Machine learning. AI. Skynet. Where does the future of automation go?
Author : Robert Campbell, Senior SEO/Content Manager, Wake Industrial LLC
09 April 2019
Shutterstock image
Over the years, machine learning, the idea that machines can teach themselves to operate more efficiently, has seen many advances and many potential paths laid out. Here, Robert Campbell, Senior SEO/Content Manager, Wake Industrial LLC discusses machine learning in relation to Industry 4.0 and what the future might hold for Industry 5.0.
My introduction to machine learning was a modeling system we called Alpha, named for the computer it ran on. Or rather, computers. The Alpha was the modeling system for the hot rolling mill I worked at, where we made sheet steel from slabs (those donuts of steel going down the road, that’s what we made). Originally the mill was manually set up, with operators using experience and guesswork to get the settings close enough to give the feedback system a shot to correct and make a good coil of steel. With the Alpha, we now had something that could use data from previous runs and the resulting outcomes, and teach itself to set the mill properly. This was a semi-automatic system, there was still programming required to make sure the calculations were correct. Operationally it was fully automatic, learning from each run to affect the next. It even advanced to where new products could use similar products for setup. From manual setup to automatic, we saw a huge decrease in waste, increase in quality, and a lot less work to set up an extremely complicated process.
That was 13 years ago. Today, machine learning has advanced to the point where programmers almost aren’t needed once the machine is commissioned. Sensors give information to the processors detailing not just what they detect, but how they’re detecting, what their condition is, and so much more. Computing power has increased to the point now that all of that data can be not just collected, but analysed at extremely fast speeds. This data analysis, and the ability to act automatically on it, creates a situation where the machine can now predict not just what the product is going to do, but how the machine itself is operating. Automated maintenance routines, predictive instead of preventative maintenance, compensating for machine wear and situational changes, all of these come into play with the latest machine learning techniques. And that’s just one aspect of machine learning. With current advances and initiatives, automated modeling is being installed in machines, allowing the machines to take the data available and determine the causal parameters for an outcome automatically, with little or no operator interaction. Where does this put us? Today, we can build, program, and commission a machine and then merely observe as the machine optimises itself and begins improving efficiency and productivity. If you’re into automation and the future it holds, these are some exciting times.
What does this mean for the future? Just look to what the big automation companies are doing. Rockwell invested $1 Billion in PTC, a company that works in the AI and machine learning space. Why did they do that? Industry 4.0 is becoming the data revolution. The timely, proper use of data will determine who the industrial giants of the future are. Big Data has been around for quite a while but data is just that, random pieces of information that mean little by themselves. Inside Industry 4.0, you’re seeing a data driven optimisation of all processes, from physical to digital, sales to manufacturing. In the near future, operators will be guided by the machines to adjust and maintain the machine per the machine’s requirements. Augmented reality will be used to allow machines to talk to their operators and maintenance personnel. “Digital twin” plants will allow machines to simulate changes to themselves to ensure any changes are beneficial to the actual process. Pretty soon, the machines will be telling the humans what they need to do, instead of the humans telling the machines.
What does this mean for Industry 5.0? I have no idea. The idea that computing power would be at this level, allowing machines to teach themselves was unthinkable just a decade ago. What will the next revolution be? My guess is more “I Robot” than “Terminator”, but whatever it is will be unimaginable to 2019 humanity. Maybe cyborgs. Cyborgs would be cool.
(Photo Credit – Business Review)
As for the dangers of the future, a story from Facebook is telling. Two AI systems were set up to communicate with each other. Quickly they taught themselves a language no one could understand, and the language was changing so rapidly that no one could catch up. No one knew what the AI’s were saying to each other. Were they planning our demise? Calculating the next Powerball numbers? Who knows, but one scientist decided that it was time to pull the plug. At some point the machines will be able to stop us from pulling the plug, and that could prove…interesting.
Robert has a degree in IT Systems Development and has been working with and around Industrial Automation equipment for over 10 years. He joined the Wake Industrial Company in January of 2016.
Rcampbell.wakeindustrial@gmail.com