Focus on Vision

Assembly processes and services of manufacturers of high-tech systems are characterized in Northwest Europe by a high variety of products and solutions with low volume. Production automation, flexibility and optimization are essential processes for producing smaller series and at the same time realizing the wide variety of products and services.

Research question

How can industrial SMEs use machine learning frameworks within vision applications (object recognition, object tracking and quality checks) to realize more efficient production processes (cobots and autonomously moving systems)?

To enable labor productivity improvements, devices are increasingly equipped with vision systems for pick-and-place applications, quality controls, object localisations and object recognition. However, vision systems are sensitive to changes in the environment, which can cause systems to stop. Vision systems are particularly sensitive to unpredictable changes in the environment, such as lighting, shadow formation, orientation of products and large optical variations in, for example, natural products.

project approach

Machine Learning (ML), a form of artificial intelligence, can largely solve these shortcomings and can make vision systems more robust and faster to configure; ML is extremely suitable for use in vision applications. However, ML for vision is a far-from-my-bed show for many SMBs, predestined for multinationals with large budgets. Above all, the structure and knowledge about applying ML for vision is neither clear nor easily accessible. With this project, the consortium wants to provide insight into this ML structure; second, making ML available to SMEs; thirdly, to jointly investigate how ML for vision can be applied industrially through three case studies and fourthly to safeguard and disseminate the knowledge acquired within SMEs and education.

research outcome

The project is a collaboration between research groups mechatronics and ambient intelligence of Saxion, Computer Vision & Data Science of NHL Stenden. The participating companies are active as high-tech system developers, knowledge suppliers and / or end users as a production company. In addition, the Smart Industry Fieldlab TValley and branch organization BOOST are involved. This project will develop knowledge for proper application of Machine Learning algorithms in vision applications.

Duration project

Start project: 01-09-2019 - End project: 31-08-2021

Partners

logo.png

Saxion

Hogeschool in Enschede, Deventer en Apeldoorn

www.saxion.nl
riwo.png

RIWO Engineering

Website
SINGA_logo.png

Singa

Website
Clear Flight Solutions logo.jpg

Clear Flight Solutions

Website
Focus op vision 1.jpg Focus op vision 2.jpg Focus op vision 3.jpg Focus op vision 4.jpg

More information about the project?