Blended Intelligence for Safe and Efficient Nuclear Sort & Segregation

The Challenge

The Sort and Segregate Phase 2 competition was a funded initiative, which aimed to find, through technology development, a solution to the problem of sorting and segregating radioactive waste generated by nuclear decommissioning activities. The developed solution would be an autonomous, fully integrated system capable of dealing with nuclear waste in a safe and efficient way.

The Solution

The BI-SENSS (Blended Intelligence for Safe and Efficient Nuclear Sort and Segregation) approach adopted was differentiated by the layering of automation on a haptic telemanipulator rather than onto an industrial pick and place robot. In this way, the system is designed to allow skilled operator intervention should automation fail, without any human exposure to the hazardous environment. This mitigates a key operational risk - having to deal with outliers in legacy wastes, and also allows the system to easily adapt to variations in feedstock, applying the right skill and intelligence to the right task.

The project focused on the development of an inactive demonstrator tasked with integrating the key systems required for automated waste sort and segregation (object identification, grasping, manipulation, physico-chemical and radiological characterisation, consignment decision making, packing optimisation planning, packing).

VNS was the lead partner in the project and responsible for providing the Dexter™ telemanipulator and all the hardware required to successfully deliver this non-active autonomous prototype. As an expert in nuclear robotics, VNS was also able to bring to the project its specialist knowledge of haptic -telemanipulation through Dexter™.

The project team, after 15 months of dedicated work, was able to achieve its main goal of delivering a fully integrated system able to autonomously carry out the following steps:

• Take scans of waste items in a pile
• Segment the items (identify separate items)
• Create grasp poses and execute a grasp pose by picking up a selected item (see figure below)
• Carry out further characterisation of the picked up item (material type, mass, radiation level)
• Decide based on the characterisation results where the item should be packed
• Pack the item efficiently in the relevant container, in a way that minimises voidage space
• Return to the starting position and start step 1 again

The system is also able to request the operator's input at any stage of the process when dealing with complex items.

The full strength of this integrated autonomous demonstrator was displayed at a live demonstration to all the stakeholders when the system was able to successfully sort, segregate and pack a wide variety of waste items that it had never seen before. The core functionalities of the system were highlighted as well as the potential to transfer these into a fully operational active demonstrator.

The next step is to build a fully integrated active demonstrator and validate its ability to autonomously sort and segregate waste in a nuclear environment. The system is scalable for application in the nuclear industry and beyond.

SPECIALIST SUB-CONTRACTORS:

• Faculty (Artificial Intelligence)
• Createc (radiological detection equipment)
• University of Lincoln (robotics automation)
• Mott McDonald (waste management consultant)

VNS EXPERTISE:

• Robotics and Remote Handling Solutions
• Radioactive Waste Sorting and Repacking