The three-foot (one-meter) machine balances a Barrett WAM arm and a BarrettHand atop the base of a Segway personal transporter with two small training wheels. To find its way around dynamic environments, HERB uses two laser range finders and a camera that let it tell people apart from fixed and movable objects like walls and chairs. (A rough layout map of the space is first programmed into the bot.) By observing how people move, the robot uses learning algorithms and probability distributions to predict where they'll go next to avoid running into them. "HERB knows people have intent, that they don't just move in a straight line," says Intel research scientist Sidd Srinivasa, one of the project's co-leaders. To figure out what an object is, HERB compares its live camera image with a set of 3-D models in its database, built up from representative images that researchers showed it earlier.
Manipulating objects in cluttered settings, like carrying a pitcher through a house without spilling anything, takes two skills. First, HERB has randomized planning algorithms to determine the best way to grip or move something as quickly as possible. For example, the robot might be given 30 seconds to "think" of a way to pick up a mug; if it finds one in 15, it then has 15 more seconds to improve its plan. "They're not optimal algorithms, but practical," Srinivasa says.
HERB also uses imitation learning to figure out how to handle objects by watching how people handle them. "We're much better at demonstrating actions than explaining them," Srinivasa says. "HERB takes human examples and learns to generalize from them. It's not just repeating what you're doing." This helps the robot deal with new, unfamiliar items. During a daylong public demonstration in October, HERB moved around in a model kitchen, opened cabinets and a refrigerator, and handed objects to visitors or put them in a recycling bin—all with only a few missteps.
Srinivasa would eventually like HERB to learn some simple social rules—such as knowing to go around a group of people rather than through them—as well as how to deal with completely unfamiliar environments, even in the dark. A useful robotic assistant is about a decade away, he estimates. "Moore's law"—the rule of thumb first posited by Intel co-founder Gordon Moore in 1965 that the number of transistors on a chip doubles every two years—"is on our side." he says
In the meantime, learning robots will sometimes surprise their own creators. Srinivasa tells how an early version of HERB puzzled researchers when it was grabbing coffee cups to place them in a dishwasher rack. It used a strange hand position, with one of its "thumbs" pointing down. Then they realized this was a "far more efficient motion" used by professional bartenders, Srinivasa says: "They lift from underneath and pour in a single smooth motion, like in the movie Cocktail." He calls these surprises "one of joys of doing manipulation research."
Subscription Center
See what we're tweeting about
7 Comments
Add CommentWhen I was in a tech school in the early 90's, one of the teachers was a former NASA guy, and incredibly intelligent. He explained back then how robots learn. Can't remember the details, but it was pretty simple and obvious when he explained it.
Reply | Report Abuse | Link to thisSo...yeah, they can, do, and will learn from previous experiences. The oldest example that I know is the ones that play chess (the actual example he used). Once you beat them, you will never defeat them the same way again. I could give his name, but he'd probably not appreciate it, and he was such a great teacher, I don't want to make his life rough.
Why are we teaching robots to learn? Haven't you people seen Terminator!!! We are all gonna die!
Reply | Report Abuse | Link to thisYea... ill second that!!! Teaching robots to learn is a bad idea. They first will put millions of diligent workers on the streets, creating more poverty then we have now. Eventually it will get to the point were they are going to college and learning next to human students. If a robot learned at the rate or faster then humans can we will out source humans all together. It sucks getting laid off now because a younger smarter version of yourself takes your job, imagine getting laid off because a robot is taking your position. The next thing we will have is angry robots trying to get their rights. PROTESTING ROBOTS is the last thing we need!!!
Reply | Report Abuse | Link to thisAs he said above we have all seen Terminator and I-ROBOT!!!! Didn't exactly end well in either of those movies.
Perhaps we're missing the point of science fiction robot movies. Perhaps it is not the robots that are the problem, I think we may be the problem. Robots tend to act logically and when faced with illogical humans, they make the logical conclusion that we are dangerous and seek to exterminate us.
Reply | Report Abuse | Link to thisIf we treat we create androids that have human-like intelligence and we treat them well, there shouldn't be a problem. If you treat them like slaves and property and they have the capacity to learn, they will eventually seek to overthrow their oppressors. If they can learn, teach them to be good and value life. Don't teach them that we are evil, hateful, and violent creatures and we should be fine.
In other words, if we create androids in our own image, we had better make sure that the image we work from is based off of the good in this world, not the evil. If we are evil, and create in our own image, we shouldn't be surprised when our creations turn out to also be evil.
When it comes right down to it, there is a saying I heard a while back: "a computer never really has an error, it always does exactly what it was programmed to do. The human who programmed it made all the errors." If androids go crazy and kill us all, it's because we programed them with the capacity to do so.
OK, so we can teach robots simple learning algorythms. I was very pleased when I taught one to find its way out of a maze by the shortest route. But remember Eistein taught himself that E=Mc^2 without any form of computer. Just try writing code to do that!
Reply | Report Abuse | Link to thisThe greatest Android ever created became a Painter...
Reply | Report Abuse | Link to thisBut can robots learn not to drop wrenches on their motor control boards. That is the real question.
Reply | Report Abuse | Link to this