The National Science Foundation recently granted members of a research team from the University of Michigan, University of Illinois and Cornell University $4 million for a project on increasing productivity in factories that use robots.
The three universities will be working to boost productivity by teaching robots how to detect external threats that might cause them to stop working, as well as recover from their own mistakes and start new tasks without delays or interruptions.
Dmitry Berenson, professor of electrical engineering and computer science, said robots have been an instrumental part of productivity improvements over the last few decades, though the nature of their function has changed.
“Robots have generally been there to do the Dirty, Dumb, and Dangerous, or the three Ds as we call them; those are tasks that humans don’t want to do, or are too dangerous for humans to do.” he said. “But overall, the major improvements in productivity over the last 30 or 40 years involve things like automatic structuring and a big part of that has been using robots, so they’ve been very beneficial.”
In recent years, U.S. productivity gains have slowed, which worries economists as growth in productivity generally leads to higher standards of living. According to the Bureau of Labor Statistics, U.S. productivity decreased by 0.6 percent in the second quarter of 2016.
To combat this, researchers on this project are trying to find a way to equip robots with more problem-solving capabilities.
The goal of this research is to create a new system called “software-defined control.” This will primarily be used to reduce robotic failures in factories, as they can be costly and time-consuming to fix, by creating real-time monitoring and response capabilities.
Mechanic Engineering Prof. Kira Barton, a researcher working on the project, said the simulations the team will conduct will be beneficial to improving the way manufacturing works.
“The aim behind these types of simulations would be that you’d have a good model of your system that you can combine with data that you’re collecting from the system itself,” she said. “You’d be using this so that you can detect when there are anomalies in the system, or when something adverse occurs you can try to identify that it happened, isolate that specific area of the manufacturing system and make an informed decision about how to keep production going.”
The system the researchers are working on essentially compares an actual physical manufacturing plant with robots working in it to a simulated version and strives to find any discrepancies between the two. The way the parts flow through the factory can then be reprogrammed, avoiding faulty equipment.
Engineering junior Jesse Jenter said he thinks the increase in industrial productivity with robots could decrease the need for unskilled labor and require more people to get a degree. However, he acknowledged that being able to correct robots’ mistakes could create job loss for workers with a lower skill set.
According to the National Association of Manufacturers, 12 million Americans are currently employed in a manufacturing job. Between 2015 and 2018 estimations state there will be about 1.3 million new industrial robots installed in factories worldwide.
“A lot of people think the automation control using the software and the robots could possibly harm job creation and growth but still promote industry growth,” Jenter said. “I think that’s the point of the robots: to eliminate the unskilled workers whose jobs don’t generally require any problem solving skills. What these robots create instead is more skilled labor, requiring people to get some form of higher education which could definitely hurt people at the lower end of the skill spectrum.”
Berenson said robots serve to improve efficiency by eliminating jobs that often seem redundant with the advent of technological advances, noting that the workforce has adapted to changes like this before.
“The type of jobs that the robots are replacing are probably not the kind of jobs that a lot of people want to do.” he said. “They may have to do it for financial reasons and so on, but you can see many examples in history that jobs that were very tedious have been replaced by machines. For instance, we have a lot less blacksmiths and so on these days because we don’t need to make horseshoes for horses because we use cars. And I’m sure that had a huge impact on the industry at that time. But we’ve come to live with it, and society has adapted to it.”
He added that for those whose jobs are replaced by robots, there will likely be new opportunities available to them.
“With proper policies I believe we can restructure some parts of society to retrain those people, to have them do different jobs,” he said. “If the task is something that is laborious, there’s a strong pressure for a robotic solution, because not a lot of people really want to do that thing.”
Barton said overall, the research will lead to technological advancements beyond what already exists and could serve to benefit the economy as a whole.
“Manufacturing right now is really moving towards this idea of hard intelligence, digital fiber type of manufacturing, where you want to look at things from a system level,” she said. “I think this, in the success of the research we are proposing, will really advance the state of the art in this area and will lead to new capabilities in terms of security and safety.”