Developing a digital twin for validation of packing optimiser used for nuclear decommissioning

For sorting and packing solid items of nuclear waste, use of autonomous robotic systems is an accepted future technological trend, as in principle they can overcome all the major constraints of the current practice. This PhD project helps to contribute to this trend by aiming to develop a real-time validation system for packing simulations, comprising a robotic packing setup and a digital twin (DT), such that packing simulation models can be plugged in to be tested, verified and corrected.

The physical twin will be a simple setup with the bare minimums to allow real-time discrepancy detection and correction to be carried out. The emphasis will be on the digital twin which is built on an existing packing software (DigiDEM) to provide a common virtual packing platform to handles digitised irregular shapes, their packing and mechanical stability, and an API to allow different packing algorithms or heuristics to be tested and validated.

Given the overall aim, the specific objectives of the project include: (1) Build a simple physical twin using OTS components and software to allow a robotic arm to pick an object and place it at a specified location/orientation in container. (2) Equip the physical twin with vision systems to allow real-time detection of whereabout of the said object, and of configurational change of the packed bed. (3) Build a skeleton DT to interface with and to match the physical setup, including robotic arm movement and the movement of the picked object. (4) Adapt DigiDEM to establish a common virtual platform to handle digitized irregular shapes and to simulate their packing process, such that accessibility and stability are considered as part of the simulation. (5) Test the system with available packing model(s) to demonstrate DT's capabilities.