Theory and practice must work together. We need to bridge the gap that exists between idealised abstractions and messy reality. This means Felix Viberg, who conducts research on automation and machine learning as an externally employed doctoral student at SKF’s factories in Gothenburg.
Many people who have not spent a lot of time in factories become nervous by the idea of a factory floor. This has been noticed by Felix Viberg, industrial doctoral student in computer and information science at Linnaeus University. Therefore, it is challenging to place academics in such environments, something at which we must become better.
“It is necessary to be able to abstract problems based on the messy reality if one whishes to come close to the spectacular things that everyone is talking about but no one manages to achieve. Like, for instance, lights-out manufacturing”, says Viberg.
“The methods are there, but the applied science is missing. I think that one explanation is that factories may be a bit too imposing for their own good, even though this may sound banal”, Viberg continues.
Combines automation and machine learning
• Lights-out manufacturing: A factory that can be operated “with the lights switched off” – it is fully automated and does not require any on-location staff.
• Machine learning: To teach a computer system to make as precise predictions as possible. You teach an algorithm to recognise specific answers based on previous data and to then answer the same questions for future unknown data.
• PLC: Programmable logic controller – a type of computer that is used to control processes, mainly in the industry.
Viberg is an externally employed doctoral student at SKF’s factories in Gothenburg, where he works with machine learning and data-driven software systems. He is a trained engineer with specialisation in technical physics and wants to combine his knowledge about automation and machine learning.
“There is incredibly much to do! Many large industrial companies have started building their competence within this intersection, but it seems hard to get the ideas out there”, says Viberg.
At SKF, Viberg works daily with specific, practical problems in this very intersection between automation and machine learning, combined with his doctoral studies. He is expecting the path towards a doctoral hat to be challenging, but he feels motivated.
Dreams of his own factory
Felix's first real job was as an automation technician at Småland sawmills. Although he has studied for many years since then, he has never completely left the factories. His knowledge of programming PLCs has given him several fun summer jobs. Felix is simply drawn to factories and feels at home in that environment.
“I’m not a doctoral student in order to be awarded with an academic title. It’s more like I’ve set a trap for myself, to make me fall down into the well of knowledge. Once I exit the well, I want to know how to build factories that can run themselves”, Viberg adds.
“Then it might be possible for me to have my own factory. It doesn’t really matter what it makes. Perhaps just a small factory in my basement, that makes paper clips or something. The most important thing is that it operates independently and never stops!”, Viberg concludes.
Concrete solutions at an early stage
Daniel Einehag is manager for Manufacturing Reliability Engineering at SKF Sweden and Felix’s supervisor. He is happy to have an industrial doctoral student at his department and sees great possibilities for synergy effects, for both the doctoral student and the manufacturing.
“For the first time at SKF’s factory in Gothenburg, we have an industrial doctoral student working at a department that is closely linked to manufacturing. This makes it possible for idea and research to lead to concrete solutions that work and create value already at an early stage”, says Einehag.
“Through clear projects linked to the needs of manufacturing, I’m convinced that Viberg’s research will be unique and add value. It would be fantastic if the results could come to use straight away, instead of risking ending up a desk drawer, which theoretical research results that are hard to apply tend to do way too often”, Einehag concludes.
- Felix's project, Machine Learning in Manufacturing