Making nuclear decommissioning safer with AI
Veolia
We integrated AI into robotics to autonomously segregate and sort nuclear waste, to speed up the process and reduce the need for humans to get close to sites.
Background
Robotics enable nuclear sites to be decommissioned at the end of their operational life at a distance, without the need for hazmat suits. However, even with robotics involved, this process remains slow and expensive. Veolia has a robot – Dexter – which is operated remotely, tailored for use in nuclear environments. Veolia sought to explore how AI could enable its system to operate autonomously, boosting efficiency and minimising the need for human intervention, ultimately enhancing safety.
“This project shows the power of AI in robotics, even in complex environments where safety is critical.”
Iain Mackay, Customer Director
Faculty
Solution
To sort waste and segregate autonomously, Dexter needed the overall process broken down into steps. Using simulated radioactivity, Dexter now has the key information to confidently determine the object's type and material, ensuring the correct packaging drum is selected for proper disposal. The team wanted a failsafe so if the robot can’t decide, a human steps in to finish the task remotely.
Impact
NDA and Sellafield experts were impressed that the escalation element meant the robot was never stuck thanks to the integrated human-machine team. We are working to develop a real-world commercial solution based on this demonstration.
“We’re delighted by the engagement and results provided by Faculty. The team was agile and responsive but also insightful in identifying where AI could make an impact and where traditional methods were more effective. We're excited to explore new opportunities together.”