The Path To Better Robots Lies In Teaching Them Human Interaction
Inhuman, unfeeling robots are a staple of science fiction - as is the fear that one day, their cold, logical minds might deem humans unnecessary. That danger isn't really realistic - it's far likelier that cumbersome machines, unused to human beings, will cause fear, frustration or, in worst-case scenarios, death. Naturally, this isn't something we want to see happening - which is why robotics researchers across the world are breaking out of the lab and into the boardroom.
Take what happened three months ago in Vancouver, for example. In a boardroom in the University of British Columbia, over one hundred volunteers lined up in a boardroom to experience something which, years from now, will be lumped in with seeing electricity in action at the first World's Fair. They came to meet a fellow named Charlie.
Except Charlie isn't really a person at all. He's a $400,000 humanoid research robot, and all he was there to do was hand out water bottles. It was a simple enough gesture on his part - one which would be taken for granted if performed by another human being: he'd pick up a water bottle, look down at it, then meet the gaze of the person he was giving it to as he extended his hand.
The whole process took only two seconds, but was the result of months of research and hundreds of hours of programming.
In laboratories across the world, a curious thing is taking place. The much-dreaded robot takeover is already well on its way. There are robots designed to flip pancakes and mix drinks. There are robots that can care for the elderly, clean up public parks, or help on the assembly line. There are robots designed to teach, to police, and to speak...and they all have one thing in common.
They are, each and every one, completely and utterly inhuman.
The chief issue with all these machines is that they're...well, they're machines. They have no inkling of social cues, they don't understand face-to-face interaction, and they certainly aren't constructed up to Asimov's standards. As a result, the researchers responsible for the development of these machines are turning to a new class of scientist; men and women who are experts in human-computer interaction. Through them, these researchers and developers are fine-tuning their robots with human gestures, mannerisms, and even ethics and thought patterns. Little by little, these men and women are blazing the trail for the robot revolution.
"When I started down this road, I wanted polite robots; robots that understood social cues," explained robotics researcher Elizabeth Croft, founder of the Collaborative Advanced Robotics And Intelligent Systems Laboratory (CARIS). Croft, who started out as a mechanical engineer, eventually found her way into human-robot interaction while working to improve the safety of men and women in automated factories. There, she came to the realization that, unless robots and humans learned to read and understand one another, they could never truly share a workspace.
"If the person doesn't understand what the robot is up to and the robot is not aware of the person, then you cannot have safety," said Croft. The goal of her organization, then, is simple: to design a new generation of robots that are capable of working side-by-side with humans by virtue of being "as easy to interact with as other humans." These robots will know how to share and communicate, how to interrupt, and even how to understand simple hand gestures - all things we take for granted, and yet all things robots are not inherently capable of.
"I can say to a person 'can you get me that?' and they'll understand immediately. That's what I want to do with a robot," added Croft.
In order to achieve this surprisingly lofty goal, CARIS has to effectively reverse-engineer thousands of years of human social conditioning and mannerisms, and condense all of this into code. In Croft's words, they need to take human behavior and weave it into "the interactive fabric of a robot living in your life." This task requires CARIS students and researchers to delve into a wide array of disciplines, including sociology, physiology, and psychology.
CARIS agents have studied hundreds of hours of video involving human subjects performing mundane tasks to break down the mechanics of non-verbal communication. They've read all the most famous books on human interaction and social cues. They've even hooked up human volunteers to biometric sensors in order to determine - and eliminate - robot behaviors that tend to make people nervous.
"Fast trajectories" explained CARIS PhD student Matthew Pan. "Fast trajectories that came out of nowhere were really scary. People also don't like when robots perform unnatural contortions like twisting their head completely around or bending their arms backwards."
This is some very cool stuff, and the idea that we might one day see robots working seamlessly with human employees is downright incredible. Not everyone is so enthusiastic, however. Some men and women in the private sector, noted Pan, are actually opposed to the work CARIS is doing.
"Auto workers, especially auto workers' unions, are very opposed to these types of technologies because they fear that it's going to take over their jobs," said Pan. "As a result, we need to work very hard to get the message across that we aren't replacing these workers, but we're actually enhancing their ability."
Indeed, the enhancement of human life - and the abilities of human workers - is the end goal of everything CARIS- along with its sister organizations in the field of Human-Robot Interaction - is doing. The research, which began back in the 1990s, ranges from enabling robots to assist in the workplace to designing machines that will assist the disabled and the elderly.
At Brown University, for example, researchers have designed a robot that can follow a human master, knowing whether to halt, follow, or wait with a simple wave of a hand. MIT scientists have determined that humans share a closer professional bond when asked to switch roles with robotic assistants. Hungarian developers have created humanoid robots that are capable of gaining the trust of dogs. With all the work that's being done, the perception of robots - even those of the researchers- is beginning to change.
"We are, as a community, beginning to realize that robots are another category of being," said Croft. "Our human volunteers are treating robots not as people, nor as things, but as new entities occupying a gray area somewhere in between; one where human qualities like responsibility and perhaps even morality might apply."
Inhuman, unfeeling robots are a staple of science fiction...but they aren't what's being designed here. Instead, researchers across the globe are hard at work designing robots which don't send us scrabbling for the walls of the uncanny valley. Cold, logical machines are likely to remain where they belong - in science fiction horror stories. What we're going to see instead is a new breed of robot which, while not necessarily human, will end up being the next best thing.
Like What You're Reading? Sign Up For My RSS Feed Here!
Robotics And Artificial Intelligence Blogger