Högskolan i Skövde

his.sePublications
EnglishSvenskaNorsk
Change search
ExportLink to record
Permanent link

Direct link
BETA

Project

Project type/Form of grant
Project grant
Title [sv]
Automated quality inspection in assembly lines through low-cost vision system (VISION)
Title [en]
Automated quality inspection in assembly lines through low-cost vision system (VISION)
Abstract [sv]
Syfte och mål:Syftet med projektet är att utveckla ett billigt och högpresterande vision-system för automatiserade kvalitetsinspektioner i monteringslinor baserat på billig hårdvara och avancerade maskininlärningsalgoritmer för bildbehandling i realtid. Motivationen bakom projektet är att möjliggöra en utbredd användning av automatiserade kvalitetsinspektioner inom svensk produktionsindustri för att minska, eller till och med eliminera, kvalitetsfel i manuella monteringsprocesser.Förväntade effekter och resultat:Projektet förväntas minska, eller till och med eliminera, kvalitetsfel i manuella monteringsprocesser. Förutom ökad kvalitet är en drivande faktor bakom projektet att sänka produktionskostnaderna genom att undvika hantering av kassationer och att kunna ersätta dagens bemannade kvalitetskontrollstationer med automatiska inspektioner längs produktionslinorna, vilket också möjliggör kortare ledtider.Upplägg och genomförande:I projektet kommer en fullt fungerande lösning på TRL-nivå 6 att utvecklas och utvärderas genom ett antal industriella testfall. Lösningen kommer också att implementeras i en offentligt tillgänglig demonstrator. För att säkerställa att alla företag, oavsett storlek, enkelt kan installera och använda med lösningen kommer vision-systemet att utformas för att fungera med billiga webbkameror som kostar mindre än 300 kr. Systemet kommer att publiceras som öppen källkod för alla att kunna laddas ner och användas utan kostnad.
Abstract [en]
Purpose and goal:The aim of the project is to develop a low-cost, high-performing vision system for automated quality inspections in assembly lines based on cheap hardware and state-of-the art machine learning algorithms for real-time image processing. The motivation behind the project is to enable a massive use of automated quality inspections within the Swedish production industry in order to reduce, or even eliminate, quality defects in manual assembly processes.Expected results and effects:The project is expected to reduce, or even eliminate, quality defects in manual assembly processes. Besides increased quality, a driving factor behind the project is to lower production costs through avoid handling rejections and being able to replace today’s manned quality inspection stations with automated inspections along the production line, which does also enable shorter lead times.Approach and implementation:In the project, a fully functional solution on TRL 6 will developed and evaluated through a number of industrial test cases. It will also be implemented in a publically available demonstrator. To ensure that any company, regardless of size, can easily install and afford the solution the vision system will be designed for cheap off-the-self cameras costing less than 300 SEK. The system will be published as open source software available for anyone to download and use without cost.
Publications (1 of 1) Show all publications
Syberfeldt, A. & Vuoloterä, F. (2020). Image Processing based on Deep Neural Networks for Detecting Quality Problems in Paper Bag Production. Paper presented at 53rd CIRP Conference on Manufacturing Systems, July 1-3, 2020. Procedia CIRP, 93, 1224-1229
Open this publication in new window or tab >>Image Processing based on Deep Neural Networks for Detecting Quality Problems in Paper Bag Production
2020 (English)In: Procedia CIRP, E-ISSN 2212-8271, Vol. 93, p. 1224-1229Article in journal (Refereed) Published
Abstract [en]

To identify quality issues within the production and prevent defect products to be delivered to customers is critical for most manufacturing companies, and usually performed both within and at the end of each production section. In this paper we investigate the use of deep neural networks for performing automatic quality inspections based on image processing, with the aim of eliminating today’s manual inspection processes. A deep neural network is implemented on a real-world industrial case study and its performance is evaluated and analyzed when it comes to detecting quality problems in produced products. The results show that the network has an accuracy of 94.5% which is considered good in comparison to the 70-80% accuracy that a trained human inspector can achieve.
Place, publisher, year, edition, pages
Elsevier, 2020
Keywords
Deep Neural Networks, Image Processing, Quality Inspection, Industrial Vision Systems
National Category
Production Engineering, Human Work Science and Ergonomics
Research subject
INF201 Virtual Production Development; Production and Automation Engineering
Identifiers
urn:nbn:se:his:diva-19092 (URN)10.1016/j.procir.2020.04.158 (DOI)2-s2.0-85092428222 (Scopus ID)
Conference
53rd CIRP Conference on Manufacturing Systems, July 1-3, 2020
Funder
Vinnova
Note

CC BY-NC-ND 4.0

Edited by Robert X. Gao, Kornel Ehmann

The authors would like to thank Jonsac AB for their support in the study and for allowing us to work in their facility. The authors also want to thank Vinnova for financing the VISION project through the strategic innovation program Produktion2030, within which this work has been undertaken.

Available from: 2020-09-23 Created: 2020-09-23 Last updated: 2025-09-29Bibliographically approved
Principal InvestigatorSyberfeldt, Anna
Coordinating organisation
University of Skövde
Funder
Period
2018-04-16 － 2020-11-30
National Category
Other Engineering and Technologies not elsewhere specified
Identifiers
DiVA, id: project:2279Project, id: 2018-01592_Vinnova

Search in DiVA

Other Engineering and Technologies not elsewhere specified

Search outside of DiVA

GoogleGoogle Scholar
v. 2.47.0
|
WCAG
|
Skövde University Library
|
Contact