Development of hyperspectral imaging (HSI) for nuclear decommissioning

Miniaturised high-resolution hyperspectral cameras offer potential for speedy mapping of materials and environments such as waste disposal repositories, such that real time characterisation and monitoring could be undertaken in real time and remotely. Currently the techniques for achieving this in three dimensions (i.e. such that the material maps can be placed in 3D in context with other materials), have not been developed for nuclear decommissioning and waste disposal environments, so achieving this will provide a new option for real-time remote characterisation in restricted environments, leading to a much quicker decontamination response. Additionally, it is essential to tie characterisation with the safety case that defines what characterisation is required, and thus with knowledge of existing data and data gaps, thus being able to integrate output into a safety case's spatial database management system would allow for easy integration into nuclear safety case development.
The aim is therefore to develop hyperspectral imaging for a wide range of applications in the nuclear sector. One of these is to go on a robotic platform to survey sites for decommissioning and extends the EPSRC funded TORONE project (www.torone-project.com). The student will develop data analytics routines involving artificial intelligence and machine learning to analyse the images.
Hyperspectral imaging is a technique to spatially analyse an area of interest into multiple spectral bands usually in the visible or near-infrared. Many features are only visible in certain wavelength ranges.
Rapid decision making during nuclear decommissioning and radioactive waste management, including deep geological disposal, is key to reducing risk and costs. This project aims to develop a novel capability to undertake speedy, remote, in-situ mapping of materials in extreme environments using hyperspectral imaging in 3D, through (i) integration with 3D digital data (e.g. from 3D laser scanning) and (ii) the use of photogrammetry-style techniques to create 3D hyperspectral material images without the need for 3D geometrical data collection. This will be undertaken within the framework of a GIS-style spatial database building management system (DBMS) to set the 3D context and allow real-time decision making for safety cases.