Slashdot Mirror
Teaching Robots New Tricks Without Programming
cylonlover writes "Maya Cakmak, a researcher from Georgia Tech, spent the summer at Willow Garage creating a user-friendly system that teaches the PR2 robot simple tasks. The kicker is that it doesn't require any traditional programming skills whatsoever – it works by physically guiding the robot's arms while giving it verbal commands. After inviting regular people to give it a try, she found that with few instructions they were able to teach the PR2 how to retrieve medicine from a cabinet and fold a t-shirt."
"Without programming?"
Bullshit. Look in the article, in the picture in the article.
Program's right there, on the right side.
"Test subjects were provided instructions on how to teach the robot similar to what you'd expect when buying a sophisticated appliance."
"Tutorial: Programming PR2 by Demonstration."
"Step 1... Say: 'TEST MICROPHONE'."
"Step 2... Say: 'RELEASE RIGHT ARM.' ... Move the arm to a neutral pose and say HOLD RIGHT ARM."
If this isn't programming, then I'm not a programmer. Instead, I'm just someone who manipulates a text editor.
its just the method that has changed
Teaching Robots New Tricks With Non-Traditional Programming
There, fixed that for you.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
I hear ya. I've found folding laundry so complicated, that I just have my mom do it for me and bring the folded clothes to my basement.
Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
Isn't that still an important and useful qualitation, since the vast, vast majority of people can't, shouldn't and don't fucking want to write code? And, I would argue, nor should they ever need to. Writing arbitrary invented languages, with awkward syntax and extremely-non-human thought-structures, to accomplish esoteric tasks has never been an intuitive or optimal way of getting shit done.
Trust me, everybody would loooooooove for the computer to take instructions like a human but it's not going to happen because of everything that's implicitly understood. So you can teach this computer to fold a shirt, if you hand it an XS shirt and an XXL shirt will it figure out that it must adapt the folding action to the size of the shirt? I bet you any 5yo would figure that out all on their own because they've understood the basic concept of folding a shirt. Take a fundamental sentence like "put the black and white pants on the top shelf" did we mean the black pants and the white pants, or the black and white checkered pants?
All that happens is that some really smart people will try really hard to write code that guesses what it was people actually meant but without actually knowing the context and purpose they'll fail miserably. Not to mention all the times they'd have to guess at do what I meant, not what I said because normal people when facing a choice between the reasonable and the absurd pick the reasonable like. Like say you have a knife and a chicken and you ask what to do with the knife and they answer "Cut the chicken to pieces and put it in the oven" most people will understand that you're to put the chicken in the oven, not the knife - even though you didn't ask what to do with the chicken.
Or the TL;DR version: Good luck, I don't think we'll be unemployed any time soon.
Live today, because you never know what tomorrow brings
So... how did that 5yo come to "implicitly understand" so much that you never had to write code to teach her how to adapt the folding action to the size of the shirt? DNA defines how to grow a brain, not really how it will understand the world it encounters, how it will respond to that world, or the methods of thought internally used to process either of those things. Is there really any reason why artificial creatures shouldn't follow biology's lead in the whole "learning" thing?
Actually I'd say that's a pretty complex question how much the brain is "preprogrammed" by DNA, clearly all the inputs like sight, hearing, taste, smell and touch are hooked up in some fashion with some form of processing, some basic output like crying, a lot of reflexes and instincts and possibly also knowledge are considered innate and studies on twins vs siblings vs half-siblings vs adopted have shown considerable correlation on "how it will understand the world it encounters, how it will respond to that world, or the methods of thought internally used to process either of those things". For example religiousness has been shown to be a partly hereditary trait, even though the specific religion is a matter of upbringing.
But those things aside, sure the brain works on electrochemical processes not magic and pixie dust and there are projects underway to both simulate a brain and the contents in it. That said, even with today's fastest supercomputers we're not even close to simulating a full human brain with ~100 billion neurons - I think the best we've done is the Blue Brain project of ~1 million, so we're another 5 orders of magnitude away. There's a proposal for a 10 year, 1 billion euro "Human Brain Project" that may have a full brain simulator in 2023 out now. In short it might eventually happen as some LHC-class project, but it won't be folding shirts any time soon.
Oh and in case you were wondering, no you can't take whatever that 100 billion neuron brain has learned and easily transfer it to anything less, each neuron encodes parts of the learning. This is no longer a Von Neumann architecture, each neuron is it's own little processor with thousands of synapses and even the wiring of those is dynamic much like a FPGA. If you thought programming was hard before, you should really see it once we simulate the human brain. The programming model is so bizarrely complex we'll need entirely new tools to even understand the behavior, much less modify it. We'll not be happy black-box training computers the way we do people.
Live today, because you never know what tomorrow brings