Identifying Unknown Objects in Gloveboxes

A challenge launched during the summer of 2018 was to explore technologies which would enable Sellafield to analyse the contents of gloveboxes, specifically allowing them to identify the presence of potentially unknown objects.

Applications were invited for solutions to meet this challenge, the difference with this Game Changers call for innovation being that more mature technologies were required due to a more defined need by Sellafield and the short timescale to anticipated deployment.

The application from i3D Robotics (i3DR) was selected by review panel, with meetings taking place with technical teams shortly afterwards at the Sellafield site and an application for further proof of concept underway. i3DR has been developing object recognition and AI related technologies aligned with its core ‘stereo-camera’ expertise.

The stereo vision algorithms, originally developed for the Mars Rover missions, produce a dense 3D representation of a target scene. These accurate, robust algorithms are processed at fast speeds which makes the technology the preferred choice in a wide range of practical applications such as robotics, navigation and industrial processes.

i3DR has previously produced technology for 3D measurement and automated detection in harsh industrial environments ensuring quick detection of defects in as-cast steel, hot rolled plate and furnace mapping.

i3DR are already engaged with the nuclear industry working in infrastructure mapping, object detection and location, and interfacing, and vision information with robotic systems for autonomous tasks.

The unit proposed for this specific challenge is small and light enough mounted onto a ROV robot arm for insertion into a glovebox and manipulated.

About i3D Robotics

i3D Robotics are experts in 3D vision technology, particularly stereo imaging. The experienced team come from a wide-range of academic and commercial backgrounds including machine vision, space, optics and photonics. As well as providing off-the-shelf 3D imaging products, they design bespoke inspection systems for many applications.

The 3D stereo vision algorithms have been developed by academic partners and use semi-global matching (SGM), an algorithm which is considered as one of the leading dense stereo matchers. Benefits include fast processing speeds, parallelization, robustness and accuracy makes SGM the preferred choice in a wide range of practical applications such as robotics, navigation and industrial processes.

For further information about the i3DR project and opportunities for collaboration/partnership, please email