Robotics and AI systems for dust detection and decontamination on Tokamak Systems.

The aim of this project is to autonomously detect the presence of Beryllium Oxide (BeO) dust and decontaminate components removed from Tokamak reactor. This will be achieved through the use of a robotic manipulator fitted with sensors for detecting beryllium and a vision system for object detection and localisation. The system will use a chemical analysis technique, most likely Raman Spectroscopy, to detect BeO in real-time and influence the path planning of the scanning process based on data.

To achieve this aim, the development of novel algorithms will be considered in the topics of path planning, object detection and localisation to address the specific problem of detecting Beryllium Oxide dust, where sensor must be placed within a specific distance and orientation to the object by an over-actuated robotic manipulator. The approach to address the problem is given as follows:

Develop a gimbal system for the end effector of the robotic manipulator, adding two additional degrees of freedom (DOF) to a standard 6-DOF robotic manipulator. This gimbal system will have smaller, high precision motors allowing for greater accuracy and speed during the scanning process.

Develop novel path planning algorithm and inverse kinematics for the over-actuated manipulator, which will prioritise the use of the additional joints for speed and precision.

Develop a scanning algorithm that efficiently covers a components entire surface, maintaining a pre-determined standoff distance and remains at each location for a certain integration time as specified by the Beryllium sensor.

Use a vision system to recognise a component from a library of potential CAD files, for the chosen component perform 3D pose orientation detection and localisation of the component in the manipulators available space. This is necessary so that a precise standoff distance and angle can be maintained for the Beryllium sensor