What do computer chips, TVs, tomato soup cans, disposable diapers, root beer bottles, and chocolate chip cookies all have in common? They are all tracked, counted, or measured by Machine Vision, a software technology that performs inspections on thousands of products, from the electronics store to the supermarket.
Mark Keup is one of the software engineers responsible for this task. He has worked at Cognex for twenty years. He and his wife Sandy live in New Berlin, Wisconsin and have three children. Two of them participated in FIRST Lego League (FLL) and FIRST Tech Challenge (FTC) when my son Jeffrey was involved. Mark coaches robotics at Heritage Christian School alongside engineer Dan Dotson.
How did you know you wanted to be an engineer?
Mark: Growing up, I took a few courses in high school in drafting and electronics repair. This led to me getting a job as a technician in a small electronics firm. I wasn’t a motivated student in high school. I was fascinated with the emerging new technology of video games and liked to play sports. Study? Not so much.
Meantime, my best friend’s dad was an electrical engineer at Rockwell Automation (at the time Allen-Bradley) and I really looked up to him as an inspiration. I wanted to be an engineer, too. I applied to study at the University of Wisconsin – Platteville to a brand-new program in Electrical Engineering.
I didn’t start as much of a student there, either, but a professor I really looked up to inspired me to get my act together. I eventually graduated and was doing very well in most classes by the end. I suppose it took a while to mature. Along the way, I realized how much I liked the digital world and particularly software. I also discovered I was pretty good at it.
Where have you worked as an engineer?
Mark: On my first job, at Paragon Electric, I had my first chance to build a product. It was a piece of electronics that was used in supermarkets to control the defrost cycles of all the refrigerators and freezers. I learned a lot about engineering there. I also grew my enjoyment of software development.
I got married, moved to New Berlin, and took a new job as a software developer, designing software to analyze car engines for repair shops. From there I went to work for Rockwell Automation (where my inspiration – and now father-in-law had worked) and began my career in Industrial Automation.
This is the field I’ve worked in ever since,
making products that go onto machines that make things.
My specialty was the Bar Code reading industry (with lasers back then)
and after a few years, a new field called Machine Vision.
Eventually Allen-Bradley decided to sell my group of to a company named Cognex. Cognex was already a leader in this industry and has grown to be the global leader in this rapidly growing industry.
What exactly is Machine Vision?
Mark: Machine Vision uses high speed cameras to take pictures of things being manufactured and then uses advanced mathematical algorithms to analyze those images to perform inspection on the product. Sometimes it’s as simple as determining whether an object is there or not. Sometimes it means making measurements. Sometimes it means finding a pattern and reporting on the location of the pattern so a robot can move to pick it up.
Over the years, Machine Vision has evolved to inspect
just about every product made.
Everything at your supermarket.
Everything at your electronics store and more
are inspected by Machine Vision.
We like to say we inspect everything from computer chips
to chocolate chip cookies.
What else is involved within the field of Machine Vision?
The field is wildly creative because most
of what we try to do hasn’t been done before.
As products get smaller or the the need to
manufacture them faster increases,
we must find a way to adapt new technologies to our field
to keep up with the needs of manufacturers.
Sometimes that means inspecting OLED TV panels with 10 micron accuracy or doing dozens of inspections on disposable diapers, beverage bottles, or canned goods moving at 40 or 50 miles per hour down the assembly line. Sometimes that means finding a way to miniaturize our cameras so they fit into the tiny wrist cavity of a collaborative robot that works side by side with a worker in a factory, doing the jobs that a human wasn’t designed to do.
Where do you get ideas for new products?
Mark: Most of the ideas for new products come from analyzing problems our customers face. Often, more than one company faces a similar pattern of issues and we can turn that into a product that can both help them and be a worthwhile investment for us to do.
Sometimes that doesn’t work out and we just have to say no, either because there is too much work to make a profit or because the technology isn’t ready to tackle that problem yet.
Because of the long time I’ve spent doing engineering in this field,
I’ve gotten to see many advances in what we do.
Things we used to say were impossible are now very possible.
Over time, the speed of our cameras has increased
by over 100 times and the size has shrunk from needing a PC
and a large boxy camera to now
putting ten times that power into a 1-inch cube!
What is the process of product development?
Mark: We’ve found that innovation in product development proceeds in stages, kind of like a funnel. In the earliest stage there are many ideas and many solutions proposed. It takes research on each of them to determine which of them has value that might be worth investigating. In any given year we might have hundreds of these ideas, coming from customer problem statements or from engineers internally.
In product engineering, we evaluate the pros and cons
of each opportunity and select the best of them
to evaluate more deeply.
We usually do all the research and innovation
and create a prototype that shows how our ideas
can be made into a product.
At this point we try to construct a plan for how to turn the prototype into a product. We talk to customers and internal people to refine the concepts. From there we can make it into a real product. Sometimes this means making an end-to-end schedule, planning the product in phases, and then doing what we can to make that plan be real. That’s often called ‘Waterfall’ planning.
One of the newer ways people are thinking about development planning is something called “Agile.”
In Agile, instead of trying to plan everything out
from the beginning and running into unexpected surprises
along the way that ruin the plan,
you try to prioritize the work and complete
a slice of the idea all the way through.
By building that end-to-end slice, you learn things that will inform you about how to do the next slice. If discoveries are made along the way, you can adapt the plan.
It is truly fun to try out the new ideas from work on the robotics world, and the reverse, too!
What kind of process do you follow for whatever you’re involved in?
For completing projects, do you work better alone or in a group?
I’d love to hear from you!
P.S. Join me next time for Mark’s and Dan’s role in developing the FLL/FTC robotics program for students . . .