Baxter the Friendly Robot

YDNSeptcover

Nine faces gather around a conveyor belt watching miniature plastic cups roll by. One of these faces belongs to Baxter. Baxter is a robot. He — we’ll call him “he” for now — has a swimmer’s body: two big arms, a powerful chest, and tiny legs. The other eight faces include a factory owner, several automation engineers, and a reporter (that’s me). Together we wait to see if the rumors are true: could this awkward new robot with cartoon eyes save American manufacturing? And at what cost to the humans who came before him?

The floor of the plastics factory is covered with cardboard boxes emblazoned with “Made in the USA.” High-pitched crackles, like the sound of pouring a hundred boxes of cereal into a giant bowl, shoot from the ceiling as bits of plastic are sucked out of giant silos, pushed through thick tubes, and poured into the belly of a plastic molding machine.

The bits are melted and the hot liquid is squirted into a mold. The machine hisses as hydraulics squeeze the mold tight and then pull back to reveal the product: eight small cups. These cups, slightly more durable versions of the ones you’d find atop a Pepto-Bismol bottle, will be used to deliver doses of medicine to patients around the world. When they arrive at hospitals, they will be fed into automated medicine dispensers. But first, the cups need to be stacked.

An instant after the plastic hardens, a robotic hand shoots into the maw of the press. The hand belongs to a Wittman robot — a traditional industrial machine. It looks like a cross between a World War I tank and a Rube Goldberg device. The Wittman robot moves with such speed and disregard for human life that it is kept inside a Plexiglas cage. Its hand grabs the cups and lifts them straight up like lightning. Then its arm rolls 10 feet to the right and plummets to the conveyor belt below, stacking the cups.

The final challenge falls to Baxter: the towers of cups must be dropped into plastic bags. As the parts roll down the assembly line, Baxter’s head turns to look at the objects. His two big eyes squint, to let everyone know that he is focusing. He slowly moves his hand into position, gingerly lowering his claw over one of the stacks. We’re all rooting for him: It’s his first day on the job, and with the lightning-fast Wittman robot in the background, Baxter’s meditated pace makes him quite the underdog.

But Baxter keeps dropping the cups. Justin, a thin 18-year-old hacker-turned-automator moves in to help him out.

Justin leans on Baxter’s shoulder as he navigates the robot’s smartphone-like interface. Like a loving father teaching his son how to properly hold a golf club, Justin grabs Baxter’s hands and shows him how to do the task. This is one of Baxter’s most revolutionary features: teaching him how to do something doesn’t require any coding.

As we watch, the engineers and salesmen trade anecdotes about recent automation jobs and the uncertain state of the industry. “Times are lean,” says one engineer. “I get phone calls all the time from small business owners saying,  ‘Help me …  if I don’t do something soon I’m going to lose my business to China.’” Another rambles about the deficit and “Obamacare.”

American manufacturers have taken a beating for the last two decades as more and more production has been offshored. The newest industry strategy to bring factory jobs back home consists of automating as many jobs as possible. Rethink Robotics, a Boston-based startup led by some of the world’s most brilliant roboticists, hopes to use this strategy to ride to the rescue of American manufacturing. Their knight-in-shining-plastic-armor is a brand-new species of robot named “Baxter.”

Baxter provides factories of all sizes with a new form of low-cost labor. But his true strength is that he does more than just produce: Baxter is designed to seem friendly and be a fast learner. Basically, he’s a “drag-and-drop” replacement for a human being. According to Rethink, robots like Baxter empower factory workers by giving them user-friendly machines they can manage themselves. But in the past, employers have cared far more about productivity than robots’ facial expressions. Meanwhile, the workers are only concerned about keeping their jobs. Is Baxter “helping” either? Or is “saving American manufacturing” just a euphemism for taking jobs away with a smile?

 

Baxter’s stomping grounds in Boston are a sort of visual metaphor for the resurgence of American manufacturing. Rethink Robotics Inc. is housed in a giant brick building with a towering smokestack, which, no longer releasing manufacturing fumes, hosts a smattering of satellite dishes and antennae.

I went to Rethink to spend some quality time with Baxter and his folks. I was curious to find out if the robot lived up to all the hype it has been getting in the tech world and the media. I also wanted to understand Rethink’s vision of how technology can save manufacturing. Is the future of robot-filled industrial America as utopian as they make it out?

Baxter was dreamed up by one of the world’s most prolific geeks: Rodney Brooks, the Australian inventor of the Roomba robot and the former head of MIT’s Artificial Intelligence Laboratory. Brooks left MIT in 2008 to found Rethink Robotics, on a mission to rescue American manufacturing through technological innovation.

Baxter, Brooks’ most recent brainchild, solves two major problems. The first is that automation is expensive. Lots of small factories across America can’t afford to hire engineers to build customized robots. Baxter — with a sticker price of $22,000 — provides these factories with an easily trainable robot that can quickly pay for itself.

The second problem is that traditional industrial robots are scary: they are so complex that ordinary workers can’t interact with them — much less program them — and they often pose safety hazards that require them to operate within clunky cages. Baxter, on the other hand, works at a human’s pace and has the common sense to stop moving when something (or someone) is in his way.

Rethink is aware that though Baxter appears physically unthreatening, his unprecedented affordability and intelligence may nonetheless pose a different kind of threat to factory workers. Brooks firmly rejects the idea that Baxter will replace humans on the factory floor. At the core of Rethink is a vision of a brighter industrial future in which mindless repetitive tasks have been automated and humans are free to do more interesting and satisfying work.

“Our vision is that Mildred-the-line-worker becomes Mildred-the-robot-trainer,” Brooks explained to a crowded auditorium during a recent TED talk. As he spoke, a projection screen showed blown-up images of smiling factory workers with their arms around Baxter.

Yet inside Rethink’s office, it is clear that making humans comfortable around a humanoid robot like Baxter is no easy task. In the space’s center is an eerie area roughly the size of a tennis court that is referred to as the “robot dance floor.” It’s filled with Baxters. A team of mostly middle-aged computer science Ph.D.s. monitors the 6-foot-tall humanoids as the robots pick up boxes from conveyor belts. Crossing the dance floor felt like walking in a room filled with people who had all frozen still simultaneously, but whose eyes still watched my every move.

Another area called “Matt’s Playroom” serves as a Blade Runner-esque robot workshop. Here, Baxters have been cut up into pieces in order to test the endurance of their parts: arms swing back and forth, screens nod left and right, claws open and close.

Designers like Matthew Klee are working hard to give Baxter a loveable personality. Klee, a slim, soft-spoken man in his 30s, is an expert in making machines that are easy to use (he also worked on Skype and the Xbox 360 Kinect).

“You don’t want it to be communicating any kind of arrogance, like it’s better than you, or anything like that,” Klee says on the day that I visit Rethink.

The design team had to walk a fine line. “What is the difference between an intelligent robot and a dishwasher?” Klee asks. “Both are machines, but there is something special about the robot that makes people treat it very differently and assign it human qualities.” The designers wanted Baxter somewhere in the middle. If Baxter was not human enough, people would not feel comfortable interacting with him. On the other hand, if he was too confident, people would be threatened and feel like he was replacing them.

The designers eventually realized it was all in the eyes. The team tried over 80 different styles. Some were incredibly realistic; others, downright cold.

“As soon as they started to look a little human, people started to freak out,” Klee tells me. In the end they settled on a pair of eyes that were cartoonish yet realistic enough to convey five emotions: neutral, surprised, focused, confused, and sleepy. I ask if Baxter can blink. “No,” Klee laughs, “that would be creepy.”

Rethink employees refer to Baxter as an “it,” not a “he.” When Product Manager Alex Goodwin accidentally uses “he,” he stops himself midsentence and restarts using the neutral pronoun. Apparently there has been a long-running internal debate over which pronoun to use when referring to Baxter. Brooks has long held that, officially, Baxter is an “it.” Customers, on the other hand, consistently refer to Baxter as “he.” Some factories have treated their Baxters like a giant family, assigning each robot its own first name and using “Baxter” as its last.

But despite the nicknames, Klee is worried that many manufacturers don’t seem to understand the importance of emotionally integrating robots into the workforce. One of the most common questions from factory owners is if Baxter can move faster.

He can, Klee says. But for some reason, thinking about speeding Baxter up makes the designer pause and look a little shaken.

“He’s made me nervous a couple of times in the past few weeks,” he admits after a second. “It’s sort of like if a dog you really get to know bites you one day. You’re like, Baxter … why’d you do that?”

I imagine Baxter executing human-like motions at three or four times human speed, and realize just how terrifying such a sight would be to factory workers worried about losing their jobs.

Klee senses my discomfort and tries to crack a joke. “Now we’re trying to figure out if we can keep Baxter from biting people,” he says, laughing.

Rethink launched a major media blitz in September 2012 to introduce Baxter to the world. The robot appeared on the cover of Time and on numerous “Game-Changing Technological Breakthroughs of the Year”-type lists. Baxter was famous long before any robots hit the factory floor.

The key to Rethink’s publicity campaign was convincing the public that Baxter was not out to steal their jobs. To help with this, the company directed reporters to talk to Chris Budnick, the owner of Vanguard Plastics. Budnick quickly became one of the most recognizable robot-allies in the manufacturing world. Nearly every article about Baxter includes a quote from Budnick: “Our folks loved it and they felt very comfortable with it,” he told The New York Times. “Even the older folks didn’t perceive it as a threat.” In an interview with CNBC, Budnick went so far as to promise that “Baxter won’t replace any of our people.”

How could a factory owner in today’s struggling manufacturing world make such a claim without losing mass amounts of revenue? Was Budnick lying when he said he wouldn’t fire people, or had he figured out some genius way to automate without sacrificing human jobs? I decided to ask Chris Budnick himself.

 

When the Industrial Revolution made its way to the countryside of New England, it traveled by barge up the Farmington Canal. Not long after the canal’s completion, the 60-mile-long channel of water was filled in and replaced by train tracks. Once-quiet villages alongside the canal line were transformed into industrial hubs. Large brick factories flanked the tracks, and the Farmington River ran brown with pollution.

Now the Farmington Canal railway has been converted into a bike trail. Cycling the 30-mile stretch from New Haven to Vanguard Plastics, I pass ghostly brick complexes that once employed thousands of people. The air is clear, and the foothills of the Appalachians loom regal on the horizon over empty industrial parks. It’s hard to imagine the puffing smokestacks and churning engines that once earned Southington, Conn., its motto,  “The City of Progress.”

Yet nestled between a forest and a highway on the north end of town, in a small building that is made of aluminum siding and flanked by two huge water silos, industrial Connecticut lives on. This is Vanguard Plastics.

I dismount in a parking lot at the end of a long driveway, completely exhausted from the ride. I expect to find a bleak industrial wasteland but the first thing I notice is a vegetable garden. It wraps around the building and is filled with basil, grapes, heirloom tomatoes, and a 20-foot-tall sunflower.

The garden is my first clue that Chris Budnick, the president of Vanguard Plastics, is not your typical factory owner. As I pause to catch my breath (and ogle the produce), Budnick emerges from the factory. He’s incredibly fit, 49 years old, and wearing a tucked-in denim shirt. He welcomes me into his air-conditioned office. The walls are covered with paintings of trout and photos of his children.

Budnick went to military college, served four years as an officer, and after surviving Ranger School, was offered a spot in one of the Army’s most elite units. He turned it down to return home and help his father run the family plastics business.

As we walk out of his office, Budnick hands me a pair of safety glasses and leads me onto the factory floor. Vanguard produces a wide array of plastic parts: grips for all-plastic Colt handguns, fittings for Dodge Ram steering columns, caps for water filters, and the small medicine dosage cups. The factory is on track to crank out 40 million cups in 2013.

Budnick is the first to admit that manufacturing plastic parts can be dull. “I like to think of my job as a way of helping people,” he says. When Budnick describes the conditions in factories overseas, he does so with a tone of moral outrage. He can’t believe that people are subjected to such cruel working conditions and paid so little. Hence the garden, and the break room table covered with fresh produce, which he says he offers to his employees to make sure they have access to good food.

It’s a nice gesture — Budnick’s desire to support his workers is undeniably sincere. But the trend in manufacturing is hard to ignore: robots like Baxter are increasingly replacing workers as American factory owners “race to the bottom” to compete with Chinese manufacturers’ absurdly low prices.

Budnick admits that these days, it’s hard to justify employing a human due to high taxes, health care costs, and the specter of Obamacare. “The state of Connecticut is making it as difficult as possible for me to hire someone,” he says. Judging by their actions, his colleagues agree. Between 1969 and 2011, the number of Connecticut manufacturing jobs has plummeted from 477,000 to 174,000. During the same period, manufacturing’s output has soared.

Budnick stands by his promise not to fire the workers whose jobs will be replaced by Baxter. Yet at the same time, he plans to expand production without hiring new workers. Vanguard has already invested heavily in new machinery, robots, and computers. The number of injection molding machines (23) nearly outstrips the number of employees (30). In the 1970s, Vanguard employed around 60 people. Now the company makes far more stuff and employs half as many people.

Budnick views Baxter as a potentially exciting new player in the game of globalized manufacturing. “We like to buy new toys,” Budnick says, “as long as the new toys pay for themselves.” Baxter is so affordable that he could pay for himself in about one year. Budnick has already purchased one “Baxter,” and if all goes well, he’ll purchase two more soon.

Automation is more than just a  way for Budnick to boost revenue. The factory needs it to survive. Without robots, Budnick says, he would be unable to offer customers low prices and pay his workers a reasonable wage (an entry-level machine operator is paid $30,000 per year, but costs the company around $40,000 after taxes and benefits). Budnick says that an equivalent facility in China — lacking American-style automation — would employ 100 people but pay them only a fraction of what he pays his workers.

So how will Baxter’s eyes help save Vanguard? Budnick isn’t so sure the eyes matter much. “Well, Baxter has to sell,” he jokes. Jay Marple, the automation engineer tasked with training Baxter, mishears Budnick. “What? He’s got a soul?” he asks. We all turn to look at Baxter and laugh. Marple admits he wishes he could replace Baxter’s face with a programming console. To manufacturers like Budnick, the Rethink vision of collaboration is secondary: what matters most is speed and reliability. The fewer humans required to watch over Baxter, the better.

Still, Budnick is excited about Baxter. He is thankful that he is in a position to experiment with what he calls “a new species of robot.” For now, Baxter isn’t very profitable, but Budnick’s betting that the robot’s descendants will be revolutionary. “It’s a Wright Flyer,” he said. “It doesn’t go very far, but it gets off the ground.”

 

The troupe of automators at Vanguard continues coaching Baxter for several hours. He keeps getting better and better at stacking the medicine cups. There are hiccups every once in a while — the cups will slip out of Baxter’s grip a moment too soon, or he will knock over neighboring towers with an enthusiastic swing of his claw. But the overall trend is clear: this task will soon be fully automated.

From time to time a 30-year-old Hispanic woman wearing safety glasses and a yellow blouse passes by and disappears behind a giant machine. I sneak away from the crowd and hide behind the machine to watch her work. Her movements are fluid and refined. She bags the towers at least five times faster than Baxter. I go back and ask Budnick who she is. “That’s Lesbia. She is a really nice lady. She has been with us for a long time. Her husband, Jaime, is over there.” He points across the factory.

“Is it OK if I replace Lesbia for a little while?” I ask. When Lesbia rotates to another machine across the factory, I start stacking. The cups are not nearly as stackable as I had anticipated. Just before I slip one giant tower into a bag, it breaks apart in six different places. The cups scatter all over the assembly line. I hear laughter and realize that Lesbia is watching. She comes over and silently starts stacking the cups with precision and aplomb.

I try again. And again. Over time I gradually become competent. I try to start a conversation about automation.

“So, how long have you been doing this job?”

“Seven years,” she replies.

I ask if she is worried about the robot. She points at Baxter and starts laughing. She clearly doesn’t feel too threatened. To her, Baxter is just another machine.

After I’ve bagged about 6,000 cups, Budnick walks over and silently watches me. “It’s a dull job,” he says. “I couldn’t do it. I would beat my head against the wall.”

When I rejoin the spectators surrounding Baxter, the conversation about automation is still running. Reina Ellis, a representative of Rethink, shares the simple heuristic she uses to justify taking jobs away from humans and giving them to robots:  “If it is something you don’t want your child to have to do when they grow up, then that is something that should be automated.”

But would that justification console Lesbia if she were replaced by Baxter?

 

For centuries, engineers have been working hard to eliminate toil from our lives. They’ve been wildly successful at making factories run more efficiently. But for the most part they’ve operated without a clear vision of how their inventions will change the work landscape. But that’s where Rethink is different. The reason that they have been able to garner so much attention is that they have attempted to fully articulate a vision of an automated future: one where man and machine work side by side, and where humans are able to pursue more satisfying and meaningful jobs.

The reality of the situation is darker. Once Baxter is deployed, what happens in the factories is increasingly out of the robot designers’ hands. Manufacturers who take control of the situation don’t have the luxury to make decisions based on utopian visions. They must lower prices, or perish. Most likely, many laborers will lose their jobs, and eventually, factories will be filled with fleets of Baxters moving at high speeds with only a few human supervisors.

Yet whether or not Baxter’s eyes betray his maker’s vision, why were we so captivated by that vision in the first place? Baxter makes us think about our answers to an important question: What do we want the automated world to look like? Do we want Lesbia to be a robot supervisor? Or perhaps doing work she finds even more satisfying?

The shape of the automated world is destined to lie beyond the control of engineers and manufacturers. Perhaps the burden is on the rest of us to figure out what we really want from machines, before workers like Lesbia end up in the unemployment line.

Comments