Understanding the type of risks in industrial automation

Several of us read the The Wall Street Journal article “Why It’s So Hard for a Robot to Straighten a Candle Wick” this last weekend. This story is a good example of “technology vs. process risk” in industrial automation projects. Even though the technology to solve this problem is available today, process complexity prevents prescriptive automated equipment and programming to solve the task at hand.

I usually thought of technology risk as getting all the components of the automated equipment or robot cell to be successfully integrated to deliver the desired performance. Process risk on the other hand is the ability for the chosen technology to repeatedly operate at desired performance when input varies significantly from one unit to the next. As an industry, we have made significant progress in extending the “technology risk frontier” of what can be successfully automated. Platforming of the “design-to-deploy” workflow, productization of key robotic applications, and plug & play automation hardware are all main contributors in the significant reduction of technology risk.

In contrast, the “process risk frontier” has not moved very much over the years. Traditional tools are not well suited to manage large process variations as a human operator does. That said, there are several reasons to be hopeful. Accurate digital twins embedding physics, skill-based robot programming, touch-sensing capabilities for robots, and synthetic training in virtual environments are all emerging technologies that will enable us to improve process risk. Until then, system integrators with deep process expertise have been making miracles at extending this frontier.

As for the candle manufacturer highlighted in the Wall Street Journal, I would like to ask them if they considered using a simple piece of disposable cardboard to hold the wick in the center of the candle during the wax-filling process. No automation is needed and operators can be reallocated to higher-value tasks. Perhaps this is another lesson in itself: to automate smartly, you need to consider the whole process flow, including the manufacturing steps taking place at upstream suppliers. And if you can avoid automating it, this is even better.