Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Defect Annotation on Objects Using a Laser Remote Conrol

 
: Lengenfelder, Christian; Holzbach, Gerrit; Voit, Michael; Beyerer, Jürgen

:

Stephanidis, Constantine (Ed.):
HCI International 2020 - Posters. 22nd International Conference, HCII 2020. Proceedings. Pt.I : Copenhagen, Denmark, July 19-24, 2020; HCII 2020 had to be held virtually
Cham: Springer Nature, 2020 (Communications in computer and information science 1224)
ISBN: 978-3-030-50725-1 (Print)
ISBN: 978-3-030-50726-8 (Online)
pp.535-542
International Conference on Human-Computer Interaction (HCI International) <22, 2020, Online>
English
Conference Paper
Fraunhofer IOSB ()
quality assurance; defect marking; augmented reality and environments

Abstract
In manufacturing, annotating defects for later correction is tedious and still not yet standardized. This task cannot be automated in the foreseeable future since finding and assessing the severity of defects is even for human workers a challenging task. As in the production of car body panels, the manufactured parts are often checked for dents or scratches. Defect slippage is cost-intensive. Therefore, thorough documentation is beneficial. Most manufacturers use direct defect annotation with a grease pencil, coarse masks on a computer or handheld, paper checklists, or simple ‘not OK’ labels on the parts for later inspection and repair. Since no accurate digital documentation of defects is available, defect slippage rates due to workers overlooking annotations are high. Moreover, the required attention shift between the part itself to a representation on a paper or computer screen and vice versa introduces inaccuracies, defect misses and is time consuming. In this contribution, a novel remote control pointing device for defect annotation on objects is proposed. A pilot study was conducted in which the accuracy, user experience and task load were evaluated against a tablet based input method. The results show an average accuracy of 1.5 cm versus 5.1 cm with the novel input method and an overall better and lower task load and user experience.

: http://publica.fraunhofer.de/documents/N-596897.html