Japanese group hones system to enable factory robots to communicate and learn to prevent glitches
With more than 400,000 of its yellow robots already reigning on the world’s factory floors, Fanuc has a new goal for the digital era: connecting the brains of industrial robots.
In a rare recent tour by the Financial Times of Fanuc’s annual robotics exhibition, the biggest crowd-pleaser was its new system that will enable its vast army of factory robots to communicate with each other and learn to prevent glitches that cause delays or halts to manufacturing lines.
The shift towards the so-called industrial internet of things comes as the Japanese manufacturer grapples with a steep decline in profits caused by an economic slowdown in China, the world’s largest user of industrial robots.
Yet those concerns were nowhere in evidence as a smiling Yoshiharu Inaba, Fanuc’s chief executive, stood at the entrance to welcome more than 7,000 clients from around the world to examine 32 of its robots on display.
The annual event is one of the few occasions when the gates of Fanuc’s headquarters at the foot of Mount Fuji are opened to large groups of outsiders. The secretive company has ruthlessly guarded its technologies inside the “yellow kingdom”, where it has honed its manufacturing skills to become the world’s largest maker of industrial robots.
The company’s robots are famous for helping build vehicles at General Motors, Volkswagen and Tesla. Less well known is Fanuc’s history in artificial intelligence research, which dates back more than a decade. The company also has a sophisticated surveillance system that notifies customers when their robots need to be upgraded or repaired.
On display at the exhibition were robots that automatically learn the fastest way to pick up randomly arranged screws, and machines equipped with sensors so they stop moving when a human approaches — painted green instead of Fanuc’s typical bright yellow, so users can easily identify that they are safe to go near.
“Our future growth potential is in network and software,” says Tetsuaki Kato, general manager of Fanuc’s robot software development laboratory. “What our customers like the best about our machines is that they don’t break down, and that’s where we have won trust so we can do the same with our network technology.”
For its new service connecting its robots, Fanuc teamed up with Cisco, US factory automation systems maker Rockwell Automation, and Preferred Networks, a Tokyo-based machine learning start-up. The service will be available from the end of this year.
The Japanese group is not alone in trying to bring artificial intelligence and connectivity to manufacturing to create smart factories. German rival Kuka recently said it would work with Chinese telecoms company Huawei in cloud computing and mobile technology to link up industrial robots.
Google, Amazon, Facebook and other US technology groups have heavily invested in deep learning — a form of AI that tries to mimic the function of layers of neutrons in the human brain.
Global spending on robotics will reach $ 135bn by 2019, estimates IDC, increasing at an annual compound rate of 17 per cent.
But executives at Fanuc and Preferred Networks say their battleground is in manufacturing, where they have an advantage thanks to the data collected by Fanuc’s 400,000 robots worldwide.
“Japan’s manufacturing is stronger than the US. We can’t fight with information stored on the cloud, but the yellow robots are everywhere in the world and they gather data. If you ask whether Google can do the same, the answer is no,” says Junichi Hasegawa, chief strategy officer at Preferred Networks.
Instead of storing data in the cloud, which can be an issue for users concerned about speed and security, Fanuc’s system envisions placing the data on a device that can be installed at factories, and connected to existing equipment and even robots made by rivals.
Similar to accessing an app store on smartphones, the system’s users will eventually be able to download applications that match their factory needs. With learning capability, the robots will share and analyse data, and collaborate to detect glitches that would interrupt production.
Nevertheless, Fanuc executives admit that it will take time to convince customers to trust the machines’ intelligence.
“Is it really OK to automatically use what the machines have learnt? For now, we are thinking of installing safety measures to monitor the system. What’s new can be scary,” Mr Kato says.
Copyright The Financial Times Limited 2016. You may share using our article tools.
Please don’t cut articles from FT.com and redistribute by email or post to the web.