Google Has Made a Mess of Robotics

Its scattered, ambiguous, frequently abandoned objectives for its string of big acquisitions have hurt the whole field.
Illustration: Oscar Bolton Green

In 1982, Blade Runner introduced the world to replicants, director Ridley Scott’s term for the alarmingly lifelike robots of author Philip K. Dick’s imagination. In the past few years, there have been two buzzy follow-ups. One, a Scott-produced sequel called Blade Runner 2049, grossed a disappointing $31.5 million during its Oct. 6 opening weekend. The other, a secretive robotics division at Google named Replicant, has flopped much harder.

Andy Rubin, the Google executive who created Android, began scooping up roboticists in 2013, acquiring at least nine companies in all. The shopping spree, while a side bet next to Google’s search-driven ad revenue, seemed to establish the company as the likely leader in building an army of intelligent machines. Its acquisitions were doing some of the most exciting work in the field, and its vast reach and resources made Google an ideal place to combine those efforts into a single vision. The industry welcomed its new overlord. “People were very enthusiastic about it,” says Rosanna Myers, chief executive officer of startup Carbon Robotics.

Not anymore. None of the acquired companies have robots in use beyond the offices of Google’s now-parent company, Alphabet Inc. At least three key robotics chiefs who joined in that 2013 wave left the company in the last few months, and, because four years is the typical vesting period for Google stock options, they probably won’t be the last. At this point, Myers says, the exodus counts as a win for robotics, since many of the brightest minds in the field have essentially spent the past few years trapped in a time capsule. Ultimately, Google’s run on roboticists “held the industry back more than moving it forward,” she says. Alexa Dennett, a spokeswoman for Alphabet’s skunk works, X, says its robotics projects will likely take at least five more years to come to market because substantial technological advances take time.

Google had tried working in the field before Replicant. As early as 2010 a team began work on “cloud robotics,” systems where groups of robots share data and code and most computing happens on remote servers. It also tinkered with toy robots for kids. The company dropped these efforts because of concerns about market size and privacy, according to people who worked on them. Rubin’s plans indicated that it was taking a much bigger swing.

The companies Google bought were working on a wide range of projects. Boston Dynamics made mobile robots resembling four-legged animals. Redwood Robotics and Schaft Inc. were working on arms, andMeka Robotics was building a humanoid automaton meant to work alongside people in factories. Industrial Perception Inc. specialized in computer vision. Bot & Dolly and Autofuss worked on related applications in media and design. Holomni’s plans vaguely involved mechanized wheels.

Most of the companies were unlikely to yield products in the near term, says Rich Mahoney, who ran the robotics division of researcher SRI International and was on the board of Redwood Robotics in 2013. “My view was that they were all acqui-hires or seeding a range of technologies Google wanted to build future opportunities on,” he says. But it never became clear what opportunities Google was pursuing. Its new robotics division communicated nothing to the outside world, and the company never articulated its vision even to employees who came along with the acquisitions, say people familiar with the division.

Rubin left Google in 2014 to create a venture firm and a hardware-focused startup incubator. Replicant was disbanded soon after his departure, and its workers fanned out across Google, other Alphabet subsidiaries, and elsewhere. Google’s failure to follow through on a broad plan for robotics mirrored similar problems it’s had elsewhere in commercializing technological gains, says Ryan Hickman, a cloud robotics pioneer who now runs home-robot startup TickTock AI Inc. “Google will continue to crank out engineering marvels, but other places are taking the lead on product innovation,” Hickman says.

Google’s difficulties with robotics are hardly unique; most companies in the field fail. But some failures build the foundation for future success. One prominent example is Willow Garage, an incubator that spun off three of the companies Google acquired in 2013. Willow Garage shut down in early 2014 but is responsible for the Robot Operating System, open source software now incorporated into a wide range of equipment. By contrast, Replicant swallowed promising projects and stifled them, says Jeremy Conrad, a partner at hardware incubator Lemnos Labs. “These were some of the most exciting robotics companies,” he says, “and they’re just gone.”

The robotics industry is now worth $24.5 billion, according to VC firm Loup Ventures and trade group International Federation of Robotics. More than half of that comes from industrial machines like those used in e-commerce warehouses. The robotics market is projected to double in five years.

Google hasn’t entirely receded from the scene. Besides Waymo, Alphabet’s autonomous-car division, Google is investing in drone development. For the last year it’s also operated what has come to be referred to as the “arm farm,” a room somewhere at its Mountain View, Calif., headquarters where at least 10 robotic limbs are being refined to grasp and manipulate various objects. “There’s still a ton of work needed to make robotics less robotic,” says Vincent Vanhoucke, principal scientist at Google. “But we believe these building blocks could lead to big breakthroughs.” A few Replicant projects moved inside X two years ago.

Other behemoths are becoming more prominent. Inc. is steadily pushing automation, including with a robotics division it built from its $775 million acquisition of Kiva Systems in 2012. Toyota Research Institute, a robotics lab the automaker started with a $1 billion investment in 2015, has also been a big draw for researchers. At the moment, none of these places, including Google, looks like the decisive leader, says Mahoney, the former SRI robotics chief. “In the end, it’s going to come down to the right set of people coming together, not just resources,” he says. “That hasn’t happened yet.”

    BOTTOM LINE - The top roboticists netted by Google starting in 2013 have been given little direction and have little to show. They’re starting to strike out on their own.
    Before it's here, it's on the Bloomberg Terminal.