Japanese robot maker Fanuc Corp. has strengthened a tie-up with an artificial-intelligence venture as part of its effort to develop industrial machines that can learn and repair themselves.
Fanuc said Friday it plans to acquire a 6% stake in Preferred Networks Inc., a Tokyo-based startup that is also known as PFN, for ¥900 million ($7.3 million). The two companies signed a basic partnership agreement in June to develop robots that require less human supervision.
“A couple of our yellow robots have already installed at the PFN’s office to collaborate in research and development,” Fanuc Chief Executive Yoshiharu Inaba said at a news conference. “We believe the PFN is the world’s most advanced in terms of deep-learning expertise.”
Preferred Networks is a small developer of AI technology, including “deep learning,” a kind of self-study program for machines. Instead of being programmed solely with fixed rules, computers learn by themselves how to achieve a task.
Companies in the business of designing machines used in factory automation are racing to make their robots smarter to improve efficiency.
Fanuc aims to use Preferred Networks’ expertise to develop robots that can optimize their own workflow and repair other robots.
“Marrying its own robots with AI technology is a necessary step for Fanuc’s long-term strategy to keep its competitive edge,” said Kenjin Hotta, an analyst at Macquarie Securities. “[Fanuc’s] hardware is great, but it needs to catch up with European rivals on the systems front,” he added.
Toru Nishikawa, chief executive of Preferred Networks, said the two companies hope to unveil prototype robots by early next year.
Fanuc produces robots for factories that help assemble products such as Apple Inc.’s iPhones and Tesla Motors Inc.’s electric cars. It has enjoyed hefty profits from the global smartphone boom, but its share price has fallen more than 15% in recent weeks on concerns about the Chinese economy.
Deep learning is one of the hottest investment targets for global tech giants, including Google Inc., which acquired a London-based startup for an estimated $500 million last year. Deep-learning technology is used in many ways, including automatically tagging names of people in photographs uploaded to the Internet.
Nippon Telegraph & Telephone Corp. paid ¥200 million for a stake of less than 10% in Preferred Networks last year. Preferred Networks has also teamed up with Toyota Motors Corp., Panasonic Corp. and Cisco Systems Inc.