By Takashi Mochizuki
(FROM THE WALL STREET JOURNAL 12/18/15)
TOKYO — Toyota Motor Corp. is acquiring a stake in Tokyo-based machine-learning venture Preferred Networks Inc., the
latest sign that the world’s largest auto maker by sales is accelerating its development of self-driving cars.
“This will strengthen our partnership as our collaboration has begun to show various results,” said Kenichi Murata, a
Toyota engineer heading the company’s connected car research.
Toyota aims to make some of its cars fully self-driving on highways by around 2020, meaning vehicles would get on and
off the highways and change lanes without driver input. That would be a step forward from the current so-called advanced
driver assistance system, which helps drivers maintain an adequate distance between cars and park smoothly.
The purchase price indicates a sharp and recent rise in the value of Preferred Networks, which spun out of software
developer Preferred Infrastructure Inc. in 2014. In August, Fanuc Corp. paid 900 million yen for a 6% stake, valuing the
venture at 15 billion yen about three months ago. Nippon Telegraph & Telephone Corp. said in October 2014 it paid 200
million yen for less than 10% of the company.
Competition in autonomous driving technology has been heating up. In addition to auto makers such as General Motors
Co. and Tesla Motors Inc., technology companies such as Google parent Alphabet Inc. have been ramping up research and
development toward the goal of driverless cars.
Toyota’s investment in Preferred Networks follows its announcement in November that it plans to set up a robotics and
artificial intelligence research center in Silicon Valley by January, pouring $1 billion into the project during the
next five years. Mr. Murata said the venture’s engineers are likely to work with people in the planned new institution.
Preferred Networks has been a partner with Toyota since October 2014, but the two companies haven’t released details
of their progress. They said Preferred Networks will show an exhibition in Toyota’s space at January’s Consumer
Electronics Show in Las Vegas, demonstrating how artificial intelligence can help teach cars not to crash.
“We hope our deep-learning expertise will contribute to making automated driving technology safer,” said Toru
Nishikawa, chief executive of Preferred Networks.
Focusing on machine-learning technology, including “deep learning,” a branch of artificial intelligence that enables
machines to learn on their own without much human supervision, Preferred Networks has entered many partnerships with
major companies, including Panasonic Corp., Nvidia Corp. and Cisco Systems Inc. since the company’s launch last year.
The partnership with Fanuc has been developing machines that can figure out how to assemble devices and even repair
At a recent robotics exhibition in Tokyo, the two companies showed off their early achievements.
In one demonstration, a Fanuc robot powered by the venture’s deep-learning software learned in eight hours how to pick
up components from a box most efficiently, a program a veteran engineer would have taken days to write.
“The achievement is impressive because the technology can be applied to many other industrial machines, and that would
quickly change how we work,” said Toshimitsu Kawano, managing director of Beckhoff Automation Japan.
(END) Dow Jones Newswires 12-18-150247ET Copyright (c) 2015 Dow Jones & Company, Inc.