Anaheim, Calif. — IBM's expert on the Internet of Things says cognitive computing, using sensors and controllers, can give factory technicians accurate predictions of coming problems by giving them probabilities of what could go wrong.
Krish Dharma said that level of technology is in the early stages of adoption, but the pace has picked up significantly. He thinks it will be five to 10 years before it becomes widely used.
"It's going to enable the technician to make the best decision," Dharma said. But the computer system is not going to make all the decisions, he added.
Krish is director of industry solutions for IBM's U.S. industrial market, focusing on automotive and aerospace, and defense. He spoke at an Industry 4.0 conference held May 10 at Antec in Anaheim.
Officials from machinery makers joined information technology gurus at the conference to describe a fully connected manufacturing world. They said many innovations, such as smart molds and remote analysis by an equipment company's experts anywhere around the world, will become standard features on new machines, not just options.
Bruce Catoen, chief technology officer at Milacron Holdings Corp., made an analogy with intermittent windshield wipers: Nobody is going to pay extra for that feature.
The term Industry 4.0 is more commonly used in Europe than the United States, so some speakers also used the Internet of Things moniker. "The IOT is a journey, it's not an event. So you will see continuous improvement on that," Catoen said.
Jurgen Giesow, director of engineering and technology at Arburg Inc. in Rocky Hill, Conn., said the concept of an Internet of Things may be different for each company.
"We need a combination of complexity and flexibility, that's really what Industry 4.0 is giving us," he said.
IBM made history with Watson, a supercomputer with artificial intelligence that won on "Jeopardy." So computers can be "smart" and can use reasoning skills. Dharma painted a future of computer systems helping humans run factory networks better — from manufacturing and the supply chain to marketing to closely teaming with customers.
Dharma, who is based in Costa Mesa, Calif., said the big problem with manufacturing is "tribal knowledge," or the knowhow of company veterans, when it doesn't get passed down to others. "Most often you don't capture the solutions from prior problems," he said.
And skilled veterans can leave a company at any time, taking the knowledge with them. Dharma said cognitive computing systems can hold onto the knowledge, after being "trained" by experienced workers. "You're also creating machine learning modules," he said.
For example, if the molding machine is running too hot, the cognitive system can analyze the data and say there is a 90 percent chance that the one issue is the cause, then list some other possible reasons.