Go back

Deep learning during the
ctrlX developR Challenge

The developers taking part in the ctrlX developR Challenge are now well into the phase where they put their ideas into practice. The aim is to develop new automation applications based on the compact ctrlX CORE control platform. Prof. Dr. Andreas Schiffler (44) from Germany and Cesare Bornaghi (38) from Italy are facing a particular challenge: “Deep learning”. In this interview, the two developers talk about what makes their projects so unique, what they personally have learned from them and more.

Cesare, you have a background in automation and control technology and you’re the CTO of Vision System. What fascinates you about this profession?

I work for a company that develops innovative computer vision applications. It’s my dream job because I’m able to help shape future technologies and have to solve numerous challenges in a range of applications. It’s a challenge every day. That’s what drives me.

Why are you taking part in the ctrlX developR Challenge?

I’ve been working with ctrlX AUTOMATION from Bosch Rexroth for a few months now. It’s fun to develop and integrate solutions on this automation platform. I can now demonstrate my ideas and experience with it in an international competition and go up against other developers from all over the world.

What is your project for the competition?

My idea is to present the results of computer vision algorithms based on deep learning on ctrlX AUTOMATION. I want to create a system which is able to find production errors and catalog images of them in order to produce a database. As a result, it will be possible to classify and analyze error types. On the basis of the findings, it could then be possible to make improvements in production.

Andreas, your idea too focuses on deep or machine learning. What exactly is it about?

I’m looking at the orchestration of OSS in the context of computer vision and machine learning algorithms in order to come up with advanced (and integrated) controller functionalities. You could call it a “soft sensor”. The measured data are the input and the output, from a user perspective, is nothing more than a new value or scale on the data level. One aim is to recognize anomalous process statuses by simultaneously evaluating various sensor measurements.

How is the solution phase going?

It really is a challenge. You have to take into account all the necessary dependencies and implement them in the given time. It isn’t trivial. But I’m receiving some very good support from my mentors. And the atmosphere within the team is so good that I can envisage working on other projects with Bosch Rexroth in the future.

What could your project result in?

At the end of the project, there’ll be a demo or a prototype which can be used in my laboratories by “Applied Machine Learning” students. After all, I’m a mechanical engineering professor at a university. Naturally, I’d be delighted if this solution was used as part of the ctrlX AUTOMATION ecosystem later on and was able to help companies with industrial automation. I look forward to seeing what will develop as a result.

If you have any questions or require further information regarding ctrlX AUTOMATION, please contact us:

Get in contact

ctrlX AUTOMATION the magazine!

The title "We're open" sums it up. Find out how ctrlX AUTOMATION is developing from an open platform to an industrial ecosystem. You can expect exciting user examples, partner statements and all the latest news about ctrlX AUTOMATION.

Get your free ctrlX AUTOMATION magazine!

Go back