Description
Challenge
At Ophardt, currently two operators work at the end of the anodizing line, where aluminium shrouds are anodised. One operator racks down the aluminium shrouds and another one checks the quality of the shrouds. De-racking the shrouds affects ergonomics, as it entails grasping the objects above head height and below hip height. Human quality checking is error prone, and often not consistent, as there is always a gap in the perception of inspection between different operators.
Solution
To address these two problems, the RANCH experiment (a) uses a collaborative robot to aid the operator in de-racking the aluminium shrouds and (b) uses a vision system together with deep learning to automate the quality inspection.
Automating de-racking is non-trivial as the titanium strip, where the shrouds adhere, is not fully static and can move a bit while racking down. Also, shrouds might be clamped extra on the strip during the anodizing process requiring variable forces during de-racking. Visual inspection is hard as even now there are differences in the judgements of different operators, especially for detecting extrusion lines. An automated quality inspection helps to avoid discussions between the operators and increase the consistency of judgements.
The HORSE framework is used to design and implement the process workflow, and to enable the communication between all actors involved.
Project:
Enterprises:
Ophardt Belgien (End user); Sirris & IMEC (RTOs) ‘
Sector
Health
Keywords
