Using deep learning techniques for automated inspections

At Prodrive Technologies I work as a software architect on the manufacturing automation of the factory, particularly the AI and vision solutions that provide the intelligence. What I find a big advantage of working here is that I have a big say in defining on what I want to work on. Currently, I do not only work on the implementation of software and algorithms but also have a more leading role in making architectural design choices and reviewing the work of other engineers.

When I start my day, I typically check if there is any review work for me to do. Reviewing each other’s work is very important so we can use the lessons learned, do not invent the wheel again and it also results in having everything documented as a bonus. I get a kick out of helping colleagues with their software, calibration, qualification and vision challenges and as a result often get quite a few calls every day. I will not get bored easily!

I can also spend some time on my own work which currently involves improving camera calibration algorithms. It uses ChArUco patterns that, without going into too many technical details, result in having a better distribution of calibration points and in turn gives a more accurate camera calibration. It is quite important to get this right as all future camera setups used for inspections and localization of components depend on the accuracy of this calibration.

Finally, as an architect I am working on a roadmap to determine the future of deep learning and machine vision at Prodrive Technologies. Visual inspection processes that are currently being automated are component presence detection, cable insertion and anomaly detection for product damage. I setup requirements for a generic machine learning framework and its applications so we can leverage machine learning technologies in multiple processes. This is a big motivator for me as I can really contribute to how the company applies vision and machine learning techniques that will play a large role the fully autonomous lights-out factory.

I get a kick out of helping colleagues with their software, calibration, qualification and vision challenges

System Architect profile

• BSc, BEng, or MSc in Mechanical, Electrical, Mechatronic, Industrial Engineering, or Physics.

• Strong understanding of mechanical/mechatronics systems and technologies.

• Excellent leadership, interpersonal, and problem-solving skills.

• Analytic mindset, able to quickly assess issues and their root causes.

• Highly motivated, feel responsible and able to work effectively under minimal supervision.

• Strategic and creative thinking.

• Strong organizational and time-management abilities.

• Ability to thrive in a dynamic and fast-paced environment.

• Being decisive, a relationship-builder, and pragmatic.

Pros and cons

• Very versatile work from requirements, to implementation and qualification

• You are allowed to take responsibility if you think it can be done better

• Flexible working hours and environment

• I get a lot of calls to help colleagues

• I get a lot of calls