I first read about ‘smart advisors’ in a report published by industry analyst Gartner. The idea aligns well with our own thoughts that decision making will shift away from people to virtual assistants. Already prevalent in the banking and services industry, virtual assistants advise us on a wide variety of choices such as what loan to take out or which local restaurant to choose.
When we think about smart advisors in manufacturing we can imagine them recommending maintenance work, changes in process parameters or schedule changes. With a smart advisor we do not fully trust the virtual assistant to make a decision. We believe that it’s not yet intelligent enough to fully replicate human decision making. Why is this?
Actually our mistrust is well founded. Anyone who has worked in a plant knows that the theory and practice of data collection are not the same. If we want to automate decision making then we must be sure that data is accurate. Any of the thousands of sensors could be giving us inaccurate data, a communication channel might fail or communication might be delayed. There are a myriad of opportunities for failure.
Even so, with smart manufacturing we should be able to overcome these problems. Systems are becoming increasingly intelligent, especially when programmers take account of potential points of failure. Smart manufacturing needs smart programmers, the so called super engineers who truly understand virtual and physical manufacturing. In Industry 4.0, people provide the intelligence to autonomous systems, those people have to be really smart. The smart advisor is programmed by a super engineer, it will be iterative and be constantly improved with experience and knowledge.
I think smart advisors are a perfect starting point for the journey into smart manufacturing and Industry 4.0. Eventually we expect to see the plant fully autonomous and self-organising. Smart advisors can be programmed for specific tasks, once we are confident they can make good decisions all or most of the time then we could approve them for autonomous operations. Many autonomous agents, working together, will become the building bricks of autonomous manufacturing.
About the Author