Probably NOT artificial intelligence. It’s already a thing, but not THE big thing. AI is just one tool among many, and suitable for some tasks but not others.
There are serious problems using AI in production environments for some tasks, which is one reason machine learning—which has been around for decades—hasn’t been as widespread as one might expect.
Imagine you’re building some product that, when it rolls off a production line, sells for $10k to $100k, and you make 1000 or 5000 units a day. Serious care is taken not to damage one of those items during production.
Assembly of the product will be dependent on industrial robots for reasons of cost, safety, and quality. Robots report the same motions over and over, and for many operations typically need expensive jigs or machine vision to endure the location and orientation of the product is known precisely so that the robot can move in and so it’s job without colliding or missing.
if you use AI-based guidance, you might at first be happy that in the lab it correctly identifies an object’s position and orientation 99% of the time, and to within a millimeter. Wow, right?
But 99% in a production environment is miserably bad. If a robot doesn’t finish its job, or crashes 1 - 10 times per day, the AI vision vendor will get tossed out with lots of screaming, and possibly suffer legal consequences.
Also, when the AI fails, it may be brutally difficult to determine why. The mass of trained parameters are difficult to interpret.
There’s a lot of money spent on applications like this, and AI-based solutions aren’t well suited to it. AI can do a pretty good job of detecting quality issues that don’t cause catastrophic failure. AI can identify and possibly prevent equipment failure, because even 80% or better accuracy in preventing a CNC or robot or conveyor motor failure can reduce downtime.
It’s also not immediately clear how AR/VR will benefit production. A lot of ideas are being tried, and many look good, but cost/benefit isn’t an easy matter to figure out. A lot of money is being spent, and some applications will work out, but others won’t.
”Revolutions” in industrial automation take years of stressful work and long stretches sitting inside assembly plants and factories. Vendors that arrive on site “knowing” their technology is smart and inevitable nearly always fail miserably.
Can't agree more with you, I would never trust a black box blindly! It still could be used though if there is verification afterwards, or if its used as an aid to a human.
The AR/VR trend is interesting, because this technology is aimed at humans, and automation is exactly aimed at replacing them. Instead of implementing AR/VR so that a human can use it, why not just automate the process and skip the human component altogether?
1 of 1 people found this helpful
Industrial automation has been evolving steadily since the start of the industrial revolution. There are so many industries that a giant leap forward in any industry would not be considered the next big thing in industrial automation.
I think there will be a noticeable trend to developing industrial automation that travels to the location it is needed rather than just making bigger factories (which will also happen).
For example automated machines that build buildings in their final locations. Such machines are so large they will need to assemble themselves on-site. They will also need coordinated robot armies to handle a myriad of detailed tasks.
The tech is already here, but the existing wired infrastructure is allowing it to grow very slowly typically only in remote locations.
Are you talking about wireless communication?
I'm basing this on something that I've not come across before, which is distributed or wide area automation. With 5g you can essentially run or coordinate production of parts in multi geographic regions which will then seemlessly and automatically (thanks to driverless transport) come together just when needed.
99.999% of those reading this discussion would probably answer my title's question with: Artificial Intelligence. I would not necessarily disagree.
AI will surely impact industrial automation. But I was thinking a bit differently. Something along the lines of Augmented or Virtual Reality.
While VR/AR technology will improve automation, it can also aid in the manufacturing assembly process for sophisticated products, as well as improve maintenance, QA, and tech support. The possibilities are amazing.
What do you think?