Mobile Autonomous Sort and Segregate System

A robot mounted on a mobile modular system that can be deployed to identify, sort and segregate radioactive waste for safe recycling or disposal.

The robot will confront a mass of radioactive waste ranging from metals, to plastics, electrical equipment, soil and more. In order to process it, the robot will first identify an individual waste item using its vision system. The robot will recognise some waste, but otherwise it can be trained by an operative to identify a new waste type by sight. Through machine learning, the more the vision system is used, the more autonomous the process becomes. The system will also measure each item's weight, size, shape, surface area and composition for efficient sorting and packing.

Visual recognition of waste is combined with radiometric and chemical characterisation to classify the waste for sorting. After visual identification, each item is picked up by the robotic arm and its level of radioactivity is monitored and it is chemically analysed. The information on the item's physical characteristics, material type and radioactivity level is used to sort the item into the correct waste-stream for safe recycling or storage.

Records of this information, together with images of the items will be stored as a record of what items have been placed into each waste-stream.

This project innovates on current state-of-the-art by removing the person from the process. This means that there is less risk to the operators from working in hazardous environments. There is less risk of human error in this repetitive task. The process will also be quicker and cheaper than a manual system, offering savings to the UK taxpayer on the cost of decommissioning redundant nuclear equipment and facilities.

Combining a robot arm with vision systems, machine learning, nuclear and chemical characterisation systems will mark a new development for nuclear decommissioning. The robot arm can be of any model and size to suit the waste type. A further key innovation comes via the intelligent vision system, which automates the recognition of different forms of waste through machine learning. Human interaction is minimised, creating an efficient, waste-minimising workflow that can adapt to location, segregate waste by various measurable criteria, and will improve the more it is deployed. Waste generated by nuclear decommissioning is therefore dealt with safely, quickly and cheaply, with minimal human interaction, efficiently packing waste containers, and with a diligent recycling process.