The Year Ahead (And Behind) in Robotics

2014 was light on cool new robots. Or, to be more accurate, robots that actually did anything. The studious sort might point out that automation powered some of the biggest technology stories of the year, including the Chinese “Foxbots” that helped assemble the latest generation of iPhones, and the European Space Agency’s Philae lander, which became the first probe to relay images from the surface of a comet. But these aren’t the bots that most of us are looking for. Some of the most exciting robotic and artificial intelligence (AI) projects that made news last year won’t make an impact for years, or possibly decades. That’s how robot-spotting goes. To follow this field closely is to watch some of the most powerful technology in human history unfold at an often maddeningly slow pace. Despite what you may have seen or read, we’re still in the opening frames of that great, era-spanning time-lapse video, the one that ends with the android underclass marching for the right to reproduce.

But the trends that dominated robotics in 2014 weren’t flash-in-the-pan fads. Last year’s embarrassing drone antics, from buzzing sporting events to crashing in national parks, already seem quaint compared to the meth-hauling drone discovered in Mexico this week. And the climate of AI panic created last year by Elon Musk and Stephen Hawking has already spilled over into 2015, with the disastrous coverage of an open letter on AI safety, and $10M in funding from Musk himself towards research that would avert a superintelligent apocalypse. Hollywood, meanwhile, is still oscillating between the occasional fascinating thought experiment about intelligent machines, and its standard fever dreams of killbots as blood-thirsty as they are boring.

Still, barring some unforeseeable breakthrough, this won’t be the year of the robot. The age of ubiqitous automation is still lurking somewhere over the horizon. But these are pivotal times, nonetheless, when investors are deciding which aspects of the field are the most promising, and the general public is deciding which machines are the most terrifying. Here are the trends that were a big deal in 2014, and that will continue to define robotics in 2015.

Investors Finally Care About Robotics

2013 seemed like a banner year for robotics acquisitions, when Google assimilated seven promising companies, including Boston Dynamics (makers of the Atlas, Big Dog and other Pentagon-funded bots that people have decided to be afraid of). But the search giant wasn’t done, and neither was the rest of Silicon Valley. Google acquired London-based AI startup DeepMind in January 2014 for a reported $400M. In March, Facebook bought a five-person startup called Ascenta for $20M, with the goal of using solar-powered unmanned aircraft to increase access to the internet. Less than a month later, Google purchased solar drone-maker Titan Aerospace for between $60M and $75M, for reasons almost identical to Facebook’s (Titan had originally been in acquisition talks with Facebook). Meanwhile, AI startup Vicarious closed a $40M funding round whose investors included Facebook’s Mark Zuckerberg, as well as Elon Musk, founder of SpaceX and Tesla Motors.

Without context, all of these funding injections might seem sinister, like the text crawl preceding a movie about intelligent machines run amok. With one exception, the reasoning behind these expenditures is more practical than lurid (see the next entry). Google and Facebook want robots to expand their customer base, and with it their advertising revenue. They’re also both interested in deep learning, an offshoot of machine learning that includes algorithms that can, among other things, make sense of natural language, whether spoken or read. Deep learning is an easy target for robo-phobics, because of its evocative name, as well as its neural networks and roots in traditional cognitive science. Then again, it’s also the basis for Apple’s Siri. If you find Siri’s trick ear and constant intellectual pratfalls chilling, the prospect of an AI apocalypse is the least of your worries.

Though we’re less than a month into 2015, it’s already apparent that last year’s spike in robotics investment wasn’t a fluke. As GigaOM reported yesterday, funding rounds for MIT spinoff Jibo (whose social robot will launch later this year) and industrial bot-maker Rethink Robotics account for $51.9M. That’s big money for this field, and Jibo’s financial support is a strong indicator that investors are suddenly willing to gamble on AI-powered bots as high-profile consumer products, rather than restricting their backing to unseen assembly-line machines.

And while this meet seem like a stretch, the $1B in funding recently secured by SpaceX should count as a win for robotics. Among the rocket-maker’s biggest innovations is its reliance on partially to fully autonomous flight control, including its plans to land a self-piloted rocket stage on a self-steered robotic barge.

A.I. Panic Is On The Rise

The only Hollywood-worthy robotics investment in 2014 was Elon Musk’s, a contribution to Vicarious that became the foundation of his ongoing crusade to protect humanity from a threat that doesn’t exist. In comments scattered across the latter half of last year, Musk described AI as an immediate threat, “potentially more dangerous than nukes,” with concrete risks that could materialize in as few as five years. Musk presented his investor status in Vicarious, as well as in DeepMind, as evidence that he had peeked behind the curtain, and seen researchers actively working to counter the doom that is AI that’s capable of “recursive self-improvement.” And Musk has esteemed company—in an op-ed last May, and then again in an interview in December, legendary cosmologist Stephen Hawking issued similar warnings about superintelligent AI as an extinction level event.

The circumstances of Hawking’s op-ed for The Independent are beyond unfortunate: the story was tied to the release of the movie Transcendence, a junk-science bomb that starred Johnny Depp as a dying AI researcher who becomes an evil AI. But Hawking went back into the breach in his BBC interview, saying, “Once humans develop artificial intelligence, it would take off on its own and re-design itself at an ever increasing rate,” and “The development of full artificial intelligence could spell the end of the human race.”

There were other newsworthy discussions of automated doom throughout the year, including two UN conferences that addressed the military deployment of lethal autonomous robots. But, as it turns out, 2014 was little more than a hand-wringing preamble to the state of AI panic currently unfolding in 2015. A widely-covered open letter on AI safety signed by an impressive array of AI researchers has seen its message instantly garbled by Terminator-obsessed reporters. The intellectual breadth and measured, academic tone of the letter was also undermined by the announcement, less than a week after its release, that Elon Musk was committing $10M to AI safety-related research.

The problem isn’t that Musk is putting his money where his unfiltered mouth is. It’s that the video accompanying the announcement completely glosses over sober, near-term safety issues, such as building privacy measures into data-mining algorithms, and leaps ahead to robot overlords. “You can construct scenarios where recovery of human civilization does not occur,” says Musk. That’s very true. But when science fiction, rather than science, is your springboard, you can also construct literally any scenario. Expect 2015 to be rife with torrid, unbounded “what ifs” about imminent robot holocausts. The fact remains, however, that software did not decide to murder anyone in 2014, and is far too rudimentary to make any meaningful decisions in 2015.

Hollywood Is Still Crazy, And Mostly Stupid, About Robots

In Hollywood, on the other hand, intelligent bots are everywhere. In fact, in terms of sheer quantity, 2014 might have been the biggest year yet for fictional robots on TV and in film. They populated smaller movies, like Automata *and *The Machine, along with big budget projects like Interstellar, RoboCop *and *Transformers: Age of Extinction. The Steven Spielberg-produced television series *Extant *decided to not only add an unnerving robot child to the already busy story of an alien-impregnated astronaut, but to kick off its recurring discussions of robot uprisings right in the pilot From AI machinations on the TV series Person of Interest to the scene-stealing soft-bot from the Disney animated movie Big Hero 6, the media landscape was overflowing with intelligent machines.

This is not necessarily a good thing. Setting aside the quality of these movies and shows—whether they worked as stories—the robots and AIs that they showcased often had nothing to with how they work now, or how researchers expect they will in the foreseeable future. Hollywood proceeded as usual, and mostly used bots as visual and narrative tropes. The Autobots were giant aliens who happen to bleed sparks and motor oil, and the lanky humanoids in *Automata *were haunting to look at, but ridiculous as functional technology (as most humanoid designs are).

Person of Interest continues to do fascinating things with AI, and Baymax from Big Hero 6 was apparently inspired by soft robotics research at Carnegie Mellon University. Then again, Baymax is a hyper-competent superhero, while the goal of real-life softbots is to develop some of the weakest and most intentionally harmless of automatons. So there were slick fictional bots and silly ones, but nothing stood out, on a conceptual level, as much as the Scarlet Johansson-voiced AI from 2013’s Her. This is a pretty fallow period for fictional robots, either as explorations of existing tech, or as inspiration for future roboticists.

There’s hope for more interesting machines in 2015, though. Movies like Chappie and Ex Machina will feature bots as apparently sympathetic protagonists, as well as the AMC show Humans. As for HBO’s reboot of Westworld, it shouldn’t be too difficult to make the series’ machines more thought-provoking than the original movie’s blandly menacing robo-gunslinger and android prostitutes with a heart of… nothing. The same goes for The Avengers: Age of Ultron, whose central robotic villain was never a compelling figure in the comics, but, as voiced by James Spader, might at least class up a tired old trope.

Drone Crime Is Here To Stay

While the Federal Aviation Administration spun its wheels on the issue of commercial drone regulations—rules that many thought were coming this year are now expected to be pushed to 2016 or beyond—private drones took to the skies in record numbers. Good times presumably resulted, but so did the stunts, the self-destructive high jinks, and the inevitable crimes. As it turns out, flying a DJI Phantom 2 with a GoPro 3 action camera over a fireworks display results in undeniably beautiful footage. Using one to dangle an Albanian flag over a World Cup qualifying soccer match between Albania and Serbia? That leads to fans roaring en masse at a flying robot, and then chasing soccer players with chairs. And piloting a quadcopter over the National Tennis Center during the U.S. Open can get you arrested, because being a filmmaker doesn’t mean you’re above the law.

The worst offenders of the 2014, however, were a pair of tourists who, in two apparently unrelated incidents, crashed their drones in Wyoming’s Yellowstone Park. One drone wound up in Yellowstone Lake, while the other flew into the famed Grand Prismatic Spring, the largest hot springs in the United States. In the latter incident, the authorities learned of the crash after the drone’s clueless owner asked if they could retrieve it. Both crashes came after the National Park Service had banned the use of unmanned aircraft in all of its 401 parks.

But there’s another dimension to illegal drone operation that isn’t quite so zany. This past week, police in Mexico found a crashed drone with an undelivered payload of 6.6 pounds of meth. There’s no reason to think that an epidemic of drug smuggling by UAV is underway—most commercially-available drones can barely fly for a half-hour per battery charge, which limits their effective range significantly. But what’s obvious, even at this early stage in the history of commercially-available unmanned aerial systems, is that drone crime is a new fact of life. The FAA’s regulations, whenever they come, are important for businesses hoping to fly drones for profit. Those rules won’t, however, make an impact on the dummies, or the drug dealers, who are already breaking existing laws. And as more models hit the market this year, robots are bound to start showing up on more and more police blotters around the country.

Wearable Robots Have Arrived. No, Really.

Finally, a robotics development that comes with no caveats or downsides. 2014 was the year when exoskeletons became legal to operate at home, and out in the world, following the FDA clearance of ReWalk’s Personal Exoskeleton System. This is an interesting milestone, in part because of the generally misleading media coverage of exoskeletons for the disabled. For years, devices from companies like ReWalk and Ekso Bionics have been featured in stories about exos finally becoming a reality. What’s downplayed, or even omitted, in most cases is that these exos were restricted to rehabilitation facilities, and classified as experimental devices. If a test subject or pilot participant were to fall and injure him or herself while using the system, health insurance likely wouldn’t cover the related medical costs. This was true even in rehabilitation facilities, ever since the FDA reclassified exoskeletons in 2013 (they were previously in the same, low-risk category as exercise equipment).

People who would otherwise be confined to wheelchairs are already walking the Earth, strapped into their own personal robots.

What should have been a major story—ReWalk becoming the first and only company (in June) whose exos could go essentially anywhere in the United States—was overshadowed by coverage of restricted-use systems, as well as exoskeleton R&D funded by the Pentagon. That coverage included, to a degree, brain-controlled prosthetics and robot arms, devices that have managed incredible experimental progress, but whose use is still limited to a handful of subjects in controlled laboratory environments. But from news that special effects prop-makers were working on the military TALOS suit, which is intended to increase the amount of body armor carried by special operators, to footage of a double amputee with a pair of nerve-attached bionic limbs, 2014 may have seemed like the year of the body-worn robot, even if none of those high-profile devices were ready for deployment.

Meanwhile, ReWalk’s Personal Exoskeleton System didn’t make many headlines, but it did make history. A handful of customers with spinal cord injuries, people who would otherwise be confined to wheelchairs, are already walking the Earth, strapped into their own personal robots. And just yesterday, ReWalk announced the first reimbursement for an exoskeleton by a private insurer in the United States. The significance of this precedent can’t be overstated. Previous ReWalk systems in this country have been paid for by the Veteran’s Administration, limiting their use to veterans. Later this month, a 29-year old man who’s been in a wheelchair for 7 years due to a workplace injury will take home a personal exoskeleton, which is his to own and operate wherever he chooses.

As other companies work towards FDA clearance, this kind of story is likely to become so common in 2015, it won’t merit any coverage at all. That’s the funny thing about robotics. The more it integrates into human life, the less noise humans make about it.