ARRANGE – classification, pose determination and tracking of industrial objects to support sorting work

Name of the participant: Fabian Rücker

Description of the IT research project: Although sorting tasks do not appear very complicated at first glance, in industry – despite progressive digitization – they still often have to be carried out by hand, because, unlike for humans, it is not so easy for a computer to distinguish geometric shapes from one another and to order the chaos according to certain specifications, which requires an understanding at the machine level that provides camera shots with semantics and then classifies and detects objects correctly. These are central problems in the field of computer vision.

The aim of this research project is to record a sorting process in sheet metal production using computer vision and deep learning methods and to support the operator with the aid of augmented reality glasses: The AR-glasses are to identify the individual components and display relevant information about the sorting process in the field of vision by means of the built-in camera. In addition, in order to display digital information next to a classified object, the pose of the object in the camera image must be estimated.

Within the framework of ARRANGE the described computer vision procedures are developed, combined and executed on AR-glasses.

Software Campus partner: Fraunhofer ICT Group, TRUMPF

Implementation period: 02/2020 – 01/2022