Home > Cool Tech > Why is Google building a robot army?

Why is Google building a robot army?

Meka UT Dreamer

Google’s recent acquisitions include Meka, the company behind the humanoid Dreamer “sociable humanoid head.”

It’s not unusual for Google to capture headlines in the technology press, but one of Google’s latest moves raised even the most-jaded technorati eyebrows: Google acquired functional robot maker Boston Dynamics for an undisclosed amount. But that’s just the icing on a huge robotics cake Google’s been shoveling into its maw. This year alone, Google has acquired no fewer than eight robotics firms, including Industrial Perception, Redwood Robotics, Meka, Schaft, Holomni, plus the startups Bot & Dolly and Autofuss. And Google is committing brainpower too: In addition to putting former Android chief Andy Rubin in charge of robotics, earlier this year Google hired machine learning expert Geoffrey Hinton, and last year brought famed inventor and AI advocate Ray Kurzweil on board. Between initiatives like Google Glass, Knowledge Graph, and now robotics, Google is positioning itself as a hotbed for future technology development.

But seriously: robots?! Sure, industrial robots have been around for decades, and a few us of even having Roombas patrolling our homes for stray breadcrumbs. But could Google finally be the company to bring us closer to the robots we’ve always dreamed of from fiction? Let’s take a closer look at what Google’s new toys do – and what its own interests may be.

What is Google buying?

Google has been tight lipped about robotics: The company has not announced any products or plans, or even uttered some broad suggestions about its robotics research. That means any Google robotics products and platforms are years away. However, we can glean some insight from the companies and technologies Google buys.

Could Google finally be the company to bring us closer to the robots we’ve always dreamed of from fiction?

Boston Dynamics comes closest to a classic vision of autonomous robots that operate in the real world, or (in movies anyway) take over the real world. Boston Dynamics makes some of the most successful functional robots on the planet. It’s PETMAN humanoid robot is a fully-articulated humanoid — although it’s designed to wear and test chemical protection gear, not roam around independently. However, Boston Dynamics is probably better known for Cheetah and BigDog: slightly creepy four-legged robots that can follow leaders and navigate difficult, uneven terrain inaccessible to wheeled vehicles. Cheetah is faster than any human, while BigDog (and its cousin LS3) can carry up to 400 pounds of gear, and another version can also famously hurl objects — a useful trick for clearing rubble, or getting supplies to people cut off from roads in war or following a natural disaster. Unsurprisingly, Boston Dynamics’ main client is the Defense Advanced Research Projects Agency (DARPA)—and Google has indicated it intends to honor those contracts.

Schaft is a Japanese company, and it’s claim to fame involves capacitor-driven, water-cooled motor systems that are much more powerful than the battery-driven electric servo motors used in most robotic limbs – that means stronger, smaller robotic limbs. Although Schaft makes bipedal robots designed for first-responder situations, the company had more recently been focusing on developing and marketing small, powerful, highly-articulated robot arms as standalone components in other robots or in research fields.

Boston Dynamics LS3 AlphaDog

Boston Dynamics’ LS3 AlphaDog can go where few wheeled ‘bots can – and carry up to 400 pounds along with it.

Another Google acquiree, Redwood Robotics, is all about arms: specifically arms that are cheap, easy to configure, and designed to be operated safely alongside real people without safety cages or complex programming. Redwood’s target market is most likely manufacturing and warehousing, although programmable arms can also be valuable in labs – all areas where Boston-based Rethink Robotics has been making some inroads with its Baxter robots.

Same story with Industrial Perception, which has focused on developing robots with 3D vision capabilities to identify and interact with objects in their environment, like picking items out of bins, and processing packages of varying size and weight. Again, the initial focus is on manufacturing and logistics — think shipping, warehousing, and transport. Need to move around warehouses? Google acquisition Holomni essentially makes powered casters that can move in any direction. They probably aren’t much use in the muck and mire, but should be great for getting around warehouses and hospitals.

Meka Simon Robot Torso

Meka’s Simon robot moves with an impressive 13 degrees of freedom, plus moving eyes, eyelids and ears.

On the cuter side, Google has also acquired Meka, which helped produce Dreamer, a “sociable humanoid head” designed to interact with people. The idea is that folks are more likely to feel affinity with a character-like robotic face than with a steel box and a row of blinky lights. However, Meka is better known for robots with limbs featuring compliant force control: pincers, hands, and arms that can be pushed out of the way and won’t accidentally crush your fingers in a handshake. Meka even built a $340,000 research bot with a Microsoft Kinect sensor in its head (plus a high-res camera for 3D vision). Meka’s designs are intended to be safe around people in everyday situations and work environments; it was also part of the joint venture that founded Redwood Robotics.

What about doing something real with robots today? That’s what creative sister company’s Autofuss and Bot & Dolly do, except they don’t make robots. Bot & Dolly’s systems automate very complex photography for motion pictures, stop motion animation, and more – their systems helped propel this year’s feature film Gravity. Autofuss is more of a traditional creative agency, although they’ve been using industrial robots from the automotive industry reprogrammed by Bot & Dolly to produce work, including a number of ads for Google’s Nexus product line.

Why robots?

At a basic level, robots are machines that do things humans are unable or unwilling to do. Sometimes, robots can do those tasks faster, more accurately, and tirelessly — perhaps better than humans — but only if the tasks are well-defined and the robots’ environment tightly controlled. The classic examples are in manufacturing and industry — the sorts of repetitive, precise tasks that can be mind-numbing (or even dangerous) for people. Unimate, the first programmable digital robot, went online in 1961, lifting and stacking hot pieces of metal from die-casting machines at General Motors. It spawned a new era in factory automation. The last half-century has seen many advances in sensors, software design, materials, motors, and more, but the basic idea is unchanged. Several of Google’s robots acquisitions — Industrial Perception, Redwood Robotics, Holomni — fall into this category.

Paro elder care

The Paro elder-care robot gives patients many of the benefits of a real pet, without the upkeep.

Google isn’t moving around slabs of red-hot metal, but warehousing and shipping are strong possibilities for what Google ‘bots might do. Like manufacturing plants, many modern warehouses are highly structured facilities, with floorplans designed to handle carts and forklifts, and sophisticated tracking systems using RFID and barcode technologies. These environments are ripe for automation using robotic systems: Instead of sending human runners out into a warehouse to assemble individual customer orders, an order could be placed into an inventory system automatically once a customer’s payment is approved, then robotic systems could find the inventory, pack the box, and even load it onto a truck for delivery. Other automated systems could receive new inventory — perhaps loading it right off trucks — and store it appropriately in the warehouse. For high-turnover businesses — like distribution or food, groceries, and other perishables — automated warehousing could bring major efficiencies. It’s not as sexy (or creepy) as a self-driving car pulling up to your curb, then a four-legged robot scampering up your stairs to deliver a package, but it’s a far more likely to happen first. Other areas ripe for automation include pharmacies and libraries: Robots are already big in drug manufacturing and packaging, and have been preparing chemotherapy medication for years. Similarly, automated storage and retrieval systems have been a big business for decades.

They probably aren’t much use in the muck and mire, but should be great for getting around warehouses and hospitals.

But the tech Google is buying doesn’t just apply to boring old warehouses. Recent advances in robotics have some startling new possibilities. Believe it or not, one is patient care. For more than a decade, therapeutic robots like the cute, plush Paro have proven remarkably effective at engaging and calming patients in hospitals and nursing homes. It’s similar to pet therapy, without the responsibility of a real animal. Patient-care robots of any size can also keep a virtual eye on patients day and night, monitoring their health, dispensing or preparing medication, or summoning help if they detect something wrong. Human-sized robots with compliant force control have tremendous potential for providing physical assistance – reaching shelves, opening doors, retrieving objects, helping with meal preparation and personal hygiene. However, these applications are likely to be further in the future than warehouse automation. Right now, unstructured home environments with stairs, thresholds, raised floors, furniture, stuck doors, rugs, clutter, furniture, variable lighting, and pets are tough for robotic systems to manage. After all, the Roomba can’t even shake off a cat. You don’t want a robot ripping the door off your dishwasher because it can’t figure out the latch.

Bot & Dolly Scout

Bot & Dolly’s incredibly precise, programmable camera robots open a whole new world of special effects for filmmakers.

Robots can also do things that are too difficult or dangerous for humans. A great example is the PackBots that were sent into the World Trade Center after the 9-11 attacks and into the Fukushima Daiichi nuclear plant after the 2011 earthquake an tsunami — although they’re teleoperated by humans rather than acting autonomously. A more autonomous example is NASA’s Curiosity rover on Mars. It can analyze images of Martian terrain and compute its own safe-driving path without input from Earth-bound human operators. Technologies from Boston Dynamics and Schaft are directly applicable to these tasks. Of course, the military is keenly interested in these applications, but there are tons of civilian uses too, from emergency and disaster response, search and rescue, and exploration. Imagine robots that can look for lost climbers or skiers during a storm, roam the arctic icepack documenting climate change, or help manage natural parks, livestock, and crops.

What’s in it for Google?

Google’s bet with robotics isn’t another move for supremacy in mobile technology. The robotics acquisitions do not mean the 2014 holiday season will bring us cute, pint-sized Googlebots, ready to entertain the kids, run apps, and snap pictures of every product in your home so Google can know you better, then sell that information to advertisers.

We hope, anyway.

At a basic level, robots are machines that do things humans are unable or unwilling to do.

Google’s core business might be advertising (and search and Android) but the company is not averse to “moonshot” projects — self-driving vehicles (which it’s been working on since 2009), Google Fiber, bringing high-speed Internet to off-the-grid areas with balloons, and even Google Glass are examples of Google flexing its wallet in an effort to push technology into areas that aren’t part of its bread-and-butter businesses. Google is actively trying to create the future.

That said, Google isn’t in the habit on throwing money at any cool idea that comes along. So far, all of Google’s “moonshots” are aimed at areas that have clear commercial potential. Google won’t necessarily sell Google-branded self-driving cars or household robots. However, the company may well create platforms for driverless cars and robotics systems that other companies use for their own products. Long term, Google may earn money from that from licensing systems or patents, providing platforms or services that glue those systems together, or simply giving away the technology to broaden Google’s ecosystem — like it does with Android.

But make no mistake: The robots are coming. But they’re most likely coming to big businesses before they fly by overhead or scamper down the street.