Contrary to what you may see at the movies, robots will not come to rule the world any time soon, experts say. Ken Goldberg, University of California, Berkeley professor of industrial engineering and operations research, and Otherlab CEO Saul Griffith, spoke at event focused on the factories of the future titled “Bold Bets: Tomorrow’s Industrial Entrepreneurship.” Both agreed that having humanoid-like robots in the near future should be left to science fiction.
“Machines are very good at lots of things: precision, repetition. But they are extremely bad at any type of creativity or insights,” said Goldberg. Intuition, something humans take for granted but machines simply cannot learn on their own yet, is important. And future automation systems need to be able to understand when to call for help when faced with a decision.
“Machines are overweight, blind, stupid, slow and weak,” said Griffith, explaining that current humanoid robots have many huge components, cannot truly see what they are doing, only do what they are told (like playing back a prerecorded macro), and for the amount of power they demand, are much slower and weaker than their human counterparts. It takes a human about 100 or 200 watts of power to lift 30 pounds; it takes an industrial robot about 10 kilowatts to do the same action. Griffith believes these limitations need to be optimized if we are to truly have mobile and smarter robots.
But there is hope on better automation and robotics: what Goldberg and others call “Industry 4.0 – the Industrial Internet.” As more robots and machines are connected to the Internet, they can be linked together to share data and software. By offloading many of the complex, CPU-intensive calculations of today’s robots to the cloud, robots of the future could be lighter weight and consume much less power, with better performance.
Risk-taking at the university
While some may be funded by corporations, these types of innovations largely emerge from academic environments. “Universities are well-setup to [take risks],” explained Goldberg. “The tenure system and various approaches that let you take a chance let you do something that may fail. Industry is different; there you can fail and you can lose your life savings or lose your job. Here in the university…you can take all kinds of chances.”
Griffith wishes that universities were even riskier and bolder, but the limitation is the availability of capital.
“You have to shift the risk into the academic community and maintain very close ties to it,” said Eric Benhamou, founder and generation partner of Benhamou Global Ventures and former CEO of 3Com. “There has to be a very close ecosystem, relationships and collaboration between the researchers and the academic community and the people who do more of applied research inside the industrial complex.”
Forging a new factory future
Certain tasks that require intricacies and precision beyond a human’s means, such as computer chip production, are where machines excel. Even then, automated chip factories could see a shift.
Former Intel CTO Justin Rattner believes the factories and fabrication facilities of the future will evolve to shift from massive scales of single production units to more customized production in smaller volumes.
“There are certain things that require an economy of scale,” said Rattner regarding chip manufacturing. “We are very likely to see chip factories that shift from making a few designs at very high volume to make hundreds or thousands of designs in low volume.”
“What has happened in the last 20 to 25 years is that it has gotten that much harder to justify to shareholders spending that kind of money just to get fundamental research done,” said Rattner. “Timelines have moved in. And the most important thing is that people have to think less about invention and more about innovation…getting the idea to a form where a product organization can engage and turn that thing into something you can sell.”
This content was originally published on the Intel Free Press website.
Top image: CITRIS Berkeley