Simpler, more integrated bots on the way
Industrial robots are poised for change.
They will become more integrated, easy to use, and widely deployed, especially in China, which is emerging as a center of robotics innovation. Large fulfillment operations run by the likes of Alibaba and Amazon will be drivers in the next stage of their growth.
Those were some of the views of Rodney Brooks, a robotics pioneer and current Chairman and CTO of Rethink Robotics, speaking in a keynote at an event here.
Today, more than half the cost of a factory robot goes to systems integrators who configure it with sensors and train it, typically writing custom programs and generating proprietary data that stays on the factory floor.
By contrast, tomorrow’s industrial robots will come with integrated sensors and computer vision. They will be trained without elaborate coding on open platforms that send their data to cloud services. And widely used programmable logic controllers (PLCs) will become “art projects,” predicted Brooks.
“Today’s business model is going away…we are in an industry where deployment speed is like molasses, but it’s not going to be like that forever,” he said.
Brooks imagined a future where untethered robots respond to voice commands, freeing their supervisors from today’s interactions via scripting languages. “In the last five years, we have seen a tremendous increase in functionality in speech systems…speech is going to be fine in factories.”
“The robot industry is squarely stuck in the 20th century… [but] there are so many little startups coming along that things are going to happen, so start worrying because hundreds of thousands of entrepreneurs are coming,” he told an audience of several hundred industrial robot builders.
China alone has several hundred robotics startups today. They are fueled by a government industrial policy that wants to maintain China’s standing as a global manufacturing center. Robots are also seen as key to dealing with a labor shortage for factory jobs in which turnover rates vary from 16% to 30%, said Brooks.
“A lot of these China startups are low-end manufacturers of cheaper, light industrial robots with six degrees of freedom, dragging prices down, so it’s hard for U.S. and European companies to compete there…we can scoff at their level of innovation, but it’s only a matter of time before it increases.”
So far, promises of millions of robots on Foxconn lines in China have not come true, in part due to the challenges with programming today’s robots, said Brooks, who had his original Roomba robotic vacuum cleaners made by toy manufacturers in Shenzhen. With 20 million units sold, the Roomba became the largest selling robot to date.
Giant fulfillment operations run by the likes of Alibaba and Amazon will also be big drivers for next-gen robots, probably using machine learning.
“Amazon Robotics employs 700 people just in Boston, but the robots can’t do pick-and-pack operations, so they are hiring every Christmas,” said Brooks, noting that Amazon funds a university challenge for robots that pack.
“These fulfillment centers have an incredible need for pick-and-pack operations where every package is unique. That will drive development in robotic arms in ways we haven’t seen, and that will, in turn, impact factory automation in ways we can’t predict,” he said, calling it a “tremendous” demand that “will disintermediate many people in robotics.”
One of the big challenges ahead is dealing with safety, an issue for which the Robotics Industry Association hosting the event has set up several working groups.
Rethink’s own robots currently set payload and velocity constraints as a first step toward safety, but that won’t work for factories that need heavy loads handled quickly. Some companies are now using external cameras and controls to monitor and manage robots, but more solutions are needed, said Brooks.
Long-term, an aging population may demand “speech-controlled robots that help you in and out of bed — that’s an incredible driver of safety,” he said.
Rethink’s current robotic arms already embed cameras, force sensors, and smarts to learn and remember actions. “We can’t do all the things high-end PLCs do, but we can do a lot of it,” he said.
Another big change will be a move to wireless networking on the factory floor, he predicted.
“Many of our customers don’t have any network on the factory floor, and many keep their data walled off…wireless is changing things rapidly; there will be a whole different set of infrastructure and policies for what comes on and off the factory floor,” he said.
Unlike startup Embodied Intelligence, recently launched by Berkeley robotics researchers, Brooks does not believe that VR headsets will find wide use as an interface to robots. Instead, he sees supervisors monitoring and managing automated factories with tablets.
Written by: Rick Merritt, Silicon Valley Bureau Chief, for EE Times.