Robot can be programmed by casually talking to it

Robot can be programmed by casually talking to it

Robots are getting smarter, but they still need step-by-step instructions for tasks they haven't performed before. Before you can tell your household robot "Make me a bowl of ramen noodles," you'll have to teach it how to do that. Since we're not all computer programmers, we'd prefer to give those instructions in English, just as we'd lay out a task for a child.

But human language can be ambiguous, and some instructors forget to mention important details. Suppose you told your household robot how to prepare ramen noodles, but forgot to mention heating the water or tell it where the stove is.

In his Robot Learning Lab, Ashutosh Saxena, assistant professor of computer science at Cornell University, is teaching robots to understand instructions in natural language from various speakers, account for missing information, and adapt to the environment at hand.

Saxena and graduate students Dipendra K. Misra and Jaeyong Sung will describe their methods at the Robotics: Science and Systems conference at the University of California, Berkeley, July 12-16.

The robot may have a built-in programming language with commands like find (pan); grasp (pan); carry (pan, water tap); fill (pan, water); carry (pan, stove) and so on. Saxena's software translates human sentences, such as "Fill a pan with water, put it on the stove, heat the water. When it's boiling, add the noodles." into robot language. Notice that you didn't say, "Turn on the stove." The robot has to be smart enough to fill in that missing step.

Saxena's robot, equipped with a 3-D camera, scans its environment and identifies the objects in it, using computer vision software previously developed in Saxena's lab. The robot has been trained to associate objects with their capabilities: A pan can be poured into or poured from; stoves can have other objects set on them, and can heat things. So the robot can identify the pan, locate the water faucet and stove and incorporate that information into its procedure. If you tell it to "heat water" it can use the stove or the microwave, depending on which is available. And it can carry out the same actions tomorrow if you've moved the pan, or even moved the robot to a different kitchen.

Other workers have attacked these problems by giving a robot a set of templates for common actions and chewing up sentences one word at a time. Saxena's research group uses techniques computer scientists call "machine learning" to train the robot's computer brain to associate entire commands with flexibly defined actions. The computer is fed animated video simulations of the action -- created by humans in a process similar to playing a video game -- accompanied by recorded voice commands from several different speakers.

The computer stores the combination of many similar commands as a flexible pattern that can match many variations, so when it hears "Take the pot to the stove," "Carry the pot to the stove," "Put the pot on the stove," "Go to the stove and heat the pot" and so on, it calculates the probability of a match with what it has heard before, and if the probability is high enough, it declares a match. A similarly fuzzy version of the video simulation supplies a plan for the action: Wherever the sink and the stove are, the path can be matched to the recorded action of carrying the pot of water from one to the other.

This is pretty cool but of course one can't forget about the risk that robots having intelligent life can mean for mankind.
I think if we are going to get the point of having intelligent robots, we need to make them sympathetic to humans in order for us to survive. Otherwise they will most likely just kill us off, maybe not right away but eventually.

a reply to: knoledgeispower

Other wise they will most likely just kill us off, maybe not right away but eventually.

They already are. Whats a drone but an intelligently guided killing machine?

All weapons of war are the perfect killing machine. And the people that man them execute their tasks robotically.

Whats intelligent about that?

"Robot can be programmed by casually talking to it?"

So what? So can husbands.


Sorry, off-topic, but I had to, I had to, I tell ya.

originally posted by: intrptr
a reply to: knoledgeispower

Other wise they will most likely just kill us off, maybe not right away but eventually.

They already are. Whats a drone but an intelligently guided killing machine?

All weapons of war are the perfect killing machine. And the people that man them execute their tasks robotically.

Whats intelligent about that?

That's not what I'm talking about when I say intelligent robots and you know that. I'm talking about a robot that can think for itself not be controlled by man.

originally posted by: charles1952
"Robot can be programmed by casually talking to it?"

So what? So can husbands.


Sorry, off-topic, but I had to, I had to, I tell ya.

Bummer you didn't have anything of value to say about the subject, I would have been interested to know what you had to say on the subject instead of some wise crack.

Sorry, off-topic but I had to, I had to, I tell ya.
~~~~~~~~~~

To keep on topic: Here is the video that is included in the article

You're assuming that a robot can ever be made which can think for itself, unfortunately this is a long way from being achieved, if indeed it would be possible at all. The robot is simply following what it has been told to do and so is a long way from needing Asimov type coding to prevent it from harming humans.

Sweety a drone is u dumb piece of equipment that can not do nothing if not guided and controlled by humans. a reply to: intrptr

Robots, programmable with the spoken language, yet prone to not understanding said language and getting it all wrong?

you mean government employees?

Theyve been around for yonks mate... I know, I've seen them.. no AI there.. not even I.

a reply to: knoledgeispower

And visions of 'the sorcerer's apprentice' spring to mind



And 30 years of programming conjures up something frightening:

Try [
Make ramen
] catch [
Gas clogged, try [
Unclog gas ] catch [
Stove explodes
]
....million lines later...
Mold in Ramen, try [
Order takeaways
] catch [
Credit card out of money
........


There is something to human intelligence that may require quantum intelligence.

But programming a computer by voice is NEAT. the problem is NP-complete though.

a reply to: knoledgeispower

That's not what I'm talking about when I say intelligent robots and you know that.

I know that. What you're proposing doesn't exist yet. I think you know that, too.

I'm talking about a robot that can think for itself not be controlled by man.

They already do that too. Besides autopilots, software programs run autonomously quite well.

The difference being sentience. AI implies that but is a long way off from real world.

Computers will never know that they know. At best they are only capable of executing the next instruction. Thats rote programming.

My comparison to supposedly intelligent people controlling bombers, missiles, tanks, drones, whatever is they are "programmed" to kill and they act this out robotically, without question. Thats akin to a computer program that "just runs programs".

If that helps.

originally posted by: intrptr
a reply to: knoledgeispower

That's not what I'm talking about when I say intelligent robots and you know that.

I know that. What you're proposing doesn't exist yet. I think you know that, too.

I'm talking about a robot that can think for itself not be controlled by man.

They already do that too. Besides autopilots, software programs run autonomously quite well.

The difference being sentience. AI implies that but is a long way off from real world.

Computers will never know that they know. At best they are only capable of executing the next instruction. Thats rote programming.

My comparison to supposedly intelligent people controlling bombers, missiles, tanks, drones, whatever is they are "programmed" to kill and they act this out robotically, without question. Thats akin to a computer program that "just runs programs".

If that helps.

Of course I know that there is no sentient robots. This work is a vital stepping stone and unlike the stupid chat-bot that was deemed a success because it was able to fool people into thinking it was human, this could very well herald in an era of human and AI interaction beyond trying to trick people with preset communication abilities. This is the beginning steps to a learning machine that is being taught how to grasp abstract concepts and learn what it needs to do to complete the required task beyond simple instruction. Once the foundation for this kind of learning has been set, applying this concept to communication will allow it to learn on it's own how to interact with people. That learning template can be applied to all areas of knowledge and will simply require exposure. Having the ability to analyze on it's own what needs to take place in order to accomplish task "X" (any objective you can think of) could be dangerous....a simple request such as "save human kind" from what we see may not go anywhere, but it's not a far reaching concept for it to realize it doesn't have enough information how to accomplish this and understands it needs to learn further and take up that "step in the process" to complete it's task.


At this rate, I do not think it will take long for robots to become sentient.

Yes your drone example is akin to the robot in the article but I when I said intelligent I was meaning sentient and I think you know that too.

a reply to: knoledgeispower

Dear knowledgeispower,

Sorry to have disappointed you, I'll try again.

I remember going to an IBM center and getting in on the standard programming methods of the day. At the time, that meant punching holes in a Hollerith card. From decade to decade, programming has been getting easier on the human as the computer has more speed and memory to store its sub-programs, data, languages and so on.

The world was exposed to Stanley Kubrick's "HAl 9000" over 45 years ago. Speech to text software is standard issue on Windows machines now, "Smart" phones recognize voice commands which cause small programs to start and stop.

Japan has been leading the world in robotics for some time, more than a decade ago they showcased the Actroid:

The Actroid woman is a pioneer example of a real machine similar to imagined machines called by the science fiction terms android or gynoid, so far used only for fictional robots. It can mimic such lifelike functions as blinking, speaking, and breathing. The "Repliee" models are interactive robots with the ability to recognize and process speech and respond in kind.

en.wikipedia.org...

I have no doubt that voice communication will be (or has been?) augmented to understand International Sign Language, and eventually, with brain waves from human to robot.

What is being discussed here is one more step on a long and predictable path. The ability to question and change your own programming will be an important step, but, for robots, I suspect that it will be based on another program telling the robot to question its program under certain conditions.

I understand that some phones also serve as TV remotes. So we have gone from getting up and turning dials, to sitting back and pushing buttons, to telling our phones what program we want the TV to show.

I'm not bad mouthing scientific exploration, but I wanted to explain in a more serious and lengthy manner what I was trying to convey with

So, what?

a reply to: charles1952

The ability to question and change your own programming will be an important step, but, for robots, I suspect that it will be based on another program telling the robot to question its program under certain conditions.

I like that. A difference engine is all they are. The more "advanced" software becomes the more differences they will be programmed to choose from and the faster they will execute the solutions. Already quicker and more capable than man his machines will eventually simulate sentience so well it will be hard to tell the difference.

Still not sentient, though. Not self aware. Not life as we define it.

Only Hollywood proposes that distinction.

Sean Carroll, a cosmologist at Cal Tech, once made an insightful observation about the human condition when he said, “We are part of the universe that has developed a remarkable ability: We can hold an image of the world in our minds. We are matter contemplating itself.” That has always stuck with me. In a nutshell, I think that’s what it means to be sentient.

I would guess consciousness is an emergent property of matter under certain conditions and configurations. I’m not sure, though, how far along we are in truly understanding just what those conditions are. Consciousness, in turn, includes numerous branches (including objective and subjective states), each of which has it’s own set of properties. I think of sentience as being one of these branches and that it’s properties include various states of subjective awareness. I think the combination of our objective awareness of the events and things around us, along with our subjective awareness (sentience) of our place within that tapestry, constitutes the foundation of our perceived reality.

There’s a good chance that what I just said is pure gibberish and that I’m painfully misguided. Believe me, it wouldn’t be the first time...

Anyway, I’ve always had a “feeling” that sentience will likely emerge one day within our machine friends. But, feeling is not knowing, and all I have to go on is what I’ve read. On the one hand I can somewhat understand things from a programming POV, since I’ve been a system programmer/analyst and software developer for the past 20 years. It’s what I do for a living. I can clearly see that it will not be long before we reach a level of sophistication (both hardware and software) that will allow for the creation of a machine that is nearly indistinguishable from a sentient being. I think we will not only converse with machines on a personal level, but will also become emotionally attached to them in many cases. These machines will out-think us, out-philosophize us, and out-perform us on most levels. But, will they be sentient? I don’t know. Unless their design will somehow allow for the emergence of “subjective processing”, feelings, empathy, compassion, etc., I don’t think they could be considered sentient. However, by all appearances they will seem to be.

Maybe we place too much emphasis on sentience, anyway. It’s not like it’s a requirement. Actually, a super intelligent machine with a sense of self may pose a far greater danger than one without an identity. Eventually, I believe certain classes of machines may become self-sustaining; machines will create other machines, they will repair themselves, program themselves, and even undergo a natural evolution (they will implement improvements to each successive generation). I don’t think being sentient, though, is a requirement for a machine to act autonomously.

It’s a fascinating topic and fun to think about. I don’t have any answers, but have a feeling that reality will far surpass the limited projections being made today. If we don’t annihilate ourselves, the next 100 years could prove to be quite interesting...