Quadruped robots, such as Spot, which are controlled remotely via a handset and video feed, are a valuable tool for use in hazardous areas on nuclear decommissioning sites. The four-legged robots have the manoeuvrability to navigate stairs and rubble, whilst their body and manipulator arms can be mounted with various grippers and sensors for carrying out inspections, moving objects around, operating switches and performing tasks that may pose significant risks to humans.
However, while such robots remove many of the challenges of sending humans into hazardous environments, they still take time for operators to get to grips with.
End users at Sellafield report that it can take a new operator three days of training to get completely familiar with the controller handset and the full range of movements of the robot and its manipulator arms, and up to two weeks to become skilled enough to confidently use the robot. Melanie Brownridge, Chief R&D Officer at the Nuclear Decommissioning Authority (NDA), explains: “having the right skills in place is also key to accelerate the deployment of these transformational technologies.”
Nonetheless, most operators train using the robots themselves, and since most sites have a limited number of robots, training operators at scale is slow. Furthermore, every hour the robot is used for training diverts it from executing the valuable tasks for which it was designed. It also creates risks – an unpracticed user may accidentally damage the expensive robot or its environment.
To reduce the cost and risk associated with training robot operators using real robots, RAICo, the Robotics and Artificial Intelligence Collaboration which is a collaboration between NDA, Sellafield, AWE Nuclear Security Technologies, the University of Manchester and the UKAEA, has developed the Quadruped Familiarisation Tool (QFT). This allows multiple operators to be trained in parallel using simulations rather than real robots.
Simulation-led training
Using an off-the-shelf controller similar to the quadrupeds’ own control device, the system replicates robot behaviour in photorealistic simulation environments. The system reflects the challenges of navigating nuclear decommissioning sites. It is also possible to build precise digital replicas of specific facilities, letting operators train in exact copies of the working environment in a safe and controlled manner.
Interactive tutorials guide operators through everything from basic motions like standing and walking to advanced tasks such as using manipulator arms, operating cameras, climbing stairs, and even recovering from a fall. Finally, progress tracking and performance modules enable managers to monitor learnings and identify where extra support is needed or how training can be improved.
Three units have already been deployed at Sellafield Ltd, and two at Dounreay – both sites in the UK – where operators have highlighted its role in reducing risks and improving operator readiness. Calvin Smye, an engineer in the ROV Equipment Programme at Sellafield Ltd, noted that the QFT tool has “proven highly beneficial in maintaining maximum competencies among Remotely Operated Vehicle operators” helping them to “practice and refine their skills without risking equipment or human safety, and remain prepared for high-pressure situations, ultimately boosting efficiency and confidence in real-life”. Recently, two modules were also provided to AWE, which will integrate the QFT into its own robotics training programme, supporting greater use of advanced robotics for decommissioning.
As Varun Kumar, Team Lead and Senior Robotics Engineer at RAICo, says: “Robotics are only as effective as the people operating them. Tools like the QFT give new operators the chance to build confidence and capability in a safe environment before they touch a real machine. That protects valuable assets and accelerates the safe deployment of robotics across the sector.”
