Ursula works at Brown University learning how to pick flower petals from a daisy. It’s not difficult work for a human, but Ursula is a robot and picking flower petals represents a skill set automation hasn’t been able to master yet. The Humans to Robots Laboratory at Brown, featured in a recent New Yorker article titled “Welcoming Our New Robot Overlords,” is actively working to change that.
Ursula is one of three Baxter robots at the Brown lab, less than half a block removed from Thayer Street. It’s a $30,000 piece of equipment that, like humans, has two arms. But Ursula’s arms have an extra elbow joint for increased mobility. And where the human elbow can only bend in two directions, Usula’s work more like a shoulder and can move in any direction.
The robotic arm, in and of itself, isn’t all that new of technology. Ursula is a 2012 model, and there are more than 300 like her at research universities across the country, exponentially more being used in private industry. Robotic arms have been assembling cars and other factory-made goods for years.
What makes Ursula unique is what PhD students John Oberlin, Rebecca Pankow, and computer science professor Stephanie Tellex are teaching it to do.
Ursula has a rudimentary, robotic version of vision, and it is learning to comprehend what it sees. There is a digital camera at the end of the arm that feeds photos filled with information about the outside world to a computer program that tells Ursula what to do about it.
“By seeing we mean using the images to identify the location of the flower petal,” said Pankow, whose PhD focus is computer vision. “It’s using cameras to guide machines. Being able to take those images, analyze them and identify where the flower is in the real world. We have to create this software to give it the capability to utilize the camera in a similar way that we use our eyes and our brains to navigate the world.”
Under the tutelage of Pankow, Oberlin and Tellex, Ursula’s vision, and its understanding of what it is seeing, are rapidly increasing.
“The point we are at now,” explained Oberlin, “if you brought a blueberry bush in, the two of us could sit down and train the robot over the course of an hour or two and it would isolate parts of it. We have a process that makes it easy for us to make models of new plants.”
This advancement, they said, is helping to usher in the era when the robotic arm is as ubiquitous as the washer/dryer.
“The tools that we are working on are the first that would allow a pedestrian human to teach a robot,” Oberlin said. “We have a process now that a pedestrian human could use.”
He added, “We used to not have cars, now a lot of have cars. We used to not have refrigerators, now a lot of us have refrigerators. Well, there’s this arm. I think it’s not going to be too long until there is a cheap, useful arm that you can have in your home. We have a vacuum that drives around in your home, how long is it going to be until there’s an arm in your home?”
Their research focuses on finding ways to make robotic arms more useful in small-scale sustainable agriculture. Oberlin talked about teaching a robot to tend a home garden, Pankow said that such an arm could help the elderly care for and harvest fruit in their backyards.
They both recognize the technology also has the potential to further erode humans role in labor market. If Ursula can help grandma pick apples, it can help corporate farmers and other industries, too. The New Yorker called it a “multibillion-dollar business opportunity” and profiled a Michigan factory that is increasingly automated and has shrunk from 10,000 employees to 2,000 since the 1990s.
Oberlin has a keen sense of how automation has already affected humanity. He spoke about how automation began with the rock, which gave way to the shovel, which gave way to the plow, which was advanced first by the horse, and then the motor. He dismisses the idea that it will be the blueberry-picking robot that dooms the human labor market.
“The things that human hands can be useful for in agriculture now tend to be artisinal and luxury items already,” he said, explaining that staple food items of American’s diets – such as wheat, soy, meat and dairy products are already largely automated.
“Look at the amount of work that has already been taken away,” Oberlin explained. “Are the jobs disintegrating because of this tail end of automation we are in right now or have the jobs really been disintegrating all along and we’re just now starting to realize the consequences? It’s not because of the .5 percent of human attention that we are learning how to address now, it’s because of the 99.5 percent of human labor that we removed during industrialization and economically-motivated progress.”
He added, “Automation isn’t really the problem, profit is the problem. There’s a lot of sort of dirty practices and unsustainable practices that occur with automation at the moment. It’s fueled a little bit by greed and people who want to own the amplifier. That’s the danger here. A small number of people can acquire a small amount of material resources that can change the game relative to other people and will out compete other competitors and even co-existers.”
He and Pankow both agreed a universal basic income, a government subsidy to cover living expenses for individuals, is one way society will need to adapt to increasing automation. Tellex, who wasn’t at the lab the day I visited, also supported the concept in the New Yorker article.
“I think it’s certainly something that can be a viable solution to the problem of factory jobs, and as we get better at automation higher skilled jobs, going away,” Pankow said.
They also hope automation can help cure whatever problems it created in the first place
“Is it possible we can now go back and make it less wasteful,” Oberlin asked. “Make it better, make it more efficient. Make those cars that are going to be good, and last forever. Not make a thousand light bulbs. Make food the right way. If we can do that, do we really need all these huge factories?”