel at the Fukushima Daiichi Nuclear Power Plant in Japan, and in decommissioning operations at Sellafield nuclear site in the UK.

Due to radiation and environmental uncertainties, standard robotic manipulators used in industrial applications have been recognised as unsuitable for nuclear decommissioning scenarios.

Principal investigator Dr Romeo Glovnea — head of Sussex University’s Department of Engineering and Design, and the department’s Robotics and Mechatronic Systems Research Group — explained that through proposing a robotic manipulator based on a novel Variable Impedence Actuator (VIA) the group hopes to develop a ‘collision-safe’ robotic solution that can navigate and interact in cluttered environments without malfunction or added sensor feedback.

“We expect that this manipulator will solve the challenges associated with limited sensor availability due to radiation hazards: safely interacting with sensitive materials, and resilience in events of communication and sensor failure or during misoperation in teleoperation protocols,” Dr Glovnea told The Engineer.

The approach proposed for the manipulator is a novel Continuously Variable Transmissions (CVT)-based VIA, an approach used in automotive applications, which can automatically change its internal configuration and transmission ratio in response to a variation of the external torque, Glovnea added. 

Rescue robot set for Sellafield nuclear site

Toshiba robot set for new Fukushima reactor mission

“For example, if the limb of a robotic manipulator hits an obstacle while moving, the new actuator will immediately reduce the speed of that limb bringing it to a halt, thus avoiding damaging the obstacle or itself,” he said.

“Another way we intend to use this VIA is in a robotic gripper that will be attached to the end of the manipulator to handle objects. When the fingers close around an object and start squeezing it, the force [torques] transmitted to the CVT increase; the CVT will react by reducing the output speed such that at certain values of those forces, the fingers will stop squeezing … thus avoiding crashing the object or damaging the gripper.”

The project is a collaboration between researchers from Sussex’s Robotics and Mechatronics Systems Research Group, and the University of Tokyo’s Service Robotics Laboratory, Department of Precision Engineering, led by Prof. Hajime Asama.

Whilst the Sussex team will focus on designing a system specifically for nuclear decommissioning at Sellafield, the Tokyo team’s system will be adapted to tasks specific to the Fukushima Daiichi Power Plant.

“This will be a key challenge because we need to design a system which is flexible enough to be adapted and operate successfully in both environments,” Glovnea commented. 

“At Sellafield the tasks we have in mind cover mainly grasping, sorting and removing contaminated objects of various sizes and shapes while providing safety in case of unintended interactions with pipework and various obstacles. 

“At Fukushima the manipulator will have to operate inside the primary containment vessel; the access inside the vessel is very restricted by size thus there are limits to the overall dimensions of the device. The main tasks here will be retrieving and removing pebble-like fuel debris.”

Artificial intelligence control techniques will be employed based on reinforcement learning using neural networks, and the robotic concept of ‘embodied intelligence’, to improve the robot’s accuracy and resilience whilst limiting the use of sensors to a minimum.

The team aims to have a model of a robotic manipulator tested in laboratory conditions around the middle of 2023, with demonstrations and testing to be carried out in the UK and Japan until the end of Spring 2024 when the system is planned to be operational.