Smart Assistant for Manual Production
How to ensure quality in manual assembly processes? Huber+Suhner and Bossard have teamed up with ETH Zurich / pdz and inspire to industrialise a camera-based solution.
As industrial processes become increasingly automated, manual work is shifting to more demanding and therefore more error-prone applications. Such applications must be supported to prevent errors: As an example, forgetting to apply screw lock or to insert a seal can lead to high follow-up costs. Such or similar cases are difficult or impossible to detect in a final quality control.
Existing approaches such as the four-eyes principle or checklists are inefficient because they interrupt workflows, tie up resources and depend on the conscientiousness of the worker. Smart cameras offer an efficient and reliable alternative: Cameras take over quality assurance steps in manual workflows. This is done by using existing Deep Learning based computer vision algorithms to detect manual assembly activities. This way, assembly errors can be detected and workers can be supported with information about completed, erroneously skipped and upcoming work steps.
In an Innosuisse project, the potential of smart cameras to detect manual assembly steps was evaluated. For this purpose, a prototype smart camera system including a user interface was developed. The smart camera recognizes manual work steps and communicates their completion to workers via the user interface. Furthermore, the technological expertise for setting up new smart camera applications was transferred to the implementation partners within the framework of the project.